Deep Research (DR) requires LLM agents to autonomously perform multi-step information seeking, processing, and reasoning to generate comprehensive reports. In contrast to existing studies that mainly focus on unstructured web content, a more challenging DR task should additionally utilize structured knowledge to provide a solid data foundation, facilitate quantitative computation, and lead to in-depth analyses. In this paper, we refer to this novel task as Knowledgeable Deep Research (KDR), which requires DR agents to generate reports with both structured and unstructured knowledge. Furthermore, we propose the Hybrid Knowledge Analysis framework (HKA), a multi-agent architecture that reasons over both kinds of knowledge and integrates the texts, figures, and tables into coherent multimodal reports. The key design is the Structured Knowledge Analyzer, which utilizes both coding and vision-language models to produce figures, tables, and corresponding insights. To support systematic evaluation, we construct KDR-Bench, which covers 9 domains, includes 41 expert-level questions, and incorporates a large number of structured knowledge resources (e.g., 1,252 tables). We further annotate the main conclusions and key points for each question and propose three categories of evaluation metrics including general-purpose, knowledge-centric, and vision-enhanced ones. Experimental results demonstrate that HKA consistently outperforms most existing DR agents on general-purpose and knowledge-centric metrics, and even surpasses the Gemini DR agent on vision-enhanced metrics, highlighting its effectiveness in deep, structure-aware knowledge analysis. Finally, we hope this work can serve as a new foundation for structured knowledge analysis in DR agents and facilitate future multimodal DR studies.

Search agents enables large language models (LLMs) to iteratively interleave retrieval and reasoning, yielding strong performance on knowledge-intensive tasks. However, their multi-step autonomy also introduces substantial inefficiencies. In practice, search agents often exhibit overretrieval, where redundant or irrelevant documents are repeatedly fetched, and overthinking, where reasoning steps become excessive or unproductive. Both behaviors significantly inflate retrieval and inference cost, yet remain poorly understood, particularly under model scaling. In this work, we conduct a systematic study of overthinking and overretrieval in search agents from a scaling perspective. We formalize both phenomena at the trajectory level and propose fine-grained evaluation protocols that combine automatic statistics with LLM-based judgments. Through controlled experiments across search agents built on LLMs of varying sizes, we find that increasing model capacity generally alleviates both behaviors, but to markedly different extents. Building on these analysis results, we further propose a lightweight post-hoc reflection framework that converts the proposed evaluation signals into explicit feedback rewards to guide agents' reasoning trajectories. Our findings provide a principled foundation for diagnosing and controlling inefficiencies in search agents.

Retrieval-augmented generation (RAG) enhances large language model (LLM) reasoning by retrieving external documents, but also opens up new attack surfaces. We study knowledge-base poisoning attacks in RAG, where an attacker injects malicious content into the retrieval corpus, which is then surfaced by the retriever and consumed by the LLM during reasoning. Unlike prior work that floods the corpus with poisoned documents, we propose AdversarialCoT, a query-specific attack that poisons only a single document in the corpus. AdversarialCoT first extracts the target LLM's reasoning framework to guide the construction of an initial adversarial chain-of-thought. The adversarial document is iteratively refined through interactions with the LLM, progressively exposing and exploiting critical reasoning vulnerabilities. Experiments on benchmark LLMs show that a single adversarial document can significantly degrade reasoning accuracy, revealing subtle yet impactful weaknesses. Our study exposes security risks in RAG systems and provides actionable insights for designing more robust LLM reasoning pipelines.

Generative recommendation (GR) represents items as discrete semantic identifiers (SIDs) and performs next-item prediction via identifier sequence generation. Most existing methods perform prefix-constrained decoding over a trie, where hard structural constraints lead to locally greedy decisions. An early suboptimal prefix irrevocably restricts subsequent decoding, causing error propagation and suboptimal retrieval. In this work, we propose OneGR, a novel generative recommendation framework that formulates SID prediction as a one-pass, structure-constrained decoding problem. OneGR computes globally consistent, position-wise SID scores in a single neural forward pass, eliminating iterative hypothesis expansion and repeated inference. To address the extreme sparsity of valid SIDs, we encode the space of valid SIDs as a deterministic weighted finite-state transducer and perform A* search with the one-pass scores as additive path costs. This structured decoding strategy explores only reachable valid prefixes while guaranteeing the optimal SID prediction. Experiments on multiple benchmarks show that OneGR consistently outperforms strong baselines.

Real-world user behaviors are often noisy due to factors such as human errors, uncertainty, and behavioral ambiguity, which can lead to degraded recommendation performance. To address this issue, recent approaches widely adopt self-supervised learning (SSL), particularly contrastive learning, by generating perturbed views of user interaction sequences and maximizing their mutual information to improve model robustness. However, these methods heavily rely on their pre-defined static augmentation strategies~(where the augmentation type remains fixed once chosen) to construct augmented views, leading to two critical challenges: (1) the optimal augmentation type can vary significantly across different scenarios; (2) inappropriate augmentations may even degrade recommendation performance, limiting the effectiveness of SSL. To overcome these limitations, we propose an adaptive augmentation framework. We first unify existing basic augmentation operations into a unified formulation via structured transformation matrices. Building on this formulation, we introduce AsarRec, an Adaptive Sequential Augmentation for Robust Sequential Recommendation. To enable stable end-to-end optimization of discrete and strongly constrained augmentations, AsarRec learns to generate transformation matrices by encoding user sequences into probabilistic transition matrices and projecting them into hard semi-doubly stochastic matrices via a differentiable Semi-Sinkhorn algorithm. To ensure that the learned augmentations benefit downstream performance, we jointly optimize three objectives: diversity (encouraging distinct views), semantic invariance (preserving semantic consistency among views), and informativeness (identifying augmentations most beneficial to recommendation). Extensive experiments on four benchmarks under varying noise levels validate the effectiveness of AsarRec, demonstrating its superior robustness and consistent improvements.

As more content generated by large language models (LLMs) floods into the Internet, information retrieval (IR) systems now face the challenge of distinguishing and handling a blend of human-authored and machine-generated texts. Recent studies suggest that neural retrievers may exhibit a preferential inclination toward LLM-generated content, while classic term-based retrievers like BM25 tend to favor human-written documents. This paper investigates the influence of LLM-generated content on term-based retrieval models, which are valued for their efficiency and robust generalization across domains. Our linguistic analysis reveals that LLM-generated texts exhibit smoother high-frequency and steeper low-frequency Zipf slopes, higher term specificity, and greater document-level diversity. These traits are aligned with LLMs being trained to optimize reader experience through diverse and precise expressions. Our study further explores whether term-based retrieval models demonstrate source bias, concluding that these models prioritize documents whose term distributions closely correspond to those of the queries, rather than displaying an inherent source bias. This work provides a foundation for understanding and addressing potential biases in term-based IR systems managing mixed-source content. Code and supplementary material are available at https://github.com/Trustworthy-Information-Access/LLM-Impact-Term-Retrieval.

Large Language Models (LLMs) have demonstrated strong adaptation to unseen tasks. However, their performance in Information Extraction (IE) under unseen schemas remains limited. Actually, IE requires both general abilities for understanding natural language and semantic concepts, and specialized abilities for aligning extracted information to various human-defined schemas. Training an LLM jointly on multiple schemas, or adapting it to a specific schema, often results in performance drops on datasets with other schemas, especially when conflicts arise between schemas. We refer to this phenomenon as the schema alignment tax in this paper. To alleviate this, we propose a schema-adaptive three-phase framework, Identify–Conceptualize–Align (ICA), which enables LLMs to focus on general abilities such as identifying entity and trigger spans and assigning corresponding concepts to them, while delegating schema-specific alignment to lightweight models. Specifically, in the Identification phase, we train an LLM to identify entity and trigger spans on multiple datasets, with cross-dataset annotation to boost span recall. In the Conceptualization phase, the LLM is used to assign semantic concepts to each span. In the Alignment phase, we train different lightweight alignment models to map these concepts to different human-defined schemas. The first two phases are fully reusable across tasks, so adapting to a new schema requires retraining only the alignment model. We evaluate ICA on entity recognition and event detection on 26 commonly adopted datasets with diverse schemas. Experimental results show that our method not only surpasses state-of-the-art approaches under supervised settings, achieving an average F1 improvement of 1.6%, but also attains a remarkable 11.5% average F1 gain on NER and ED in the 10-shot setting.

Implicit hate speech often hides harmful intent behind innocuous wording, metaphors, or hostile tone, making it difficult for detectors that rely on surface cues. We observe that large language models (LLMs) frequently exhibit pseudo-reasoning that shows over-sensitivity to spurious cues while missing the latent semantic units that actually realize hateful intent. We call these units Latent Hate Components (LHCs) and argue they should be the anchors of inference. We propose DuPL, a Dual-Process argumentation framework that centers detection on LHCs. DuPL separates (i) Mining of LHCs via a high-recall Critical Miner and a Confusion Judge that early-exits on clear cases, from (ii) Deliberation of LHCs via component-wise argumentation and a final Integrative Decision. Across IHC, SBIC, and ToxiGen, DuPL consistently outperforms prompt-learning baselines, improving accuracy by +8.36, +4.93, and +3.78 percentage points and macro-F1 by +7.18, +5.00, and +3.99 percentage points, respectively. DuPL also lowers both false positive rate and false negative rate in most settings, indicating balanced mitigation of the two failure modes common in LLM detectors. By explicitly mining and deliberating over LHCs, DuPL turns opaque, uncontrolled reasoning into structured argumentation, yielding more accurate and interpretable decisions for web moderation.

Large language models (LLMs) have demonstrated significant potential in listwise document reranking. Due to their limited context length, LLM-based listwise reranking methods often rely on a sliding window strategy that only processes a small subset of documents at a time. While effective, this approach lacks interactions between documents, increases computational overhead, and results in significant API costs. It is crucial to develop long-context LLMs for enabling the full ranking of all documents in one pass.

A critical limitation of conventional sequential recommendation models (SRMs) is their reliance on observed user-item interaction sequences within a closed-world setting, which hinders their ability to generalize to unseen or infrequent items. Recently, Large Language Models (LLMs) have shown remarkable promise in recommendation systems due to their vast world knowledge and advanced reasoning capabilities. Current research has predominantly explored two approaches: using LLMs to directly generate recommendations and distilling knowledge from LLMs to enhance conventional SRMs. However, these approaches face two major challenges: (1) high inference costs, as they require LLM responses during inference, either for generating predictions or as supplementary input; (2) inadequate distillation of the reasoning process, as existing methods focus mainly on improving embeddings or aligning outputs, without fully integrating LLMs' inherent reasoning capabilities. To address these issues, we propose LCKD-SR, anLLM-drivenCascadedKnowledgeDistillation framework forSequentialRecommendation. In this framework, an LLM, a Teacher SRM, and a Student SRM form a hierarchical distillation structure, enabling an LLM-free inference by using only the Student model. Beyond traditional embedding and ranking distillation, our framework abstracts the LLM's sequential reasoning abilities by identifying key interactions that subsequently guide the Teacher's attention using learnable markers. The Student model, which mirrors the architecture of the Teacher, achieves seamless knowledge alignment from the Teacher across all three aspects. Extensive experiments demonstrate the effectiveness and efficiency of the proposed LCKD-SR, showcasing its scalability to perform multi-level knowledge transfer while enabling LLM-independent inference, thereby overcoming the inference cost and reasoning limitations of existing methods.

Knowledge-intensive language tasks (KILTs) typically require retrieving relevant documents from trustworthy corpora, e.g., Wikipedia, to produce specific answers. Very recently, a pre-trained generative retrieval model for KILTs, named CorpusBrain, was proposed and reached new state-of-the-art retrieval performance. However, most research on KILTs, including CorpusBrain, has predominantly focused on a static document collection, overlooking the dynamic nature of real-world scenarios, where new documents are continuously being incorporated into the source corpus. To address this gap, it is crucial to explore the capability of retrieval models to effectively handle the dynamic retrieval scenario inherent in KILTs. In this work, we first introduce the continual document learning (CDL) task for KILTs and build a novel benchmark dataset named KILT++ based on the original KILT dataset for evaluation. Then, we conduct a comprehensive study of the use of pre-trained CorpusBrain on KILT++. Unlike the promising results in the stationary scenario, CorpusBrain is prone to catastrophic forgetting in the dynamic scenario, hence hampering retrieval performance. To alleviate this issue, we propose CorpusBrain++, a continual generative pre-training framework that enhances the original model along two key dimensions: (i) We employ a backbone-adapter architecture: the dynamic adapter is learned for each downstream KILT task via task-specific pre-training objectives; the backbone parameters that are task-shared are kept unchanged to offer foundational retrieval capacity. (ii) We use an experience replay strategy based on exemplar documents that are similar to new documents, to prevent catastrophic forgetting of old documents. Empirical results demonstrate the effectiveness and efficiency of CorpusBrain++ in comparison to both traditional and generative information retrieval methods.

Recent advances in neural information retrieval models have significantly enhanced these models’ effectiveness across information retrieval tasks. The robustness of these models, which is essential for ensuring their reliability in practice, has also garnered significant attention. With a wide array of research on robust information retrieval being published, we believe it is the opportune moment to consolidate the current status, glean insights from existing methodologies, and lay the groundwork for future development. Robustness of information retrieval is a multifaceted concept and we emphasize the importance of robustness against performance variance, out-of-distribution scenarios, and adversarial attacks. With a focus on out-of-distribution and adversarial robustness, we dissect robustness solutions for dense retrieval models and neural ranking models, respectively, recognizing them as pivotal components of the neural information retrieval pipeline. We provide an in-depth discussion of methods, datasets, and evaluation metrics, shedding light on challenges and future directions in the era of large language models. To accompany this survey, we release three additional resources: (1) a curated list of publications related to robust information retrieval, (2) a tutorial based on this survey, and (3) a heterogeneous benchmark for robust information retrieval, BestIR, that collects all known datasets for evaluating information retrieval systems for robustness. We hope that this study provides useful clues for future research on the robustness of information retrieval models and helps to develop trustworthy IR systems.

Open-domain table retrieval plays a vital role in accessing information from structured formats on the web, yet it remains less explored than text retrieval. Table cells primarily consist of phrases and words, which include numerous entities, such as times, locations, persons, and organizations. While emphasizing entities in text retrieval has been extensively studied, there is a significant lack of research on their applications in table retrieval. In this work, we explore how to leverage entities in tables to improve retrieval performance. We investigate the important role of entities in table retrieval from a statistical perspective and propose an Entity-Centric Alignment framework for Table retrieval (ECAT). Specifically, we use entity types to highlight entities appearing in queries and tables. Then, we propose an entity-driven late interaction paradigm based on entity representations for dense and sparse retrievers, respectively. Our proposed framework is plug-and-play and flexible, making it easy to integrate into existing table retrievers. Empirical results on table retrieval benchmarks, NQ-TABLES and OTT-QA, show that our proposed ECAT is effective in enhancing existing retrievers. Extensive analyses confirm the efficacy of ECAT's different components. Our code and dataset are available at https://github.com/Trustworthy-Information-Access/ECAT.

The advent of universal time series forecasting models has revolutionized zero-shot forecasting across diverse domains, yet the critical role of data diversity in training these models remains underexplored. Existing large-scale time series datasets often suffer from inherent biases and imbalanced distributions, leading to suboptimal model performance and generalization. To address this gap, we introduce BLAST, a novel pre-training corpus designed to enhance data diversity through a balanced sampling strategy. First, BLAST incorporates 321 billion observations from publicly available datasets and employs a comprehensive suite of statistical metrics to characterize time series patterns. Then, to facilitate pattern-oriented sampling, the data is implicitly clustered using grid-based partitioning. Furthermore, by integrating grid sampling and grid mixup techniques, BLAST ensures a balanced and representative coverage of diverse patterns. Experimental results demonstrate that models pre-trained on BLAST achieve state-of-the-art performance with a fraction of the computational resources and training tokens required by existing methods. Our findings highlight the pivotal role of data diversity in improving both training efficiency and model performance for the universal forecasting task.

In book search, relevant book information should be returned in response to a query. Books contain complex, multi-faceted information such as metadata, outlines, and main text, where the outline provides hierarchical information between chapters and sections. Generative retrieval (GR) is a new retrieval paradigm that consolidates corpus information into a single model to generate identifiers of documents that are relevant to a given query. How can GR be applied to book search? Directly applying GR to book search is a challenge due to the unique characteristics of book search: The model needs to retain the complex, multi-faceted information of the book, which increases the demand for labeled data. Splitting book information and treating it as a collection of separate segments for learning might result in a loss of hierarchical information.

Robustness and Effectiveness are critical aspects of developing dense retrieval models for real-world applications. It is known that there is a trade-off between the two. Recent work has addressed scaling laws of effectiveness in dense retrieval, revealing a power-law relationship between effectiveness and the size of models and data. Does robustness follow scaling laws too? If so, can scaling improve both robustness and effectiveness together, or do they remain locked in a trade-off?

Large language models (LLMs) have shown success in knowledge-intensive tasks, including closed-book question answering and entity linking. However, their susceptibility to hallucination undermines their reliability. Retrieval-augmented generation (RAG) partially addresses this issue by combining a retriever to locate relevant documents and a generator to produce responses grounded in the retrieved evidence. Despite its advantages, RAG faces challenges: (i) the structural gap between traditional dense retrievers and autoregressive generators, and (ii) limited generation performance due to insufficient contextual guidance returned by the retriever. To tackle these limitations, we propose MINT, a novel framework that enhances RAG by co-training Retrieval-augMented generatIon and geNeration-augmented reTrieval (GAR). MINT (i) bridges the gap between the retriever and generator using a unified encoder-decoder structure (ii) incorporates an iterative co-training strategy between RAG and GAR, enabling mutual enhancement through pseudo-samples generation, and (iii) introduces three heuristic inference strategies to generate relevant document identifiers and answers. We conduct an empirical study on the KILT benchmark, and MINT is found to yield significant improvements in both retrieval and generation tasks compared with prevailing baselines.

As an important graph pre-training method, Graph Contrastive Learning (GCL) continues to play a crucial role in the ongoing surge of research on graph foundation models or LLM as enhancer for graphs. Traditional GCL optimizes InfoNCE by using augmentations to define self-supervised tasks, treating augmented pairs as positive samples and others as negative. However, this leads to semantically similar pairs being classified as negative, causing significant sampling bias and limiting performance. In this paper, we argue that GCL is essentially a Positive-Unlabeled (PU) learning problem, where the definition of self-supervised tasks should be semantically guided, i.e., augmented samples with similar semantics are considered positive, while others, with unknown semantics, are treated as unlabeled. From this perspective, the key lies in how to extract semantic information. To achieve this, we propose IFL-GCL, using InfoNCE as a "free lunch" to extract semantic information. Specifically, We first prove that under InfoNCE, the representation similarity of node pairs aligns with the probability that the corresponding contrastive sample is positive. Then we redefine the maximum likelihood objective based on the corrected samples, leading to a new InfoNCE loss function. Extensive experiments on both the graph pretraining framework and LLM as an enhancer show significantly improvements of IFL-GCL in both IID and OOD scenarios, achieving up to a 9.05% improvement, validating the effectiveness of semantically guided. Code for IFL-GCL is publicly available at: https://github.com/Camel-Prince/IFL-GCL.

Large language models (LLMs) have achieved remarkable progress across various domains, yet their tendency to generate hallucinations remains a critical barrier to their practical reliability.Confidence calibration addresses this challenge by aligning a model's confidence with its actual accuracy, improving self-evaluation and trustworthiness.However, traditional confidence calibration, operating at response level, are inadequate for long-form generation, which involve complex outputs composed of multiple atomic facts, each with varying confidence, correctness, and relevance to the query.To overcome this limitation, we propose a fact-level confidence calibration framework that evaluates and adjusts confidence at the granularity of individual facts, incorporating both relevance and correctness.This framework identifies finer-grained calibration discrepancies, reduces overconfidence, and reveals confidence variance.Based on this framework, we introduce CARE (Confidence-Aware Fact Correction), a method that leverages high-confidence facts to iteratively refine and correct low-confidence ones.Experimental results demonstrate that our CARE effectively improves the quality of generated content.Our code is available at this link.

While implicit feedback is foundational to modern recommender systems, factors such as human error, uncertainty, and ambiguity in user behavior inevitably introduce significant noise into this feedback, adversely affecting the accuracy and robustness of recommendations. To address this issue, existing methods typically aim to reduce the training weight of noisy feedback or discard it entirely, based on the observation that noisy interactions often exhibit higher losses in the overall loss distribution. However, we identify two key issues: (1) there is a significant overlap between normal and noisy interactions in the overall loss distribution, and (2) this overlap becomes even more pronounced when transitioning from pointwise loss functions (e.g., BCE loss) to pairwise loss functions (e.g., BPR loss). This overlap leads traditional methods to misclassify noisy interactions as normal, and vice versa. To tackle these challenges, we further investigate the loss overlap and find that for a given user, there is a clear distinction between normal and noisy interactions in the user's personal loss distribution. Based on this insight, we propose a resampling strategy to Denoise using the user's Personal Loss distribution, named PLD, which reduces the probability of noisy interactions being optimized. Specifically, during each optimization iteration, we create a candidate item pool for each user and resample the items from this pool based on the user's personal loss distribution, prioritizing normal interactions. Additionally, we conduct a theoretical analysis to validate PLD's effectiveness and suggest ways to further enhance its performance. Extensive experiments conducted on three datasets with varying noise ratios demonstrate PLD's efficacy and robustness.

Entity linking (EL) is a challenging task as it typically requires matching an ambiguous entity mention with its corresponding entity in a knowledge base (KB). The mainstream studies focus on learning and evaluating linking models on the same corpus and obtained significant performance achievement, however, they often overlook the generalization ability to out-of-domain corpus, which is more realistic yet much more challenging. To address this issue, we introduce a novel neural-symbolic model for entity linking, which is inspired by the symbol-manipulation mechanism in human brains. Specifically, we abstract diverse features into unified variables, then combine them using neural operators to capture diverse relevance requirements, and finally aggregate relevance scores through voting. We conduct experiments on eleven benchmark datasets with different types of text, and the results show that our method outperforms nearly all baselines. Notably, the best performance of our method on seven out-of-domain datasets highlights its generalization ability.

Multivariate Time Series (MTS) analysis is crucial to understanding and managing complex systems, such as traffic and energy systems, and a variety of approaches to MTS forecasting have been proposed recently. However, we often observe inconsistent or seemingly contradictory performance findings across different studies. This hinders our understanding of the merits of different approaches and slows down progress. We address the need for means of assessing MTS forecasting proposals reliably and fairly, in turn enabling better exploitation of MTS as seen in different applications. Specifically, we first propose BasicTS+, a benchmark designed to enable fair, comprehensive, and reproducible comparison of MTS forecasting solutions. BasicTS+ establishes a unified training pipeline and reasonable settings, enabling an unbiased evaluation. Second, we identify the heterogeneity across different MTS as an important consideration and enable classification of MTS based on their temporal and spatial characteristics. Disregarding this heterogeneity is a prime reason for difficulties in selecting the most promising technical directions. Third, we apply BasicTS+ along with rich datasets to assess the capabilities of more than 30 MTS forecasting solutions. This provides readers with an overall picture of the cutting-edge research on MTS forecasting.

The first-stage retrieval aims to retrieve a subset of candidate documents from a huge collection both effectively and efficiently. Since various matching patterns can exist between queries and relevant documents, previous work tries to combine multiple retrieval models to find as many relevant results as possible. The constructed ensembles, whether learned independently or jointly, do not care which component model is more suitable to an instance during training. Thus, they cannot fully exploit the capabilities of different types of retrieval models in identifying diverse relevance patterns. Motivated by this observation, in this article, we propose a Mixture-of-Experts (MoE) model consisting of representative matching experts and a novel competitive learning mechanism to let the experts develop and enhance their expertise during training. Specifically, our MoE model shares the bottom layers to learn common semantic representations and uses differently structured upper layers to represent various types of retrieval experts. Our competitive learning mechanism has two stages: (1) a standardized learning stage to train the experts equally to develop their capabilities to conduct relevance matching; (2) a specialized learning stage where the experts compete with each other on every training instance and get rewards and updates according to their performance to enhance their expertise on certain types of samples. Experimental results on retrieval benchmark datasets show that our method significantly outperforms the state-of-the-art baselines in the in-domain and out-of-domain settings.

Community search is a personalized community discovery problem designed to identify densely connected subgraphs containing the query node. Recently, community search in heterogeneous information networks (HINs) has received considerable attention. Existing methods typically focus on modeling relationships in HINs through predefined meta-paths or user-specified relational constraints. However, metapath-based methods are primarily designed to identify single-type communities with nodes of the same type rather than multi-type communities involving nodes of different types. Constraint-based methods require users to have a good understanding of community patterns to define a suitable set of relational constraints, which increases the burden on users. In this paper, we propose FCS-HGNN, a novel method for flexibly identifying both single-type and multi-type communities in HINs. Specifically, FCS-HGNN extracts complementary information from different views and dynamically considers the contribution of each relation instead of treating them equally, thereby capturing more fine-grained heterogeneous information. Furthermore, to improve efficiency on large-scale graphs, we further propose LS-FCS-HGNN, which incorporates i) the neighbor sampling strategy to improve training efficiency, and ii) the depth-based heuristic search strategy to improve query efficiency. We conducted extensive experiments to demonstrate the superiority of our proposed methods over state-of-the-art methods, achieving average improvements of 14.3% and 11.1% on single-type and multi-type communities, respectively.

Sparse Knowledge Graphs (KGs), frequently encountered in real-world applications, contain fewer facts in the form of (head entity, relation, tail entity) compared to more populated KGs. The sparse KG completion task, which reasons answers for given queries in the form of (head entity, relation, ?) for sparse KGs, is particularly challenging due to the necessity of reasoning missing facts based on limited facts. Path-based models, known for excellent explainability, are often employed for this task. However, existing path-based models typically rely on external models to fill in missing facts and subsequently perform path reasoning. This approach introduces unexplainable factors or necessitates meticulous rule design. In light of this, this paper proposes an alternative approach by looking inward instead of seeking external assistance. We introduce a two-stage path reasoning model called LoGRe (Look Globally and Reason) over sparse KGs. LoGRe constructs a relation-path reasoning schema by globally analyzing the training data to alleviate the sparseness problem. Based on this schema, LoGRe then aggregates paths to reason out answers. Experimental results on five benchmark sparse KG datasets demonstrate the effectiveness of the proposed LoGRe model.

Query understanding is an essential part in search systems to improve the recall. Unlike prior works focusing on word expansions, in this paper, we leverage the comprehension ability of LLM to generate detailed queries from a global semantic perspective. To this end, we introduce an efficient GaQR to reformulate a question into several queries using Chain of Thought (CoT) and make it more efficient through knowledge distillation. Specifically, we first prompt a teacher model to generate indicative queries by considering answer generation one step ahead. Then, we filter out low-quality queries by validating the effectiveness of all generated queries in retrieving useful passages. Finally, we distill a student rewriter based on the verified results to improve efficiency. Our experimental results demonstrate that the rewriter improves the retrieval performance by 3% to 15% on the Miracl and NFCorpus datasets and shows good generalisation ability across different retrieval methods. Moreover, the efficiency of the rewriter after knowledge distillation is improved by as much as 5 times. Code is available at https://github.com/youngbeauty250/GaQR.

Community search is a personalized community discovery problem aimed at finding densely-connected subgraphs containing the query vertex. In particular, the search for com-munities with high-importance vertices has recently received a great deal of attention. However, existing works mainly focus on conventional homogeneous networks where vertices are of the same type, but are not applicable to heterogeneous information networks (HINs) composed of multi-typed vertices and different semantic relations, such as bibliographic networks. In this paper, we study the problem of high-importance community search in HINs. A novel community model is introduced, named heterogeneous significant community (HSC), to unravel the closely connected vertices of the same type with high attribute values through multiple semantic relationships. An HSC not only maximizes the exploration of indirect relationships across entities of the anchor-type but incorporates their significance. To search the HSCs, we first develop online algorithms by exploiting both segmented-based meta-path expansion and significance incrernent. Specially, a solution space reuse strategy based on structural nesting is designed to boost the efficiency. In addition, we further devise a two-level index to support searching HSCs in optimal time, based on which a space-efficient compact index is proposed. Extensive experiments on real-world large-scale HINs demonstrate that our solutions are effective and efficient for searching HSCs, and the index-based algorithms are 2–4 orders of magnitude faster than online algorithms.

Entity alignment (EA) identifies equivalent entities that locate in different knowledge graphs (KGs), and has attracted growing research interests over the last few years with the advancement of KG embedding techniques. Although a pile of embedding-based EA frameworks have been developed, they mainly focus on improving the performance of entity representation learning, while largely overlook the subsequent stage that matches$KGs$in entity embedding spaces. Nevertheless, accurately matching entities based on learned entity representations is crucial to the overall alignment performance, as it coordinates individual alignment decisions and determines the global matching result. Hence, it is essential to understand how well existing solutions for matching KGs in entity embedding spaces perform on present benchmarks, as well as their strengths and weaknesses. To this end, in this article we provide a comprehensive survey and evaluation of matching algorithms for KGs in entity embedding spaces in terms of effectiveness and efficiency on both classic settings and new scenarios that better mirror real-life challenges. Based on in-depth analysis, we provide useful insights into the design trade-offs and good paradigms of existing works, and suggest promising directions for future development.

Video Corpus Moment Retrieval (VCMR) is a new video retrieval task aimed at retrieving a relevant moment from a large corpus of untrimmed videos using a text query. The relevance between the video and query is partial, mainly evident in two aspects: (1) Scope: The untrimmed video contains many frames, but not all are relevant to the query. Strong relevance is typically observed only within the relevant moment. (2) Modality: The relevance of the query varies with different modalities. Action descriptions align more with visual elements, while character conversations are more related to textual information. Existing methods often treat all video contents equally, leading to sub-optimal moment retrieval. We argue that effectively capturing the partial relevance between the query and video is essential for the VCMR task. To this end, we propose a Partial Relevance Enhanced Model (PREM) to improve VCMR. VCMR involves two sub-tasks: video retrieval and moment localization. To align with their distinct objectives, we implement specialized partial relevance enhancement strategies. For video retrieval, we introduce a multi-modal collaborative video retriever, generating different query representations for the two modalities by modality-specific pooling, ensuring a more effective match. For moment localization, we propose the focus-then-fuse moment localizer, utilizing modality-specific gates to capture essential content. We also introduce relevant content-enhanced training methods for both retriever and localizer to enhance the ability of model to capture relevant content. Experimental results on TVR and DiDeMo datasets show that the proposed model outperforms the baselines, achieving a new state-of-the-art of VCMR. The code is available at https://github.com/hdy007007/PREM.

Recommender systems play a pivotal role in mitigating information overload in various fields. Nonetheless, the inherent openness of these systems introduces vulnerabilities, allowing attackers to insert fake users into the system’s training data to skew the exposure of certain items, known as poisoning attacks. Adversarial training has emerged as a notable defense mechanism against such poisoning attacks within recommender systems. Existing adversarial training methods apply perturbations of the same magnitude across all users to enhance system robustness against attacks. Yet, in reality, we find that attacks often affect only a subset of users who are vulnerable. These perturbations of indiscriminate magnitude make it difficult to balance effective protection for vulnerable users without degrading recommendation quality for those who are not affected. To address this issue, our research delves into understanding user vulnerability. Considering that poisoning attacks pollute the training data, we note that the higher degree to which a recommender system fits users’ training data correlates with an increased likelihood of users incorporating attack information, indicating their vulnerability. Leveraging these insights, we introduce the Vulnerability-aware Adversarial Training (VAT), designed to defend against poisoning attacks in recommender systems. VAT employs a novel vulnerability-aware function to estimate users’ vulnerability based on the degree to which the system fits them. Guided by this estimation, VAT applies perturbations of adaptive magnitude to each user, not only reducing the success ratio of attacks but also preserving, and potentially enhancing, the quality of recommendations. Comprehensive experiments confirm VAT’s superior defensive capabilities across different recommendation models and against various types of attacks.

How does the graph change if we summarize it by merging nodes? How can we summarize the graph while preserving its spectral characteristics? Graph summarization aims to present a graph in a compact summary graph form while keeping its important structural information. Existing methods primarily focus on preserving the adjacency matrix. In contrast, spectral graph theory provides a powerful tool to describe the characteristics of a graph. In this paper, we propose a novel graph summarization method that preserves the spectral characteristics, including spectral moments and heat traces. We analyze the change of the spectral characteristics after summarization and design a simple yet effective summarization method based on agglomerative clustering. Our approach is extensively evaluated on real-world datasets. The experimental results show that our method excels in preserving the spectral characteristics and obtains better performance on the subsequent graph classification task.

Adversarial ranking attacks have gained increasing attention due to their success in probing vulnerabilities, and, hence, enhancing the robustness, of neural ranking models. Conventional attack methods employ perturbations at a single granularity, e.g., word or sentence level, to target documents. However, limiting perturbations to a single level of granularity may reduce the flexibility of adversarial examples, thereby diminishing the potential threat of the attack. Therefore, we focus on generating high-quality adversarial examples by incorporating multi-granular perturbations. Achieving this objective involves tackling a combinatorial explosion problem, which requires identifying an optimal combination of perturbations across all possible levels of granularity, positions, and textual pieces. To address this challenge, we transform the multi-granular adversarial attack into a sequential decision-making process, where perturbations in the next attack step build on the perturbed document in the current attack step. Since the attack process can only access the final state without direct intermediate signals, we use reinforcement learning to perform multi-granular attacks. During the reinforcement learning process, two agents work cooperatively to identify multi-granular vulnerabilities as attack targets and organize perturbation candidates into a final perturbation sequence. Experimental results show that our attack method surpasses prevailing baselines in both attack effectiveness and imperceptibility.

Graph contrastive learning (GCL), standing as the dominant paradigm in the realm of graph pre-training, has yielded considerable progress. Nonetheless, its capacity for out-of-distribution (OOD) generalization has been relatively underexplored. In this work, we point out that the traditional optimization of InfoNCE in GCL restricts the cross-domain pairs only to be negative samples, which inevitably enlarges the distribution gap between different domains. This violates the requirement of domain invariance under OOD scenario and consequently impairs the model's OOD generalization performance. To address this issue, we propose a novel strategy ''Negative as Positive'', where the most semantically similar cross-domain negative pairs are treated as positive during GCL. Our experimental results, spanning a wide array of datasets, confirm that this method substantially improves the OOD generalization performance of GCL.

With the application of generation models, internet is increasingly inundated with AI-generated content (AIGC), causing both real and AI-generated content indexed in corpus for search. This paper explores the impact of AI-generated images on text-image search in this scenario. Firstly, we construct a benchmark consisting of both real and AI-generated images for this study. In this benchmark, AI-generated images possess visual semantics sufficiently similar to real images. Experiments on this benchmark reveal that text-image retrieval models tend to rank the AI-generated images higher than the real images, even though the AI-generated images do not exhibit more visually relevant semantics to the queries than real images. We call this bias as invisible relevance bias. This bias is detected across retrieval models with different training data and architectures. Further exploration reveals that mixing AI-generated images into the training data of retrieval models exacerbates the invisible relevance bias. These problems cause a vicious cycle in which AI-generated images have a higher chance of exposing from massive data, which makes them more likely to be mixed into the training of retrieval models and such training makes the invisible relevance bias more and more serious. To mitigate this bias and elucidate the potential causes of the bias, firstly, we propose an effective method to alleviate this bias. Subsequently, we apply our proposed debiasing method to retroactively identify the causes of this bias, revealing that the AI-generated images induce the image encoder to embed additional information into their representation. This information makes the retriever estimate a higher relevance score. We conduct experiments to support this assertion.

Retrieval-augmented generation (RAG) is considered to be a promising approach to alleviate the hallucination issue of large language models (LLMs), and it has received widespread attention from researchers recently. Due to the limitation in the semantic understanding of retrieval models, the success of RAG heavily lies on the ability of LLMs to identify passages with utility. Recent efforts have explored the ability of LLMs to assess the relevance of passages in retrieval, but there has been limited work on evaluating the utility of passages in supporting question answering.

Sequential recommender systems stand out for their ability to capture users' dynamic interests and the patterns of item transitions. However, the inherent openness of sequential recommender systems renders them vulnerable to poisoning attacks, where fraudsters are injected into the training data to manipulate learned patterns. Traditional defense methods predominantly depend on predefined assumptions or rules extracted from specific known attacks, limiting their generalizability to unknown attacks. To solve the above problems, considering the rich open-world knowledge encapsulated in Large Language Models (LLMs), we attempt to introduce LLMs into defense methods to broaden the knowledge beyond limited known attacks. We propose LoRec, an innovative framework that employs LLM-Enhanced Calibration to strengthen the robustness of sequential Recommender systems against poisoning attacks. LoRec integrates an LLM-enhanced CalibraTor (LCT) that refines the training process of sequential recommender systems with knowledge derived from LLMs, applying a user-wise reweighting to diminish the impact of attacks. Incorporating LLMs' open-world knowledge, the LCT effectively converts the limited, specific priors or rules into a more general pattern of fraudsters, offering improved defenses against poisons. Our comprehensive experiments validate that LoRec, as a general framework, significantly strengthens the robustness of sequential recommender systems.

Dense retrieval methods have been mostly focused on unstructured text and less attention has been drawn to structured data with various aspects, e.g., products with aspects such as category and brand. Recent work has proposed two approaches to incorporate the aspect information into item representations for effective retrieval by predicting the values associated with the item aspects. Despite their efficacy, they treat the values as isolated classes (e.g., "Smart Homes", "Home, Garden & Tools", and "Beauty & Health") and ignore their fine-grained semantic relation. Furthermore, they either enforce the learning of aspects into the CLS token, which could confuse it from its designated use for representing the entire content semantics, or learn extra aspect embeddings only with the value prediction objective, which could be insufficient especially when there are no annotated values for an item aspect.

Making the contents generated by Large Language Model (LLM), accurate, credible and traceable is crucial, especially in complex knowledge-intensive tasks that require multi-step reasoning and each step needs knowledge to solve. Retrieval-augmented generation is good potential to solve this problem. However, where and how to introduce Information Retrieval (IR) to LLM is a big challenge. Previous work has the problems that wrong knowledge retrieved by IR misleads the LLM and interaction between IR and LLM breaks the reasoning chain of LLM. This paper proposes a novel framework named Search-in-the-Chain (SearChain) for the interaction between LLM and IR to solve the challenges. First, LLM generates the reasoning chain named Chain-of-Query (CoQ) where each node consists of an IR-oriented query-answer pair. Second, IR verifies the answer of each node of CoQ. It corrects the answer that is not consistent with the retrieved information when IR gives high confidence, which improves the credibility. Third, LLM can indicate its missing knowledge in CoQ and rely on IR to provide this knowledge to LLM. These operations improve the accuracy in terms of reasoning and knowledge. Finally, SearChain generates the reasoning process and marks references to supporting documents for each reasoning step, which improves traceability. Interaction with IR in SearChain forms a novel reasoning path based on a tree, which enables LLM to dynamically modify the direction of reasoning. Experiments show that SearChain outperforms state-of-the-art baselines on complex knowledge-intensive tasks including multi-hop Q&A, slot filling, fact checking, and long-form Q&A.

The results of information retrieval (IR) are usually presented in the form of a ranking list of candidate documents, such as web search for humans and retrieval-augmented generation for large language models (LLMs). List-aware retrieval aims to capture the list-level contextual features to return a better list, mainly including reranking and truncation. Reranking finely re-scores the documents in the list. Truncation dynamically determines the cut-off point of the ranked list to achieve the trade-off between overall relevance and avoiding misinformation from irrelevant documents. Previous studies treat them as two separate tasks and model them separately. However, the separation is not optimal. First, it is hard to share the contextual information of the ranking list between the two tasks. Second, the separate pipeline usually meets the error accumulation problem, where the small error from the reranking stage can largely affect the truncation stage. To solve these problems, we propose a Reranking-Truncation joint model (GenRT) that can perform the two tasks concurrently. GenRT integrates reranking and truncation via a generative paradigm based on an encoder-decoder architecture with novel loss functions for joint optimization to learn both tasks. Sharing parameters by the joint model is conducive to making full use of the common modeling information of the two tasks. Besides, the two tasks are performed concurrently and co-optimized to solve the error accumulation problem between separate stages. Experiments on public learning-to-rank benchmarks and open-domain Q&A tasks show that our method achieves SOTA performance on both reranking and truncation tasks for web search and retrieval-augmented LLMs.

Graph summarization is a useful tool for analyzing large-scale graphs. Some works tried to preserve original node embeddings encoding rich structural information of nodes on the summary graph. However, their algorithms are designed heuristically and not theoretically guaranteed. In this article, we theoretically study the problem of preserving node embeddings on summary graph. We prove that three matrix-factorization-based node embedding methods of the original graph can be approximated by that of the summary graph, and we propose a novel graph summarization method, named HCSumm , based on this analysis. Extensive experiments are performed on real-world datasets to evaluate the effectiveness of our proposed method. The experimental results show that our method outperforms the state-of-the-art methods in preserving node embeddings.

How can we effectively detect fake reviews or fraudulent links on a website? How can we spot communities that suddenly appear based on users’ interactions? And how can we efficiently find the minimum cut in a large graph? All of these are related to the finding of dense subgraphs, a significant primitive problem in graph analysis with extensive applications across various domains. In this paper, we focus on formulating the problem of the densest subgraph detection and theoretically compare and contrast several correlated problems. Moreover, we propose a unified framework,GenDS, for the densest subgraph detection, provide some theoretical analysis based on the network flow and spectral graph theory, and devise simple and computationally efficient algorithms,SpecGDSandGepGDS, to solve it by leveraging the spectral properties and greedy search. We conduct thorough experiments on 40 real-world networks with up to 1.47 billion edges from various domains. We demonstrate that ourSpecGDSyields up to$58.6 \ \times$speedup and achieves better or approximately equal-quality solutions for the densest subgraph detection compared to the baselines.GepGDSalso reveals some properties of generalized eigenvalue problems for theGenDS. Also, our methods scale linearly with the graph size and are proven effective in applications such as finding collaborations that appear suddenly in an extensive, time-evolving co-authorship network.

Despite achieving great success, graph neural networks (GNNs) are vulnerable to adversarial attacks. Existing defenses focus on developing adversarial training or model modification. In this paper, we propose and formulategraph adversarial immunization, i.e., vaccinating part of graph structure to improve certifiable robustness of graph against any admissible adversarial attack. We first propose edge-level immunization to vaccinate node pairs. Unfortunately, such edge-level immunization cannot defend against emerging node injection attacks, since it only immunizes existing node pairs. To this end, we further propose node-level immunization. To avoid computationally intensive combinatorial optimization associated with adversarial immunization, we developAdvImmune-EdgeandAdvImmune-Nodealgorithms to effectively obtain the immune node pairs or nodes. Extensive experiments demonstrate the superiority of AdvImmune methods. In particular, AdvImmune-Node remarkably improves the ratio of robust nodes by 79$\%$, 294$\%$, and 100$\%$, after immunizing only 5$\%$of nodes. Furthermore, AdvImmune methods show excellent defensive performance against various attacks, outperforming state-of-the-art defenses. To the best of our knowledge, this is the first attempt to improve certifiable robustness from graph data perspective without losing performance on clean graphs, providing new insights into graph adversarial learning.

Recently, a novel generative retrieval (GR) paradigm has been proposed, where a single sequence-to-sequence model is learned to directly generate a list of relevant document identifiers (docids) given a query. Existing GR models commonly employ maximum likelihood estimation (MLE) for optimization: This involves maximizing the likelihood of a single relevant docid given an input query, with the assumption that the likelihood for each docid is independent of the other docids in the list. We refer to these models as the pointwise approach in this article. While the pointwise approach has been shown to be effective in the context of GR, it is considered sub-optimal due to its disregard for the fundamental principle that ranking involves making predictions about lists. In this article, we address this limitation by introducing an alternative listwise approach, which empowers the GR model to optimize the relevance at the docid list level. Specifically, we view the generation of a ranked docid list as a sequence learning process: At each step, we learn a subset of parameters that maximizes the corresponding generation likelihood of theith docid given the (preceding) topi-1 docids. To formalize the sequence learning process, we design a positional conditional probability for GR. To alleviate the potential impact of beam search on the generation quality during inference, we perform relevance calibration on the generation likelihood of model-generated docids according to relevance grades. We conduct extensive experiments on representative binary and multi-graded relevance datasets. Our empirical results demonstrate that our method outperforms state-of-the-art GR baselines in terms of retrieval performance.

Information retrieval aims to find information that meets users’ needs from the corpus. Different needs correspond to different IR tasks such as document retrieval, open-domain question answering, retrieval-based dialogue, and so on, while they share the same schema to estimate the relationship between texts. It indicates that a good IR model can generalize to different tasks and domains. However, previous studies indicate that state-of-the-art neural information retrieval (NIR) models, e.g., pre-trained language models (PLMs) are hard to generalize. It is mainly because the end-to-end fine-tuning paradigm makes the model overemphasize task-specific signals and domain biases but loses the ability to capture generalized essential signals. To address this problem, we propose a novel NIR training framework named NIR-Prompt for retrieval and reranking stages based on the idea of decoupling signal capturing and combination. NIR-Prompt exploits Essential Matching Module (EMM) to capture the essential matching signals and gets the description of tasks by Matching Description Module (MDM). The description is used as task-adaptation information to combine the essential matching signals to adapt to different tasks. Experiments under in-domain multi-task, out-of-domain multi-task, and new task adaptation settings show that NIR-Prompt can improve the generalization of PLMs in NIR for both retrieval and reranking stages compared with baselines.

Current natural language understanding (NLU) models have been continuously scaling up, both in terms of model size and input context, introducing more hidden and input neurons. While this generally improves performance on average, the extra neurons do not yield a consistent improvement for all instances. This is because some hidden neurons are redundant, and the noise mixed in input neurons tends to distract the model. Previous work mainly focuses on extrinsically reducing low-utility neurons by additional post- or pre-processing, such as network pruning and context selection, to avoid this problem. Beyond that, can we make the model reduce redundant parameters and suppress input noise by intrinsically enhancing the utility of each neuron? If a model can efficiently utilize neurons, no matter which neurons are ablated (disabled), the ablated submodel should perform no better than the original full model. Based on such a comparison principle between models, we propose a cross-model comparative loss for a broad range of tasks. Comparative loss is essentially a ranking loss on top of the task-specific losses of the full and ablated models, with the expectation that the task-specific loss of the full model is minimal. We demonstrate the universal effectiveness of comparative loss through extensive experiments on 14 datasets from three distinct NLU tasks based on five widely used pre-trained language models and find it particularly superior for models with few parameters or long input.

The data-hungry problem, characterized by insufficiency and low-quality of data, poses obstacles for deep learning models. Transfer learning has been a feasible way to transfer knowledge from high-quality external data of source domains to limited data of target domains, which follows a domain-level knowledge transfer to learn a shared posterior distribution. However, they are usually built on strong assumptions, e.g., the domain invariant posterior distribution, which is usually unsatisfied and may introduce noises, resulting in poor generalization ability on target domains. Inspired by Graph Neural Networks (GNNs) that aggregate information from neighboring nodes, we redefine the paradigm as learning a knowledge-enhanced posterior distribution for target domains, namely Knowledge Bridge Learning (KBL). KBL first learns the scope of knowledge transfer by constructing a Bridged-Graph that connects knowledgeable samples to each target sample and then performs sample-wise knowledge transfer via GNNs.KBL is free from strong assumptions and is robust to noises in the source data. Guided by KBL, we propose the Bridged-GNN including an Adaptive Knowledge Retrieval module to build Bridged-Graph and a Graph Knowledge Transfer module. Comprehensive experiments on both un-relational and relational data-hungry scenarios demonstrate the significant improvements of Bridged-GNN compared with SOTA methods

First-stage retrieval is a critical task that aims to retrieve relevant document candidates from a large-scale collection. While existing retrieval models have achieved impressive performance, they are mostly studied on static data sets, ignoring that in the real-world, the data on the Web is continuously growing with potential distribution drift. Consequently, retrievers trained on static old data may not suit new-coming data well and inevitably produce sub-optimal results. In this work, we study lifelong learning for first-stage retrieval, especially focusing on the setting where the emerging documents are unlabeled since relevance annotation is expensive and may not keep up with data emergence. Under this setting, we aim to develop model updating with two goals: (1) to effectively adapt to the evolving distribution with the unlabeled new-coming data, and (2) to avoid re-inferring all embeddings of old documents to efficiently update the index each time the model is updated.

Generative retrieval (GR) directly predicts the identifiers of relevant documents (i.e., docids) based on a parametric model. It has achieved solid performance on many ad-hoc retrieval tasks. So far, these tasks have assumed a static document collection. In many practical scenarios, however, document collections are dynamic, where new documents are continuously added to the corpus. The ability to incrementally index new documents while preserving the ability to answer queries with both previously and newly indexed relevant documents is vital to applying GR models. In this paper, we address this practical continual learning problem for GR. We put forward a novel Continual-LEarner for generatiVE Retrieval (CLEVER) model and make two major contributions to continual learning for GR: (i) To encode new documents into docids with low computational cost, we present Incremental Product Quantization, which updates a partial quantization codebook according to two adaptive thresholds; and (ii) To memorize new documents for querying without forgetting previous knowledge, we propose a memory-augmented learning mechanism, to form meaningful connections between old and new documents. Empirical results demonstrate the effectiveness and efficiency of the proposed model.

The mainstream of data-driven abstractive summarization models tends to explore the correlations rather than the causal relationships. Among such correlations, there can be spurious ones which suffer from the language prior learned from the training corpus and therefore undermine the overall effectiveness of the learned model. To tackle this issue, we introduce a Structural Causal Model (SCM) to induce the underlying causal structure of the summarization data. We assume several latent causal factors and non-causal factors, representing the content and style of the document and summary. Theoretically, we prove that the latent factors in our SCM can be identified by fitting the observed training data under certain conditions. On the basis of this, we propose a Causality Inspired Sequence-to-Sequence model (CI-Seq2Seq) to learn the causal representations that can mimic the causal factors, guiding us to pursue causal information for summary generation. The key idea is to reformulate the Variational Auto-encoder (VAE) to fit the joint distribution of the document and summary variables from the training corpus. Experimental results on two widely used text summarization datasets demonstrate the advantages of our approach.

Most existing methods that address out-of-distribution (OOD) generalization for node classification on graphs primarily focus on a specific type of data biases, such as label selection bias or structural bias. However, anticipating the type of bias in advance is extremely challenging, and designing models solely for one specific type may not necessarily improve overall generalization performance. Moreover, limited research has focused on the impact of mixed biases, which are more prevalent and demanding in real-world scenarios. To address these limitations, we propose a novel Causality and Independence Enhancement (CIE) framework, applicable to various graph neural networks (GNNs). Our approach estimates causal and spurious features at the node representation level and mitigates the influence of spurious correlations through the backdoor adjustment. Meanwhile, independence constraint is introduced to improve the discriminability and stability of causal and spurious features in complex biased environments. Essentially, CIE eliminates different types of data biases from a unified perspective, without the need to design separate methods for each bias as before. To evaluate the performance under specific types of data biases, mixed biases, and low-resource scenarios, we conducted comprehensive experiments on five publicly available datasets. Experimental results demonstrate that our approach CIE not only significantly enhances the performance of GNNs but outperforms state-of-the-art debiased node classification methods.

Neural ranking models (NRMs) and dense retrieval (DR) models have given rise to substantial improvements in overall retrieval performance. In addition to their effectiveness, and motivated by the proven lack of robustness of deep learning-based approaches in other areas, there is growing interest in the robustness of deep learning-based approaches to the core retrieval problem. Adversarial attack methods that have so far been developed mainly focus on attacking NRMs, with very little attention being paid to the robustness of DR models.

Privacy concerns on social networks have received extensive attention in recent years. The task of user identity linkage (UIL), which aims to identify corresponding users across different social networks, poses a threat to privacy if applied unethically. Sensitive user information would be inferred with cross-network identity linkages. A feasible solution to this issue is to design an adversarial strategy that degrades the matching performance of UIL models. Nevertheless, most of the current adversarial attacks on graphs are tailored towards models working within a single network, failing to account for the challenges presented by cross-network learning tasks such as UIL. Also, in real-world scenarios, the adversarial strategy against UIL has more constraints as service providers can only add perturbations to their own networks. To tackle these challenges, this paper proposes a novel poisoning strategy to prevent nodes in a target network from being linked to other networks by UIL algorithms. Specifically, the UIL problem is formalized in the kernelized topology consistency perspective, and the objective is formulated as maximizing the structural variations in the target network before and after modifications. To achieve this, a novel graph kernel is defined based on earth mover's distance (EMD) in the edge-embedding space. In terms of efficiency, a fast attack strategy is proposed using greedy searching and a lower bound approximation of EMD. Results on three real-world datasets demonstrate that the proposed method outperforms six baselines and reaches a balance between effectiveness and imperceptibility while being efficient.

Social network alignment is widely applied in web applications for identifying corresponding nodes across different networks, such as linking users across two social networks. Existing methods for social network alignment primarily rely on alignment consistency, assuming that nodes with similar attributes and neighbors are more likely to be aligned. However, distributional discrepancies in node attributes and neighbors across different networks would bring biases in alignment consistency, leading to inferior alignment performance. To address this issue, we conduct a causal analysis of alignment consistency. Based on this analysis, we propose a novel model called CANA that uses causal inference approaches to mitigate biases and enhance social network alignment. Firstly, we disentangle observed node attributes into endogenous features and exogenous features with multi-task learning. Only endogenous features are retained to overcome node attribute discrepancies. To eliminate biases caused by neighbors discrepancies, we propose causal-aware attention mechanisms and integrate them in graph neural network to reweight contributions of different neighbors in alignment consistency comparison. Additionally, backdoor adjustment is applied to reduce confounding effects and estimate unbiased alignment probability. Through experimental evaluation on four real-world datasets, the proposed method demonstrates superior performance in terms of alignment accuracy and top-k hits precision.

Graph domain adaptation models have become instrumental in addressing cross-network learning problems due to their ability to transfer abundant label and structural knowledge from source graphs to target graphs. A crucial step in transfer involves measuring domain discrepancy, which refers to distribution shifts between graphs from source and target domains. While conventional models simply provide a node-level measurement, exploiting information from different levels of network hierarchy is intuitive. As each hierarchical level characterizes distinct and meaningful properties or functionalities of the original graph, integrating domain discrepancy based on such hierarchies should contribute to a more precise domain discrepancy measurement. Moreover, class conditional distribution shift is often overlooked in node classification tasks, which could potentially lead to sub-optimal performance. To address the above limitations, we propose a new graph domain adaptation model and apply it to cross-network node classification tasks. Specifically, a hierarchical pooling model to extract meaningful and adaptive hierarchical structures is designed, where both marginal and class conditional distribution shifts on each hierarchical level are jointly minimized. The effectiveness is demonstrated through theoretical analysis and experimental studies across various datasets.

Graph domain adaptation models are widely adopted in cross-network learning tasks to transfer labeling or structural knowledge. Currently, there mainly exist two limitations in evaluating graph domain adaptation models. On one side, they are primarily tested for the specific cross-network node classification task, leaving tasks at edge-level and graph-level largely under-explored. Moreover, they are primarily examined in limited scenarios, such as social networks or citation networks, needing more validation in richer scenarios. As comprehensively assessing models could enhance model practicality in real-world applications, we propose a benchmark known as OpenGDA. It provides abundant pre-processed and unified datasets for different types of tasks (node, edge, graph). They originate from diverse scenarios, covering web information systems, urban systems and natural systems. Furthermore, it integrates state-of-the-art models with standardized and end-to-end pipelines. Overall, OpenGDA provides a user-friendly, scalable and reproducible benchmark for evaluating graph domain adaptation models. The benchmark experiments highlight the challenges of applying GDA models to real-world applications with consistent good performance, and they potentially provide insights to future research. As an emerging project, OpenGDA will be regularly updated with new datasets and models. It could be accessed from https://github.com/Skyorca/OpenGDA.

Recently, a new paradigm called Differentiable Search Index (DSI) has been proposed for document retrieval, wherein a sequence-to-sequence model is learned to directly map queries to relevant document identifiers. The key idea behind DSI is to fully parameterize traditional ''index-retrieve'' pipelines within a single neural model, by encoding all documents in the corpus into the model parameters. In essence, DSI needs to resolve two major questions: (1) how to assign an identifier to each document, and (2) how to learn the associations between a document and its identifier. In this work, we propose a Semantic-Enhanced DSI model (SE-DSI) motivated by Learning Strategies in the area of Cognitive Psychology. Our approach advances original DSI in two ways: (1) For the document identifier, we take inspiration from Elaboration Strategies in human learning. Specifically, we assign each document an Elaborative Description based on the query generation technique, which is more meaningful than a string of integers in the original DSI; and (2) For the associations between a document and its identifier, we take inspiration from Rehearsal Strategies in human learning. Specifically, we select fine-grained semantic features from a document as Rehearsal Contents to improve document memorization. Both the offline and online experiments show improved retrieval performance over prevailing baselines.

Unsupervised representation learning for dynamic graphs has attracted a lot of research attention in recent years. Compared with static graph, the dynamic graph is a comprehensive embodiment of both the intrinsic stable characteristics of nodes and the time-related dynamic preference. However, existing methods generally mix these two types of information into a single representation space, which may lead to poor explanation, less robustness, and a limited ability when applied to different downstream tasks. To solve the above problems, in this paper, we propose a novel disenTangled representation learning framework for discrete-time Dynamic graphs, namely DyTed. We specially design a temporal-clips contrastive learning task together with a structure contrastive learning to effectively identify the time-invariant and time-varying representations respectively. To further enhance the disentanglement of these two types of representation, we propose a disentanglement-aware discriminator under an adversarial learning framework from the perspective of information theory. Extensive experiments on Tencent and five commonly used public datasets demonstrate that DyTed, as a general framework that can be applied to existing methods, achieves state-of-the-art performance on various downstream tasks, as well as be more robust against noise.

Knowledge-intensive language tasks (KILTs) benefit from retrieving high-quality relevant contexts from large external knowledge corpora. Learning task-specific retrievers that return relevant contexts at an appropriate level of semantic granularity, such as a document retriever, passage retriever, sentence retriever, and entity retriever, may help to achieve better performance on the end-to-end task. But a task-specific retriever usually has poor generalization ability to new domains and tasks, and it may be costly to deploy a variety of specialised retrievers in practice.

Neural ranking models (NRMs) have attracted considerable attention in information retrieval. Unfortunately, NRMs may inherit the adversarial vulnerabilities of general neural networks, which might be leveraged by black-hat search engine optimization practitioners. Recently, adversarial attacks against NRMs have been explored in the paired attack setting, generating an adversarial perturbation to a target document for a specific query. In this paper, we focus on a more general type of perturbation and introduce the topic-oriented adversarial ranking attack task against NRMs, which aims to find an imperceptible perturbation that can promote a target document in ranking for a group of queries with the same topic. We define both static and dynamic settings for the task and focus on decision-based black-box attacks. We propose a novel framework to improve topic-oriented attack performance based on a surrogate ranking model. The attack problem is formalized as a Markov decision process (MDP) and addressed using reinforcement learning. Specifically, a topic-oriented reward function guides the policy to find a successful adversarial example that can be promoted in rankings to as many queries as possible in a group. Experimental results demonstrate that the proposed framework can significantly outperform existing attack strategies, and we conclude by re-iterating that there exist potential risks for applying NRMs in the real world.

Recommender systems often suffer from popularity bias, where popular items are overly recommended while sacrificing unpopular items. Existing researches generally focus on ensuring the number of recommendations (exposure) of each item is equal or proportional, using inverse propensity weighting, causal intervention, or adversarial training. However, increasing the exposure of unpopular items may not bring more clicks or interactions, resulting in skewed benefits and failing in achieving real reasonable popularity debiasing. In this paper, we propose a new criterion for popularity debiasing, i.e., in an unbiased recommender system, both popular and unpopular items should receive Interactions Proportional to the number of users who Like it, namely IPL criterion. Under the guidance of the criterion, we then propose a debiasing framework with IPL regularization term which is theoretically shown to achieve a win-win situation of both popularity debiasing and recommendation performance. Experiments conducted on four public datasets demonstrate that when equipping two representative collaborative filtering models with our framework, the popularity bias is effectively alleviated while maintaining the recommendation performance.

Graphs consisting of vocal nodes ("the vocal minority") and silent nodes ("the silent majority"), namely VS-Graph, are ubiquitous in the real world. The vocal nodes tend to have abundant features and labels. In contrast, silent nodes only have incomplete features and rare labels, e.g., the description and political tendency of politicians (vocal) are abundant while not for ordinary civilians (silent) on the twitter’s social network. Predicting the silent majority remains a crucial yet challenging problem. However, most existing Graph Neural Networks (GNNs) assume that all nodes belong to the same domain, without considering the missing features and distribution-shift between domains, leading to poor ability to deal with VS-Graph. To combat the above challenges, we propose Knowledge Transferable Graph Neural Network (KTGNN), which models distribution-shifts during message passing and learns representation by transferring knowledge from vocal nodes to silent nodes. Specifically, we design the domain-adapted "feature completion and message passing mechanism" for node representation learning while preserving domain difference. And a knowledge transferable classifier based on KL-divergence is followed. Comprehensive experiments on real-world scenarios (i.e., company financial risk assessment and political elections) demonstrate the superior performance of our method. Our source code has been open-sourced1.

Dense subtensor detection gains remarkable success in spotting anomalies and fraudulent behaviors for multi-aspect data (i.e., tensors), like in social media and event streams. Existing methods detect the densest subtensors flatly and separately, with the underlying assumption that those subtensors are exclusive. However, many real-world tensors usually present hierarchical properties, e.g., the core-periphery structure and dynamic communities in networks. It is also unexplored how to fuse the prior knowledge into dense pattern detection to capture the local behavior. In this article, we propose CatchCore , a novel framework to efficiently find the hierarchical dense subtensors. We first design a unified metric for dense subtensor detection, which can be optimized with gradient-based methods. With the proposed metric, CatchCore detects hierarchical dense subtensors through the hierarchy-wise alternative optimization and finds local dense patterns concerning some items in a query manner. Finally, we utilize the minimum description length principle to measure the quality of detection results and select the optimal hierarchical dense subtensors. Extensive experiments on synthetic and real-world datasets demonstrate that CatchCore outperforms the top competitors in accuracy for detecting dense subtensors and anomaly patterns, like network attacks. Additionally, CatchCore successfully identifies a hierarchical researcher co-authorship group with intense interactions in the DBLP dataset; it can also capture core collaboration and multi-hop relations around some query objects. Meanwhile, CatchCore also scales linearly with all aspects of tensors.

Besides entity-centric knowledge, usually organized as Knowledge Graph (KG), events are also an essential kind of knowledge in the world, which trigger the spring up of event-centric knowledge representation form like Event KG (EKG). It plays an increasingly important role in many downstream applications, such as search, question-answering, recommendation, financial quantitative investments, and text generation. This paper provides a comprehensive survey of EKG from history, ontology, instance, and application views. Specifically, to characterize EKG thoroughly, we focus on its history, definitions, schema induction, acquisition, related representative graphs/systems, and applications. The development processes and trends are studied therein. We further summarize prospective directions to facilitate future research on EKG.

With the overwhelming popularity of Knowledge Graphs (KGs), researchers have poured attention to link prediction to fill in missing facts for a long time. However, they mainly focus on link prediction on binary relational data, where facts are usually represented as triples in the form of (head entity, relation, tail entity). In practice, n-ary relational facts are also ubiquitous. When encountering such facts, existing studies usually decompose them into triples by introducing a multitude of auxiliary virtual entities and additional triples. These conversions result in the complexity of carrying out link prediction on n-ary relational data. It has even proven that they may cause loss of structure information. To overcome these problems, in this paper, we represent each n-ary relational fact as a set of its role and role-value pairs. We then propose a method called NaLP to conduct link prediction on n-ary relational data, which explicitly models the relatedness of all the role and role-value pairs in an n-ary relational fact. We further extend NaLP by introducing type constraints of roles and role-values without any external type-specific supervision, and proposing a more reasonable negative sampling mechanism. Experimental results validate the effectiveness and merits of the proposed methods.

Time series data naturally exist in many domains including medical data analysis, infrastructure sensor monitoring, and motion tracking. However, a very small portion of anomalous time series can be observed, comparing to the whole data. Most existing approaches are based on the supervised classification model requiring representative labels for anomaly class(es), which is challenging in real-world problems. So can we learn how to detect anomalous time ticks in an effective yet efficient way, given mostly normal time series data? Therefore, we propose an unsupervised reconstruction model named BeatGAN which learns to detect anomalies based on normal data, or data which majority of samples are normal. BeatGAN provides a framework to adversarially learn to reconstruct, which can cooperate with both 1-d CNN and RNN. Rarely observed anomalies can result in larger reconstruction errors, which are then detected based on extreme value theory. Moreover, data augmentation with dynamic time warping regularizes reconstruction and provides robustness. In the experiments, effectiveness and sensitivity are studied in both synthetic data and various real-world time series. BeatGAN achieves better accuracy and fast inference.

Entity alignment (EA) identifies equivalent entities that locate in different knowledge graphs (KGs), and has attracted growing research interests over the last few years with the advancement of KG embedding techniques. Although a pile of embedding-based EA frameworks have been developed, they mainly focus on improving the performance ofentity representation learning, while largely overlook the subsequent stage thatmatches KGs in entity embedding spaces. Nevertheless, accurately matching entities based on learned entity representations is crucial to the overall alignment performance, as it coordinates individual alignment decisions and determines the global matching result. Hence, it is essential to understand how well existing solutions for matching KGs in entity embedding spaces perform on present benchmarks, as well as their strengths and weaknesses. To this end, in this article we provide a comprehensive survey and evaluation of matching algorithms for KGs in entity embedding spaces in terms of effectiveness and efficiency on both classic settings and new scenarios that better mirror real-life challenges. Based on in-depth analysis, we provide useful insights into the design trade-offs and good paradigms of existing works, and suggest promising directions for future development.

Recently, we have witnessed the bloom of neural ranking models in the information retrieval (IR) field. So far, much effort has been devoted to developing effective neural ranking models that can generalize well on new data. There has been less attention paid to the robustness perspective. Unlike the effectiveness, which is about the average performance of a system under normal purpose, robustness cares more about the system performance in the worst case or under malicious operations instead. When a new technique enters into the real-world application, it is critical to know not only how it works in average, but also how would it behave in abnormal situations. So, we raise the question in this work: Are neural ranking models robust? To answer this question, first, we need to clarify what we refer to when we talk about the robustness of ranking models in IR. We show that robustness is actually a multi-dimensional concept and there are three ways to define it in IR: (1) the performance variance under the independent and identically distributed (I.I.D.) setting; (2) the out-of-distribution (OOD) generalizability ; and (3) the defensive ability against adversarial operations. The latter two definitions can be further specified into two different perspectives, respectively, leading to five robustness tasks in total. Based on this taxonomy, we build corresponding benchmark datasets, design empirical experiments, and systematically analyze the robustness of several representative neural ranking models against traditional probabilistic ranking models and learning-to-rank (LTR) models. The empirical results show that there is no simple answer to our question. While neural ranking models are less robust against other IR models in most cases, some of them can still win two out of five tasks. This is the first comprehensive study on the robustness of neural ranking models. We believe the way we study the robustness as well as our findings would be beneficial to the IR community. We will also release all the data and codes to facilitate the future research in this direction.

Neural ranking models (NRMs) have shown remarkable success in recent years, especially with pre-trained language models. However, deep neural models are notorious for their vulnerability to adversarial examples. Adversarial attacks may become a new type of web spamming technique given our increased reliance on neural information retrieval models. Therefore, it is important to study potential adversarial attacks to identify vulnerabilities of NRMs before they are deployed. In this article, we introduce the Word Substitution Ranking Attack (WSRA) task against NRMs, which aims at promoting a target document in rankings by adding adversarial perturbations to its text. We focus on the decision-based black-box attack setting, where the attackers cannot directly get access to the model information, but can only query the target model to obtain the rank positions of the partial retrieved list. This attack setting is realistic in real-world search engines. We propose a novel Pseudo Relevance-based ADversarial ranking Attack method (PRADA) that learns a surrogate model based on Pseudo Relevance Feedback (PRF) to generate gradients for finding the adversarial perturbations. Experiments on two web search benchmark datasets show that PRADA can outperform existing attack strategies and successfully fool the NRM with small indiscernible perturbations of text.

Neural ranking models (NRMs) have become one of the most important techniques in information retrieval (IR). Due to the limitation of relevance labels, the training of NRMs heavily relies on negative sampling over unlabeled data. In general machine learning scenarios, it has shown that training with hard negatives (i.e., samples that are close to positives) could lead to better performance. Surprisingly, we find opposite results from our empirical studies in IR. When sampling top-ranked results (excluding the labeled positives) as negatives from a stronger retriever, the performance of the learned NRM becomes even worse. Based on our investigation, the superficial reason is that there are more false negatives (i.e., unlabeled positives) in the top-ranked results with a stronger retriever, which may hurt the training process; The root is the existence of pooling bias in the dataset constructing process, where annotators only judge and label very few samples selected by some basic retrievers. Therefore, in principle, we can formulate the false negative issue in training NRMs as learning from labeled datasets with pooling bias. To solve this problem, we propose a novel Coupled Estimation Technique (CET) that learns both a relevance model and a selection model simultaneously to correct the pooling bias for training NRMs. Empirical results on three retrieval benchmarks show that NRMs trained with our technique can achieve significant gains on ranking effectiveness against other baseline strategies.

Knowledge-intensive language tasks (KILT) usually require a large body of information to provide correct answers. A popular paradigm to solve this problem is to combine a search system with a machine reader, where the former retrieves supporting evidences and the latter examines them to produce answers. Recently, the reader component has witnessed significant advances with the help of large-scale pre-trained generative models. Meanwhile most existing solutions in the search component rely on the traditional "index-retrieve-then-rank'' pipeline, which suffers from large memory footprint and difficulty in end-to-end optimization. Inspired by recent efforts in constructing model-based IR models, we propose to replace the traditional multi-step search pipeline with a novel single-step generative model, which can dramatically simplify the search process and be optimized in an end-to-end manner. We show that a strong generative retrieval model can be learned with a set of adequately designed pre-training tasks, and be adopted to improve a variety of downstream KILT tasks with further fine-tuning. We name the pre-trained generative retrieval model as CorpusBrain as all information about the corpus is encoded in its parameters without the need of constructing additional index. Empirical results show that CorpusBrain can significantly outperform strong baselines for the retrieval task on the KILT benchmark and establish new state-of-the-art downstream performances. We also show that CorpusBrain works well under zero- and low-resource settings.

Dense retrieval (DR) has shown promising results in many information retrieval (IR) related tasks, whose foundation is high-quality text representations for effective search. Taking the pre-trained language models (PLMs) as the text encoders has become a popular choice in DR. However, the learned representations based on these PLMs often lose the discriminative power, and thus hurt the recall performance, particularly as PLMs consider too much content of the input texts. Therefore, in this work, we propose to pre-train a discriminative language representation model, called DiscBERT, for DR. The key idea is that a good text representation should be able to automatically keep those discriminative features that could well distinguish different texts from each other in the semantic space. Specifically, inspired by knowledge distillation, we employ a simple yet effective training method, called self-teaching, to distill the model's knowledge constructed when training on the sampled representative tokens of a text sequence into the model's knowledge for the entire text sequence. By further fine-tuning on publicly available retrieval benchmark datasets, DiscBERT can outperform the state-of-the-art retrieval methods.

Pre-training and fine-tuning have achieved significant advances in the information retrieval (IR). A typical approach is to fine-tune all the parameters of large-scale pre-trained models (PTMs) on downstream tasks. As the model size and the number of tasks increase greatly, such approach becomes less feasible and prohibitively expensive. Recently, a variety of parameter-efficient tuning methods have been proposed in natural language processing (NLP) that only fine-tune a small number of parameters while still attaining strong performance. Yet there has been little effort to explore parameter-efficient tuning for IR.

Dense retrieval (DR) has shown promising results in information retrieval. In essence, DR requires high-quality text representations to support effective search in the representation space. Recent studies have shown that pre-trained autoencoder-based language models with a weak decoder can provide high-quality text representations, boosting the effectiveness and few-shot ability of DR models. However, even a weak autoregressive decoder has the bypass effect on the encoder. More importantly, the discriminative ability of learned representations may be limited since each token is treated equally important in decoding the input texts. To address the above problems, in this paper, we propose a contrastive pre-training approach to learn a discriminative autoencoder with a lightweight multi-layer perception (MLP) decoder. The basic idea is to generate word distributions of input text in a non-autoregressive fashion and pull the word distributions of two masked versions of one text close while pushing away from others. We theoretically show that our contrastive strategy can suppress the common words and highlight the representative words in decoding, leading to discriminative representations. Empirical results show that our method can significantly outperform the state-of-the-art autoencoder-based language models and other pre-trained models for dense retrieval.

Neural ranking models (NRMs) have achieved promising results in information retrieval. NRMs have also been shown to be vulnerable to adversarial examples. A typical Word Substitution Ranking Attack (WSRA) against NRMs was proposed recently, in which an attacker promotes a target document in rankings by adding human-imperceptible perturbations to its text. This raises concerns when deploying NRMs in real-world applications. Therefore, it is important to develop techniques that defend against such attacks for NRMs. In empirical defenses adversarial examples are found during training and used to augment the training set. However, such methods offer no theoretical guarantee on the models' robustness and may eventually be broken by other sophisticated WSRAs. To escape this arms race, rigorous and provable certified defense methods for NRMs are needed.

Text matching is a fundamental technique in both information retrieval and natural language processing. Text matching tasks share the same paradigm that determines the relationship between two given texts. The relationships vary from task to task, e.g. relevance in document retrieval, semantic alignment in paraphrase identification and answerable judgment in question answering. However, the essential signals for text matching remain in a finite scope, i.e. exact matching, semantic matching, and inference matching. Ideally, a good text matching model can learn to capture and aggregate these signals for different matching tasks to achieve competitive performance, while recent state-of-the-art text matching models, e.g. Pre-trained Language Models (PLMs), are hard to generalize. It is because the end-to-end supervised learning on task-specific dataset makes model overemphasize the data sample bias and task-specific signals instead of the essential matching signals, which ruins the generalization of model to different tasks. To overcome this problem, we adopt a specialization-generalization training strategy and refer to it as Match-Prompt. In specialization stage, descriptions of different matching tasks are mapped to only a few prompt tokens. In generalization stage, text matching model explores the essential matching signals by being trained on diverse multiple matching tasks. High diverse matching tasks avoid model fitting the data sample bias on a specific task, so that model can focus on learning the essential matching signals. Meanwhile, the prompt tokens obtained in the first step are added to the corresponding tasks to help the model distinguish different task-specific matching signals, as well as to form the basis prompt tokens for a new matching task. In this paper, we consider five common text matching tasks including document retrieval, open-domain question answering, retrieval-based dialogue, paraphrase identification, and natural language inference. Experimental results on eighteen public datasets show that Match-Prompt can improve multi-task generalization capability of PLMs in text matching and yield better in-domain multi-task, out-of-domain multi-task and new task adaptation performance than multi-task and task-specific models trained by previous fine-tuning paradigm.

Company financial risk is ubiquitous and early risk assessment for listed companies can avoid considerable losses. Traditional methods mainly focus on the financial statements of companies and lack the complex relationships among them. However, the financial statements are often biased and lagged, making it difficult to identify risks accurately and timely. To address the challenges, we redefine the problem as company financial risk assessment on tribe-style graph by taking each listed company and its shareholders as a tribe and leveraging financial news to build inter-tribe connections. Such tribe-style graphs present different patterns to distinguish risky companies from normal ones. However, most nodes in the tribe-style graph lack attributes, making it difficult to directly adopt existing graph learning methods (e.g., Graph Neural Networks(GNNs)). In this paper, we propose a novel Hierarchical Graph Neural Network (TH-GNN) for Tribe-style graphs via two levels, with the first level to encode the structure pattern of the tribes with contrastive learning, and the second level to diffuse information based on the inter-tribe relations, achieving effective and efficient risk assessment. Extensive experiments on the real-world company dataset show that our method achieves significant improvements on financial risk assessment over previous competing methods. Also, the extensive ablation studies and visualization comprehensively show the effectiveness of our method.

Fact verification (FV) is a challenging task which aims to verify a claim using multiple evidential sentences from trustworthy corpora, e.g., Wikipedia. Most existing approaches follow a three-step pipeline framework, including document retrieval, sentence retrieval and claim verification. High-quality evidences provided by the first two steps are the foundation of the effective reasoning in the last step. Despite being important, high-quality evidences are rarely studied by existing works for FV, which often adopt the off-the-shelf models to retrieve relevant documents and sentences in an "index-retrieve-then-rank'' fashion. This classical approach has clear drawbacks as follows: i) a large document index as well as a complicated search process is required, leading to considerable memory and computational overhead; ii) independent scoring paradigms fail to capture the interactions among documents and sentences in ranking; iii) a fixed number of sentences are selected to form the final evidence set. In this work, we proposeGERE, the first system that retrieves evidences in a generative fashion, i.e., generating the document titles as well as evidence sentence identifiers. This enables us to mitigate the aforementioned technical issues since: i) the memory and computational cost is greatly reduced because the document index is eliminated and the heavy ranking process is replaced by a light generative process; ii) the dependency between documents and that between sentences could be captured via sequential generation process; iii) the generative formulation allows us to dynamically select a precise set of relevant evidences for each claim. The experimental results on the FEVER dataset show that GERE achieves significant improvements over the state-of-the-art baselines, with both time-efficiency and memory-efficiency.

Stance detection aims to identify whether the author of a text is in favor of, against, or neutral to a given target. The main challenge of this task comes two-fold: few-shot learning resulting from the varying targets and the lack of contextual information of the targets. Existing works mainly focus on solving the second issue by designing attention-based models or introducing noisy external knowledge, while the first issue remains under-explored. In this paper, inspired by the potential capability of pre-trained language models (PLMs) serving as knowledge bases and few-shot learners, we propose to introduce prompt-based fine-tuning for stance detection. PLMs can provide essential contextual information for the targets and enable few-shot learning via prompts. Considering the crucial role of the target in stance detection task, we design target-aware prompts and propose a novel verbalizer. Instead of mapping each label to a concrete word, our verbalizer maps each label to a vector and picks the label that best captures the correlation between the stance and the target. Moreover, to alleviate the possible defect of dealing with varying targets with a single hand-crafted prompt, we propose to distill the information learned from multiple prompts. Experimental results show the superior performance of our proposed model in both full-data and few-shot scenarios.

Dense retrieval has shown promising results in many information retrieval (IR) related tasks, whose foundation is high-quality text representation learning for effective search. Some recent studies have shown that autoencoder-based language models are able to boost the dense retrieval performance using a weak decoder. However, we argue that 1) it is not discriminative to decode all the input texts and, 2) even a weak decoder has the bypass effect on the encoder. Therefore, in this work, we introduce a novel contrastive span prediction task to pre-train the encoder alone, but still retain the bottleneck ability of the autoencoder. In this way, we can 1) learn discriminative text representations efficiently with the group-wise contrastive learning over spans and, 2) avoid the bypass effect of the decoder thoroughly. Comprehensive experiments over publicly available retrieval benchmark datasets show that our approach can outperform existing pre-training methods for dense retrieval significantly.

Collaborative filtering is one of the most common scenarios and popular research topics in recommender systems. Among existing methods, latent factor models, i.e., learning a specific embedding for each user/item by reconstructing the observed interaction matrix, have shown excellent performances. However, such user-specific and item-specific embeddings are intrinsically transductive, making it difficult for them to deal with new users and new items unseen during training. Besides, the number of model parameters heavily depends on the number of all users and items, restricting their scalability to real-world applications. To solve the above challenges, in this paper, we propose a novel model-agnostic and scalable Inductive Embedding Module for collaborative filtering, namely INMO. INMO generates the inductive embeddings for users (items) by characterizing their interactions with some template items (template users), instead of employing an embedding lookup table. Under the theoretical analysis, we further propose an effective indicator for the selection of template users and template items. Our proposed INMO can be attached to existing latent factor models as a pre-module, inheriting the expressiveness of backbone models, while bringing the inductive ability and reducing model parameters. We validate the generality of INMO by attaching it to Matrix Factorization (MF) and LightGCN, which are two representative latent factor models for collaborative filtering. Extensive experiments on three public benchmarks demonstrate the effectiveness and efficiency of INMO in both transductive and inductive recommendation scenarios.

Pseudo-relevance feedback (PRF) has proven to be an effective query reformulation technique to improve retrieval accuracy. It aims to alleviate the mismatch of linguistic expressions between a query and its potential relevant documents. Existing PRF methods independently treat revised queries originating from the same query but using different numbers of feedback documents, resulting in severe query drift. Without comparing the effects of two different revisions from the same query, a PRF model may incorrectly focus on the additional irrelevant information increased in the more feedback, and thus reformulate a query that is less effective than the revision using the less feedback. Ideally, if a PRF model can distinguish between irrelevant and relevant information in the feedback, the more feedback documents there are, the better the revised query will be. To bridge this gap, we propose the Loss-over-Loss (LoL) framework to compare the reformulation losses between different revisions of the same query during training. Concretely, we revise an original query multiple times in parallel using different amounts of feedback and compute their reformulation losses. Then, we introduce an additional regularization loss on these reformulation losses to penalize revisions that use more feedback but gain larger losses. With such comparative regularization, the PRF model is expected to learn to suppress the extra increased irrelevant information by comparing the effects of different revised queries. Further, we present a differentiable query reformulation method to implement this framework. This method revises queries in the vector space and directly optimizes the retrieval performance of query vectors, applicable for both sparse and dense retrieval models. Empirical evaluation demonstrates the effectiveness and robustness of our method for two typical sparse and dense retrieval models.

Unsupervised opinion summarization techniques are designed to condense the review data and summarize informative and salient opinions in the absence of golden references. Existing dominant methods generally follow a two-stage framework: first creating the synthetic "review-summary" paired datasets and then feeding them into the generative summary model for supervised training. However, these methods mainly focus on semantic similarity in synthetic dataset creation, ignoring the consistency of aspects and sentiments in synthetic pairs. Such inconsistency also brings a gap to the training and inference of the summarization model.

Given a stream of money transactions between accounts in a bank, how can we accurately detect money laundering agent accounts and suspected behaviors in real-time? Money laundering agents try to hide the origin of illegally obtained money by dispersive multiple small transactions and evade detection by smart strategies. Therefore, it is challenging to accurately catch such fraudsters in an unsupervised manner. Existing approaches do not consider the characteristics of those agent accounts and are not suitable to the streaming settings. Therefore, we propose MonLAD and MonLAD-W to detect money laundering agent accounts in a transaction stream by keeping track of their residuals and other features; we devise AnoScore algorithm to find anomalies based on the robust measure of statistical deviation. Experimental results show that MonLAD outperforms the state-of-the-art baselines on real-world data and finds various suspicious behavior patterns of money laundering. Additionally, several detected suspected accounts have been manually-verified as agents in real money laundering scenario.

Multi-stage ranking pipelines have been a practical solution in modern search systems, where the first-stage retrieval is to return a subset of candidate documents and latter stages attempt to re-rank those candidates. Unlike re-ranking stages going through quick technique shifts over the past decades, the first-stage retrieval has long been dominated by classical term-based models. Unfortunately, these models suffer from the vocabulary mismatch problem, which may block re-ranking stages from relevant documents at the very beginning. Therefore, it has been a long-term desire to build semantic models for the first-stage retrieval that can achieve high recall efficiently. Recently, we have witnessed an explosive growth of research interests on the first-stage semantic retrieval models. We believe it is the right time to survey current status, learn from existing methods, and gain some insights for future development. In this article, we describe the current landscape of the first-stage retrieval models under a unified framework to clarify the connection between classical term-based retrieval methods, early semantic retrieval methods, and neural semantic retrieval methods. Moreover, we identify some open challenges and envision some future directions, with the hope of inspiring more research on these important yet less investigated topics.

Question generation is an important yet challenging problem in Artificial Intelligence (AI), which aims to generate natural and relevant questions from various input formats, e.g., natural language text, structure database, knowledge base, and image. In this article, we focus on question generation from natural language text, which has received tremendous interest in recent years due to the widespread applications such as data augmentation for question answering systems. During the past decades, many different question generation models have been proposed, from traditional rule-based methods to advanced neural network-based methods. Since there have been a large variety of research works proposed, we believe it is the right time to summarize the current status, learn from existing methodologies, and gain some insights for future development. In contrast to existing reviews, in this survey, we try to provide a more comprehensive taxonomy of question generation tasks from three different perspectives, i.e., the types of the input context text, the target answer, and the generated question. We take a deep look into existing models from different dimensions to analyze their underlying ideas, major design principles, and training strategies We compare these models through benchmark tasks to obtain an empirical understanding of the existing techniques. Moreover, we discuss what is missing in the current literature and what are the promising and desired future directions.

User identity linkage aims to link users with the same identities across different social networks. Recently, re- searchers model the similarities of users’ behaviors such as Point of Interests(PoIs) or User Generated Contents(UGCs) to predict the identities of users. However, it is non-trivial to solve the problem due to the following challenges: 1) PoIs are always sparse in the non-location-based social platforms, and it is impractical to measure the similarities of users solely with PoIs; 2) The similarities of hierarchical are hierarchical from the view of word, phrase, and sentence. How to model the hierarchical structure remains a key challenge; 3) The unreliable semantics of words. Two different words may refer to the same physical appearance of users, indicating that users are with the same identities.To tackle the above problems, we propose UGCLink, a knowledge distillation framework that models UGCs to predict user identities. Two main components are included in the framework, where the student network aims to model the similarities of UGCs and the teacher network guides the student network to learn better word embeddings that reveal the physical appearance of users. Besides, the teacher network, a document classification model that classifies UGCs into the categories of PoIs, is trained to guide the word embedding learning process in the student network to circumvent the unreliable semantic problem. We demonstrate that our proposed method outperforms the state- of-the-art methods by more than 11% in terms of AUC score.

Question Answering (QA), a popular and promising technique for intelligent information access, faces a dilemma about data as most other AI techniques. On one hand, modern QA methods rely on deep learning models which are typically data-hungry. Therefore, it is expected to collect and fuse all the available QA datasets together in a common site for developing a powerful QA model. On the other hand, real-world QA datasets are typically distributed in the form of isolated islands belonging to different parties. Due to the increasing awareness of privacy security, it is almost impossible to integrate the data scattered around, or the cost is prohibited. A possible solution to this dilemma is a new approach known as federated learning, which is a privacy-preserving machine learning technique over distributed datasets. In this work, we propose to adopt federated learning for QA with the special concern on the statistical heterogeneity of the QA data. Here the heterogeneity refers to the fact that annotated QA data are typically with non-identical and independent distribution (non-IID) and unbalanced sizes in practice. Traditional federated learning methods may sacrifice the accuracy of individual models under the heterogeneous situation. To tackle this problem, we propose a novel Federated Matching framework for QA, named FedMatch, with a backbone-patch architecture. The shared backbone is to distill the common knowledge of all the participants while the private patch is a compact and efficient module to retain the domain information for each participant. To facilitate the evaluation, we build a benchmark collection based on several QA datasets from different domains to simulate the heterogeneous situation in practice. Empirical studies demonstrate that our model can achieve significant improvements against the baselines over all the datasets.

Signed networks are such social networks having both positive and negative links. A lot of theories and algorithms have been developed to model such networks (e.g., balance theory). However, previous work mainly focuses on the unipartite signed networks where the nodes have the same type. Signed bipartite networks are different from classical signed networks, which contain two different node sets and signed links between two node sets. Signed bipartite networks can be commonly found in many fields including business, politics, and academics, but have been less studied. In this work, we firstly define the signed relationship of the same set of nodes and provide a new perspective for analyzing signed bipartite networks. Then we do some comprehensive analysis of balance theory from two perspectives on several real-world datasets. Specifically, in the peer review dataset, we find that the ratio of balanced isomorphism in signed bipartite networks increased after rebuttal phases. Guided by these two perspectives, we propose a novel Signed Bipartite Graph Neural Networks (SBGNNs) to learn node embeddings for signed bipartite networks. SBGNNs follow most GNNs message-passing scheme, but we design new message functions, aggregation functions, and update functions for signed bipartite networks. We validate the effectiveness of our model on four real-world datasets on Link Sign Prediction task, which is the main machine learning task for signed networks. Experimental results show that our SBGNN model achieves significant improvement compared with strong baseline methods, including feature-based methods and network embedding methods.

Neural text matching models have been widely used in community question answering, information retrieval, and dialogue. However, these models designed for short texts cannot well address the long-form text matching problem, because there are many contexts in long-form texts can not be directly aligned with each other, and it is difficult for existing models to capture the key matching signals from such noisy data. Besides, these models are computationally expensive for simply use all textual data indiscriminately. To tackle the effectiveness and efficiency problem, we propose a novel hierarchical noise filtering model, namely Match-Ignition. The main idea is to plug the well-known PageRank algorithm into the Transformer, to identify and filter both sentence and word level noisy information in the matching process. Noisy sentences are usually easy to detect because previous work has shown that their similarity can be explicitly evaluated by the word overlapping, so we directly use PageRank to filter such information based on a sentence similarity graph. Unlike sentences, words rely on their contexts to express concrete meanings, so we propose to jointly learn the filtering and matching process, to well capture the critical word-level matching signals. Specifically, a word graph is first built based on the attention scores in each self-attention block of Transformer, and key words are then selected by applying PageRank on this graph. In this way, noisy words will be filtered out layer by layer in the matching process. Experimental results show that Match-Ignition outperforms both SOTA short text matching models and recent long-form text matching models. We also conduct detailed analysis to show that Match-Ignition efficiently captures important sentences and words, to facilitate the long-form text matching process.

Nowadays, users are encouraged to activate across multiple online social networks simultaneously. User identity linkage, which aims to reveal the correspondence among different accounts across networks, has been regarded as a fundamental problem for user profiling, marketing, cybersecurity, and recommendation. Existing methods mainly address the prediction problem by utilizing profile, content, or structural features of users in symmetric ways. However, encouraged by online services, information from different social platforms may also be asymmetric, such as geo-locations and texts. It leads to an emerged challenge in aligning users with asymmetric information across networks. Instead of similarity evaluation applied in previous works, we formalize correlation between geo-locations and texts and propose a novel user identity linkage framework for matching users across networks. Moreover, our model can alleviate the label scarcity problem by introducing external text-location pairs. Experimental results on real-world datasets show that our approach outperforms existing methods and achieves state-of-the-art results.

Node injection attack on Graph Neural Networks (GNNs) is an emerging and practical attack scenario that the attacker injects malicious nodes rather than modifying original nodes or edges to affect the performance of GNNs. However, existing node injection attacks ignore extremely limited scenarios, namely the injected nodes might be excessive such that they may be perceptible to the target GNN. In this paper, we focus on an extremely limited scenario of single node injection evasion attack, i.e., the attacker is only allowed to inject one single node during the test phase to hurt GNN's performance. The discreteness of network structure and the coupling effect between network structure and node features bring great challenges to this extremely limited scenario. We first propose an optimization-based method to explore the performance upper bound of single node injection evasion attack. Experimental results show that 100%, 98.60%, and 94.98% nodes on three public datasets are successfully attacked even when only injecting one node with one edge, confirming the feasibility of single node injection evasion attack. However, such an optimization-based method needs to be re-optimized for each attack, which is computationally unbearable. To solve the dilemma, we further propose a Generalizable Node Injection Attack model, namely G-NIA, to improve the attack efficiency while ensuring the attack performance. Experiments are conducted across three well-known GNNs. Our proposed G-NIA significantly outperforms state-of-the-art baselines and is 500 times faster than the optimization-based method when inferring.

Given a large graph, how can we summarize it with fewer nodes and edges while maintaining its key properties, e.g.node degrees and graph spectrum?As a solution, graph summarization, which aims to find the compact representation for optimally describing and reconstructing a given graph, has received much attention, and numerous methods have been developed for it.However, many existing methods adopt the uniform reconstruction scheme, which is an unrealistic assumption as most real-world graphs have highly skewed node degrees, even within communities.Therefore we propose a degree-preserving graph summarization model, DPGS, with a novel reconstruction scheme based on the configuration model.To optimize the Minimum Description Length of our model, we deisgn a linearly scalable algorithm using hashing techniques.We theoretically show that the minimized reconstruction error bounds the perturbation of graph spectral information.Extensive experiments on realworld datasets show that DPGS yields more accurate summary graphs than several well-known baselines.Moreover, our reduced summary graphs can effectively train graph neural networks (GNNs) while saving computational cost.

Knowledge Graph (KG) reasoning that predicts missing facts for incomplete KGs has been widely explored. However, reasoning over Temporal KG (TKG) that predicts facts in the future is still far from resolved. The key to predict future facts is to thoroughly understand the historical facts. A TKG is actually a sequence of KGs corresponding to different timestamps, where all concurrent facts in each KG exhibit structural dependencies and temporally adjacent facts carry informative sequential patterns. To capture these properties effectively and efficiently, we propose a novel Recurrent Evolution network based on Graph Convolution Network (GCN), called RE-GCN, which learns the evolutional representations of entities and relations at each timestamp by modeling the KG sequence recurrently. Specifically, for the evolution unit, a relation-aware GCN is leveraged to capture the structural dependencies within the KG at each timestamp. In order to capture the sequential patterns of all facts in parallel, the historical KG sequence is modeled auto-regressively by the gate recurrent components. Moreover, the static properties of entities, such as entity types, are also incorporated via a static graph constraint component to obtain better entity representations. Fact prediction at future timestamps can then be realized based on the evolutional entity and relation representations. Extensive experiments demonstrate that the RE-GCN model obtains substantial performance and efficiency improvement for the temporal reasoning tasks on six benchmark datasets. Especially, it achieves up to 11.46% improvement in MRR for entity prediction with up to 82 times speedup compared to the state-of-the-art baseline.

Pre-training and fine-tuning have achieved remarkable success in many downstream natural language processing (NLP) tasks. Recently, pre-training methods tailored for information retrieval (IR) have also been explored, and the latest success is the PROP method which has reached new SOTA on a variety of ad-hoc retrieval benchmarks. The basic idea of PROP is to construct therepresentative words prediction (ROP) task for pre-training inspired by the query likelihood model. Despite its exciting performance, the effectiveness of PROP might be bounded by the classical unigram language model adopted in the ROP task construction process. To tackle this problem, we propose a bootstrapped pre-training method (namely B-PROP) based on BERT for ad-hoc retrieval. The key idea is to use the powerful contextual language model BERT to replace the classical unigram language model for the ROP task construction, and re-train BERT itself towards the tailored objective for IR. Specifically, we introduce a novel contrastive method, inspired by the divergence-from-randomness idea, to leverage BERT's self-attention mechanism to sample representative words from the document. By further fine-tuning on downstream ad-hoc retrieval tasks, our method achieves significant improvements over PROP and other baselines, and further pushes forward the SOTA on a variety of ad-hoc retrieval tasks.

Recently pre-trained language representation models such as BERT have shown great success when fine-tuned on downstream tasks including information retrieval (IR). However, pre-training objectives tailored for ad-hoc retrieval have not been well explored. In this paper, we propose Pre-training with Representative wOrds Prediction (PROP) for ad-hoc retrieval. PROP is inspired by the classical statistical language model for IR, specifically the query likelihood model, which assumes that the query is generated as the piece of text representative of the "ideal" document. Based on this idea, we construct the representative words prediction (ROP) task for pre-training. Given an input document, we sample a pair of word sets according to the document language model, where the set with higher likelihood is deemed as more representative of the document. We then pre-train the Transformer model to predict the pairwise preference between the two word sets, jointly with the Masked Language Model (MLM) objective. By further fine-tuning on a variety of representative downstream ad-hoc retrieval tasks, PROP achieves significant improvements over baselines without pre-training or with other pre-training methods. We also show that PROP can achieve exciting performance under both the zero- and low-resource IR settings.

Community Question Answering (CQA) sites such as Yahoo! Answers and Baidu Knows have emerged as rich knowledge resources for information seekers. However, answers posted to CQA sites often vary a lot in their qualities. User votes from the community may partially reflect the overall quality of the answer, but they are often missing. Hence, automatic selection of "good'' answers becomes a practical research problem that will help us manage the quality of accumulated knowledge. Without loss of generality, a good answer should deliver not only relevant but also trustworthy information that can help resolve the information needs of the posted question, but the latter has received less investigation in the past. In this paper, we propose a novel matching-verification framework for automatic answer selection. The matching component assesses the relevance of a candidate answer to a given question as conventional QA methods. The major enhancement is the verification component, which aims to leverage the wisdom of crowds, e.g., some big information repository, for trustworthiness measurement. Given a question, we take the top retrieved results from the information repository as the supporting evidences to distill the consensus representation. A major challenge is that there is no guarantee that one can always obtain reliable consensus from the wisdom of crowds for a question due to the noisy nature and the limitation of the existing search technology.Therefore, we decompose the trustworthiness measurement into two parts, i.e., a verification score which measures the consistency between a candidate answer and the consensus representation, and a confidence score which measures the reliability of the consensus itself. Empirical studies on three real-world CQA data collections, i.e. YahooQA, QuoraQA and AmazonQA, show that our approach can significantly outperform the state-of-the-art methods on the answer selection task.

Despite achieving strong performance in semi-supervised node classification task, graph neural networks (GNNs) are vulnerable to adversarial attacks, similar to other deep learning models. Existing researches focus on developing either robust GNN models or attack detection methods against adversarial attacks on graphs. However, little research attention is paid to the potential and practice of immunization to adversarial attacks on graphs. In this paper, we propose and formulate the graph adversarial immunization problem, i.e., vaccinating an affordable fraction of node pairs, connected or unconnected, to improve the certifiable robustness of graph against any admissible adversarial attack. We further propose an effective algorithm, called AdvImmune, which optimizes with meta-gradient in a discrete way to circumvent the computationally expensive combinatorial optimization when solving the adversarial immunization problem. Experiments are conducted on two citation networks and one social network. Experimental results demonstrate that the proposed AdvImmune method remarkably improves the ratio of robust nodes by 12%, 42%, 65%, with an affordable immune budget of only 5% edges.

Relevance plays a central role in information retrieval (IR), which has received extensive studies starting from the 20th century. The definition and the modeling of relevance has always been critical challenges in both information science and computer science research areas. Along with the debate and exploration on relevance, IR has already become a core task in many real-world applications, such as Web search engines, question answering systems, conversational bots, and so on. While relevance acts as a unified concept in all these retrieval tasks, the inherent definitions are quite different due to the heterogeneity of these tasks. This raises a question to us: Do these different forms of relevance really lead to different modeling focuses? To answer this question, in this work, we conduct an empirical study on relevance modeling in three representative IR tasks, i.e., document retrieval, answer retrieval, and response retrieval. Specifically, we attempt to study the following two questions: 1) Does relevance modeling in these tasks really show differences in terms of natural language understanding (NLU)? We employ 16 linguistic tasks to probe a unified retrieval model over these three retrieval tasks to answer this question. 2) If there do exist differences, how can we leverage the findings to enhance the relevance modeling? We proposed three intervention methods to investigate how to leverage different modeling focuses of relevance to improve these IR tasks. We believe the way we study the problem as well as our findings would be beneficial to the IR community.

Information gathered from multiple sources on the Web often exhibits conflicts. This phenomenon motivates the need of truth discovery, which aims to automatically find the true claim among multiple conflicting claims. Existing truth discovery methods are mainly based on iterative updates, optimization or probabilistic models. Although these methods have shown their own effectiveness, they have a common limitation. These methods do not model relationships between each pair of source and target such that they do not well capture the underlying interactions in the data. In this paper, we propose a new model for truth discovery, learning the representations of sources and claims automatically from the interactions between sources and targets. Our model first constructs a heterogenous network including source-claim, source-source and truth-claim relationships. It then embeds the network into a low dimensional space such that trustworthy sources and true claims are close. In this way, truth discovery can be conveniently performed in the embedding space. Moreover, our model can be implemented in both semi-supervised and un-supervised manners to deal with the label scarcity problem in practical truth discovery. Experiments on three real-world datasets demonstrate that our model outperforms existing state-of-the-art methods for truth discovery.

How to effectively utilize the dialogue history is a crucial problem in multi-turn dialogue generation. Previous works usually employ various neural network architectures (e.g., recurrent neural networks, attention mechanisms, and hierarchical structures) to model the history. However, a recent empirical study by Sankar et al. has shown that these architectures lack the ability of understanding and modeling the dynamics of the dialogue history. For example, the widely used architectures are insensitive to perturbations of the dialogue history, such as words shuffling, utterances missing, and utterances reordering. To tackle this problem, we propose a Ranking Enhanced Dialogue generation framework in this paper. Despite the traditional representation encoder and response generation modules, an additional ranking module is introduced to model the ranking relation between the former utterance and consecutive utterances. Specifically, the former utterance and consecutive utterances are treated as query and corresponding documents, and both local and global ranking losses are designed in the learning process. In this way, the dynamics in the dialogue history can be explicitly captured. To evaluate our proposed models, we conduct extensive experiments on three public datasets, i.e., bAbI, PersonaChat, and JDC. Experimental results show that our models produce better responses in terms of both quantitative measures and human judgments, as compared with the state-of-the-art dialogue generation models. Furthermore, we give some detailed experimental analysis to show where and how the improvements come from.

Knowledge Graph Question Answering aims to automatically answer natural language questions via well-structured relation information between entities stored in knowledge graphs. When faced with a complex question with compositional semantics, query graph generation is a practical semantic parsing-based method. But existing works rely on heuristic rules with limited coverage, making them impractical on more complex questions. This paper proposes a Director-Actor-Critic framework to overcome these challenges. Through options over a Markov Decision Process, query graph generation is formulated as a hierarchical decision problem. The Director determines which types of triples the query graph needs, the Actor generates corresponding triples by choosing nodes and edges, and the Critic calculates the semantic similarity between the generated triples and the given questions. Moreover, to train from weak supervision, we base the framework on hierarchical Reinforcement Learning with intrinsic motivation. To accelerate the training process, we pre-train the Critic with high-reward trajectories generated by hand-crafted rules, and leverage curriculum learning to gradually increase the complexity of questions during query graph generation. Extensive experiments conducted over widely-used benchmark datasets demonstrate the effectiveness of the proposed framework.

Machine reading comprehension (MRC) has become a core component in a variety of natural language processing (NLP) applications such as question answering and dialogue systems. It becomes a practical challenge that an MRC model needs to learn in non-stationary environments, in which the underlying data distribution changes over time. A typical scenario is the domain drift, i.e. different domains of data come one after another, where the MRC model is required to adapt to the new domain while maintaining previously learned ability. To tackle such a challenge, in this work, we introduce the Continual Domain Adaptation (CDA) task for MRC. So far as we know, this is the first study on the continual learning perspective of MRC. We build two benchmark datasets for the CDA task, by re-organizing existing MRC collections into different domains with respect to context type and question type, respectively. We then analyze and observe the catastrophic forgetting (CF) phenomenon of MRC under the CDA setting. To tackle the CDA task, we propose several BERT-based continual learning MRC models using either regularization-based methodology or dynamic-architecture paradigm. We analyze the performance of different continual learning MRC models under the CDA task and show that the proposed dynamic-architecture based model achieves the best performance.

Query understanding is a fundamental problem in information retrieval (IR), which has attracted continuous attention through the past decades. Many different tasks have been proposed for understanding users' search queries, e.g., query classification or query clustering. However, it is not that precise to understand a search query at the intent class/cluster level due to the loss of many detailed information. As we may find in many benchmark datasets, e.g., TREC and SemEval, queries are often associated with a detailed description provided by human annotators which clearly describes its intent to help evaluate the relevance of the documents. If a system could automatically generate a detailed and precise intent description for a search query, like human annotators, that would indicate much better query understanding has been achieved. In this paper, therefore, we propose a novel Query-to-Intent-Description (Q2ID) task for query understanding. Unlike those existing ranking tasks which leverage the query and its description to compute the relevance of documents, Q2ID is a reverse task which aims to generate a natural language intent description based on both relevant and irrelevant documents of a given query. To address this new task, we propose a novel Contrastive Generation model, namely CtrsGen for short, to generate the intent description by contrasting the relevant documents with the irrelevant documents given a query. We demonstrate the effectiveness of our model by comparing with several state-of-the-art generation models on the Q2ID task. We discuss the potential usage of such Q2ID technique through an example application.

In learning-to-rank for information retrieval, a ranking model is automatically learned from the data and then utilized to rank the sets of retrieved documents. Therefore, an ideal ranking model would be a mapping from a document set to a permutation on the set, and should satisfy two critical requirements: (1) it should have the ability to model cross-document interactions so as to capture local context information in a query; (2) it should be permutation-invariant, which means that any permutation of the inputted documents would not change the output ranking. Previous studies on learning-to-rank either design uni-variate scoring functions that score each document separately, and thus failed to model the cross-document interactions; or construct multivariate scoring functions that score documents sequentially, which inevitably sacrifice the permutation invariance requirement. In this paper, we propose a neural learning-to-rank model called SetRank which directly learns a permutation-invariant ranking model defined on document sets of any size. SetRank employs a stack of (induced) multi-head self attention blocks as its key component for learning the embeddings for all of the retrieved documents jointly. The self-attention mechanism not only helps SetRank to capture the local context information from cross-document interactions, but also to learn permutation-equivariant representations for the inputted documents, which therefore achieving a permutation-invariant ranking model. Experimental results on three benchmarks showed that the SetRank significantly outperformed the baselines include the traditional learning-to-rank models and state-of-the-art Neural IR models.

Community Question Answering (CQA) has become a primary means for people to acquire knowledge, where people are free to ask questions or submit answers. To enhance the efficiency of the service, similar question identification becomes a core task in CQA which aims to find a similar question from the archived repository whenever a new question is asked. However, it has long been a challenge to properly measure the similarity between two questions due to the inherent variation of natural language, i.e., there could be different ways to ask a same question or different questions sharing similar expressions. To alleviate this problem, it is natural to involve the existing answers for the enrichment of the archived questions. Traditional methods typically take aone-side usage, which leverages the answer as some expanded representation of the corresponding question. Unfortunately, this may introduce unexpected noises into the similarity computation since answers are often long and diverse, leading to inferior performance. In this work, we propose atwo-side usage, which leverages the answer as a bridge of the two questions. The key idea is based on our observation that similar questions could be addressed by similar parts of the answer while different questions may not. In other words, we can compare the matching patterns of the two questions over the same answer to measure their similarity. In this way, we propose a novel matching over matching model, namely Match2, which compares the matching patterns between two question-answer pairs for similar question identification. Empirical experiments on two benchmark datasets demonstrate that our model can significantly outperform previous state-of-the-art methods on the similar question identification task.

Graph neural networks (GNNs) achieve remarkable success in graph-based semi-supervised node classification, leveraging the information from neighboring nodes to improve the representation learning of target node. The success of GNNs at node classification depends on the assumption that connected nodes tend to have the same label. However, such an assumption does not always work, limiting the performance of GNNs at node classification. In this paper, we propose label-consistency based graph neural network (LC-GNN), leveraging node pairs unconnected but with the same labels to enlarge the receptive field of nodes in GNNs. Experiments on benchmark datasets demonstrate the proposed LC-GNN outperforms traditional GNNs in graph-based semi-supervised node classification. We further show the superiority of LC-GNN in sparse scenarios with only a handful of labeled nodes.

This paper concerns reinforcement learning~(RL) of the document ranking models for information retrieval~(IR). One branch of the RL approaches to ranking formalize the process of ranking with Markov decision process~(MDP) and determine the model parameters with policy gradient. Though preliminary success has been shown, these approaches are still far from achieving their full potentials. Existing policy gradient methods directly utilize the absolute performance scores (returns) of the sampled document lists in its gradient estimations, which may cause two limitations: 1) fail to reflect the relative goodness of documents within the same query, which usually is close to the nature of IR ranking; 2) generate high variance gradient estimations, resulting in slow learning speed and low ranking accuracy. To deal with the issues, we propose a novel policy gradient algorithm in which the gradients are determined using pairwise comparisons of two document lists sampled within the same query. The algorithm, referred to as Pairwise Policy Gradient (PPG), repeatedly samples pairs of document lists, estimates the gradients with pairwise comparisons, and finally updates the model parameters. Theoretical analysis shows that PPG makes an unbiased and low variance gradient estimations. Experimental results have demonstrated performance gains over the state-of-the-art baselines in search result diversification and text retrieval.

The abductive natural language inference task (αNLI) is proposed to evaluate the abductive reasoning ability of a learning system. In the αNLI task, two observations are given and the most plausible hypothesis is asked to pick out from the candidates. Existing methods simply formulate it as a classification problem, thus a cross-entropy log-loss objective is used during training. However, discriminating true from false does not measure the plausibility of a hypothesis, for all the hypotheses have a chance to happen, only the probabilities are different. To fill this gap, we switch to a ranking perspective that sorts the hypotheses in order of their plausibilities. With this new perspective, a novel L2R2 approach is proposed under the learning-to-rank framework. Firstly, training samples are reorganized into a ranking form, where two observations and their hypotheses are treated as the query and a set of candidate documents respectively. Then, an ESIM model or pre-trained language model, e.g. BERT or RoBERTa, is obtained as the scoring function. Finally, the loss functions for the ranking task can be either pair-wise or list-wise for training. The experimental results on the ART dataset reach the state-of-the-art in the public leaderboard.

Predicting the popularity of online content on social platforms is an important task for both researchers and practitioners. Previous methods mainly leverage demographics, temporal and structural patterns of early adopters for popularity prediction. However, most existing methods are less effective to precisely capture the cascading effect in information diffusion, in which early adopters try to activate potential users along the underlying network. In this paper, we consider the problem of network-aware popularity prediction, leveraging both early adopters and social networks for popularity prediction. We propose to capture the cascading effect explicitly, modeling the activation state of a target user given the activation state and influence of his/her neighbors. To achieve this goal, we propose a novel method, namely CoupledGNN, which uses two coupled graph neural networks to capture the interplay between node activation states and the spread of influence. By stacking graph neural network layers, our proposed method naturally captures the cascading effect along the network in a successive manner. Experiments conducted on both synthetic and real-world Sina Weibo datasets demonstrate that our method significantly outperforms the state-of-the-art methods for popularity prediction.

Reading comprehension (RC) aims to locate a text span from a context passage to answer the given question. Despite the effectiveness of modern neural RC models, most existing work relies on maximum likelihood estimation (MLE) and ignores the structure of the output space. That is during training, one treats all the text spans do not match the ground truth as equally poor, leading to overconfident predictions on ground truth labels and reduced generalization ability in test. One way to bridge the gap between training and test is to take into account the task reward of alternative outputs using the reinforcement learning (RL) algorithms, which is often deficient in optimization as compared with MLE. In this paper, we propose a new learning criterion for the RC task which combines the merits of both MLE and RL-based methods. Specifically, we show that we are able to derive the distribution of the outputs, i.e., label distribution, using their corresponding task rewards based on the decomposition property of the RC problem. We then optimize the RC model by directly learning towards the auxiliary label distribution, instead of the ground truth label, using the MLE framework. In this way, we can make use of the structure of the output space for better generalization (as RL) via efficient optimization (as MLE). We name our approach as Label Distribution augmented MLE (LD-MLE), which is a general learning criterion that could be adopted by almost all the existing RC models. Experiments on three representative benchmark datasets demonstrate that RC models learned with the LD-MLE criterion can achieve consistently improved results over those based on the traditional MLE and RL-based criteria.

The conversation task is usually formulated as a conditional generation problem, i.e., to generate a natural and meaningful response given the input utterance. Generally speaking, this formulation is apparently based on an oversimplified assumption that the response is solely dependent on the input utterance. It ignores the subjective factor of the responder, e.g., his/her emotion or knowledge state, which is a major factor that affects the response in practice. Without explicitly differentiating such subjective factor behind the response, existing generation models can only learn the general shape of conversations, leading to the blandness problem of the response. Moreover, there is no intervention mechanism within the existing generation process, since the response is fully decided by the input utterance. In this work, we propose to view the conversation task as a dual-factor generation problem, including an objective factor denoting the input utterance and a subjective factor denoting the responder state. We extend the existing neural sequence-to-sequence (Seq2Seq) model to accommodate the responder state modeling. We introduce two types of responder state, i.e., discrete and continuous state, to model emotion state and topic preference state, respectively. We show that with our dual-factor generation model, we can not only better fit the conversation data, but also actively control the generation of the response with respect to sentiment or topic specificity.

Text matching is the core problem in many natural language processing (NLP) tasks, such as information retrieval, question answering, and conversation. Recently, deep leaning technology has been widely adopted for text matching, making neural text matching a new and active research domain. With a large number of neural matching models emerging rapidly, it becomes more and more difficult for researchers, especially those newcomers, to learn and understand these new models. Moreover, it is usually difficult to try these models due to the tedious data pre-processing, complicated parameter configuration, and massive optimization tricks, not to mention the unavailability of public codes sometimes. Finally, for researchers who want to develop new models, it is also not an easy task to implement a neural text matching model from scratch, and to compare with a bunch of existing models. In this paper, therefore, we present a novel system, namely MatchZoo, to facilitate the learning, practicing and designing of neural text matching models. The system consists of a powerful matching library and a user-friendly and interactive studio, which can help researchers: 1) to learn state-of-the-art neural text matching models systematically, 2) to train, test and apply these models with simple configurable steps; and 3) to develop their own models with rich APIs and assistance.

Web question answering (QA) has become an dispensable component in modern search systems, which can significantly improve users' search experience by providing a direct answer to users' information need. This could be achieved by applying machine reading comprehension (MRC) models over the retrieved passages to extract answers with respect to the search query. With the development of deep learning techniques, state-of-the-art MRC performances have been achieved by recent deep methods. However, existing studies on MRC seldom address the predictive uncertainty issue, i.e., how likely the prediction of an MRC model is wrong, leading to uncontrollable risks in real-world Web QA applications. In this work, we first conduct an in-depth investigation over the risk of Web QA. We then introduce a novel risk control framework, which consists of a qualify model for uncertainty estimation using the probe idea, and a decision model for selectively output. For evaluation, we introduce risk-related metrics, rather than the traditional EM and F1 in MRC, for the evaluation of risk-aware Web QA. The empirical results over both the real-world Web QA dataset and the academic MRC benchmark collection demonstrate the effectiveness of our approach.

In this paper, we introduce and tackle the Outline Generation (OG) task, which aims to unveil the inherent content structure of a multi-paragraph document by identifying its potential sections and generating the corresponding section headings. Without loss of generality, the OG task can be viewed as a novel structured summarization task. To generate a sound outline, an ideal OG model should be able to capture three levels of coherence, namely the coherence between context paragraphs, that between a section and its heading, and that between context headings. The first one is the foundation for section identification, while the latter two are critical for consistent heading generation. In this work, we formulate the OG task as a hierarchical structured prediction problem, i.e., to first predict a sequence of section boundaries and then a sequence of section headings accordingly. We propose a novel hierarchical structured neural generation model, named HiStGen, for the task. Our model attempts to capture the three-level coherence via the following ways. First, we introduce a Markov paragraph dependency mechanism between context paragraphs for section identification. Second, we employ a section-aware attention mechanism to ensure the semantic coherence between a section and its heading. Finally, we leverage a Markov heading dependency mechanism and a review mechanism between context headings to improve the consistency and eliminate duplication between section headings. Besides, we build a novel Wriptsize IKI OG dataset, a public collection which consists of over 1.75 million document-outline pairs for research on the OG task. Experimental results on our benchmark dataset demonstrate that our model can significantly outperform several state-of-the-art sequential generation models for the OG task.

Knowledge graphs have recently been extensively applied in many different areas (e.g., disaster management and relief, disease diagnosis). For example, event-centric knowledge graphs have been developed to improve decision making in disaster management and relief. This paper focuses on the task of event detection, which is the precondition of event extraction for constructing event-centric knowledge graphs. Event detection identifies trigger words of events in the sentences of a document and further classifies the types of events. It is straightforward that context information is useful for event detection. Therefore, the feature-based methods adopt cross-sentence information. However, they suffer from the complication of human-designed features. On the other hand, the representation-based methods learn document-level embeddings, which, however, contain much noise caused by unsupervised learning. To overcome these problems, in this paper we propose a new model based on Semi-supervised Auto-Encoder, which learns Context information to Enhance Event Detection, thus called SAE-CEED. This model first applies large-scale unlabeled texts to pre-train an auto-encoder, so that the embeddings of segments learned by the encoder contain the semantic and order information of the original text. It then uses the decoder to extract the context embeddings and fine-tunes them to enhance a bidirectional neural network model to identify event triggers and their types in sentences. Through experiments on the benchmark ACE-2005 dataset, we demonstrate the effectiveness of the proposed SAE-CEED model. In addition, we systematically conduct a series of experiments to verify the impact of different lengths of text segments in the pre-training of the auto-encoder on event detection.

With the overwhelming popularity of Knowledge Graphs (KGs), researchers have poured attention to link prediction to complete KGs for a long time. However, they mainly focus on promoting the performance on binary relational data, where facts are usually represented as triples in the form of (head entity, relation, tail entity). In practice, n-ary relational facts are also ubiquitous. When encountering such facts, existing studies usually decompose them into triples by introducing a multitude of auxiliary virtual entities and additional triples. These conversions result in the complexity of carrying out link prediction concerning more than two arities. It has even proven that they may cause loss of structural information. To overcome these problems, in this paper, without decomposition, we represent each n-ary relational fact as a set of its role-value pairs. We further propose a method to conduct Link Prediction on N-ary relational data, thus called NaLP, which explicitly models the relatedness of all the role-value pairs in the same n-ary relational fact. Experimental results validate the effectiveness and merits of the proposed NaLP method.

Answer selection is an important problem in community question answering (CQA), as it enables the distilling of reliable information and knowledge. Most existing approaches tackle this problem as a text matching task. However, they ignore the influence of the community in voting the best answers. Answer quality is highly correlated with semantic relevance and user expertise in CQA. In this paper, we formalize the answer selection problem from the user expertise view, considering both the semantic relevance in question-answer pair and user expertise in question-user pair. We design a novel matching function, explicitly modeling the influence of user expertise in community acceptance. Moreover, we introduce latent user vectors into the representation learning of answer, capturing the implicit topic interests in learned user vectors. Extensive experiments on two datasets from real world CQA sites demonstrate that our model outperforms state-of-the-art approaches for answer selection in CQA. Furthermore, the user representations learned by our model provide us a quantitative way to understand both the authority and topic-sensitive interests of users.

Users are usually involved in multiple social networks, without explicit anchor links that reveal the correspondence among different accounts of the same user across networks. Anchor link prediction aims to identify the hidden anchor links, which is a fundamental problem for user profiling, information cascading, and cross-domain recommendation. Although existing methods perform well in the accuracy of anchor link prediction, the pairwise search manners on inferring anchor links suffer from big challenge when being deployed in practical systems. To combat the challenges, in this paper we propose a novel embedding and matching architecture to directly learn binary hash code for each node. Hash codes offer us an efficient index to filter out the candidate node pairs for anchor link prediction. Extensive experiments on synthetic and real world large-scale datasets demonstrate that our proposed method has high time efficiency without loss of competitive prediction accuracy in anchor link prediction.

How to quantify influences between users, seeing that social network users influence each other in their temporal behaviors? Previous work has directly defined an independent model parameter to capture the interpersonal influence between each pair of users. To do so, these models need a parameter for each pair of users, which results in high-dimensional models becoming easily trapped into the overfitting problem. However, such models do not consider how influences depend on each other if influences are sent from the same user or if influences are received by the same user. Therefore, we propose a model that defines parameters for every user with a latent influence vector and a susceptibility vector, opposite to define influences on user pairs. Such low-dimensional representations naturally cause the interpersonal influences involving the same user to be coupled with each other, thus reducing the model’s complexity. Additionally, the model can easily consider the temporal information and sentimental polarities of users’ messages. Finally, we conduct extensive experiments on two real-world Microblog datasets, showing that our model with such representations achieves best performance on three prediction tasks, compared to the state-of-the-art and pair-wise baselines.

Ranking SVM, which formalizes the problem of learning a ranking model as that of learning a binary SVM on preference pairs of documents, is a state-of-the-art ranking model in information retrieval. The dual form solution of a linear Ranking SVM model can be written as a linear combination of the preference pairs, i.e., w = Σ(i,j)αijxi-xj), where αijdenotes the Lagrange parameters associated with each preference pair (i,j). It is observed that there exist obvious interactions among the document pairs because two preference pairs could share a same document as their items, e.g., preference pairs (d1,d2) and (d1,d3) share the document d1. Thus it is natural to ask if there also exist interactions over the model parameters αij, which may be leveraged to construct better ranking models. This paper aims to answer the question. We empirically found that there exists a low-rank structure over the rearranged Ranking SVM model parameters αij, which indicates that the interactions do exist. Based on the discovery, we made modifications on the original Ranking SVM model by explicitly applying low-rank constraints to the Lagrange parameters, achieving two novel algorithms called Factorized Ranking SVM and Regularized Ranking SVM, respectively. Specifically, in Factorized Ranking SVM each parameter αijis decomposed as a product of two low-dimensional vectors, i.e., αij=〈vi,vj〉, where vectors viand vjcorrespond to document i and j, respectively; In Regularized Ranking SVM, a nuclear norm is applied to the rearranged parameters matrix for controlling its rank. Experimental results on three LETOR datasets show that both of the proposed methods can outperform state-of-the-art learning to rank models including the conventional Ranking SVM.

Knowledge Graphs (KGs) have facilitated many real-world applications (e.g., vertical search and intelligent question answering). However, they are usually incomplete, which affects the performance of such KG based applications. To alleviate this problem, a number of Knowledge Graph Completion (KGC) methods have been developed to predict those implicit triples. Tensor/matrix based methods and translation based methods have attracted great attention for a long time. Recently, neural network has been introduced into KGC due to its extensive superiority in many fields (e.g., natural language processing and computer vision), and achieves promising results. In this paper, we propose a Shared Embedding based Neural Network (SENN) model for KGC. It integrates the prediction tasks of head entities, relations and tail entities into a neural network based framework with shared embeddings of entities and relations, while explicitly considering the differences among these prediction tasks. Moreover, we propose an adaptively weighted loss mechanism, which dynamically adjusts the weights of losses according to the mapping properties of relations, and the prediction tasks. Since relation prediction usually performs better than head and tail entity predictions, we further extend SENN to SENN+ by employing it to assist head and tail entity predictions. Experiments on benchmark datasets validate the effectiveness and merits of the proposed SENN and SENN+ methods. The shared embeddings and the adaptively weighted loss mechanism are also testified to be effective.

Hashing techniques for approximate nearest neighbor search (ANNS) encode data points into a set of short binary codes, while trying to preserve the neighborhood structure of the original data as much as possible. With the binary codes, the task of ANNS can be easily conducted over large-scale dataset, due to the high efficiency of pairwise comparison with the Hamming distance. Although binary codes have low computation and storage cost, the data are heavily compressed so that partial neighborhood structure information would be inevitably lost. To address this issue, we propose to introduce the k-nearest neighbors (k-NNs) in the original space into the Hamming space (i.e., associating a binary code with its original k-NNs) to enhance the effectiveness of existing hashing techniques with little overhead. Based on this idea, we develop a novel search scheme for hashing techniques namely neighborhood voting, i.e., each point retrieved by a query code will vote for its neighbors and itself, and the more voted, the better candidates. In this way, search in hashing is not simply the collision between codes (i.e., query code and candidate code), but also the collision between neighbors (i.e., neighbors of candidate points). The underlying assumption is that the true neighbors of a query point should be close to each other, while points with similar binary codes but seldom be the neighbors of other candidate points would be false positives. We introduce a novel data structure called aggregated hash table for implementing our idea and accelerating the online search process. Experimental results show that our search scheme can significantly improve the search effectiveness while having good efficiency over different existing hashing techniques.

In this paper, we introduce and tackle the Question Headline Generation (QHG) task. The motivation comes from the investigation of a real-world news portal where we find that news articles with question headlines often receive much higher click-through ratio than those with non-question headlines. The QHG task can be viewed as a specific form of the Question Generation (QG) task, with the emphasis on creating a natural question from a given news article by taking the entire article as the answer. A good QHG model thus should be able to generate a question by summarizing the essential topics of an article. Based on this idea, we propose a novel dual-attention sequence-to-sequence model (DASeq2Seq) for the QHG task. Unlike traditional sequence-to-sequence models which only employ the attention mechanism in the decoding phase for better generation, our DASeq2Seq further introduces a self-attention mechanism in the encoding phase to help generate a good summary of the article. We investigate two ways of the self-attention mechanism, namely global self-attention and distributed self-attention. Besides, we employ a vocabulary gate over both generic and question vocabularies to better capture the question patterns. Through the offline experiments, we show that our approach can significantly outperform the state-of-the-art question generation or headline generation models. Furthermore, we also conduct online evaluation to demonstrate the effectiveness of our approach using A/B test.

Assessing relevance between a query and a document is challenging in ad-hoc retrieval due to its diverse patterns, i.e., a document could be relevant to a query as a whole or partially as long as it provides sufficient information for users' need. Such diverse relevance patterns require an ideal retrieval model to be able to assess relevance in the right granularity adaptively. Unfortunately, most existing retrieval models compute relevance at a single granularity, either document-wide or passage-level, or use fixed combination strategy, restricting their ability in capturing diverse relevance patterns. In this work, we propose a data-driven method to allow relevance signals at different granularities to compete with each other for final relevance assessment. Specifically, we propose a HIerarchical Neural maTching model (HiNT) which consists of two stacked components, namely local matching layer and global decision layer. The local matching layer focuses on producing a set of local relevance signals by modeling the semantic matching between a query and each passage of a document. The global decision layer accumulates local signals into different granularities and allows them to compete with each other to decide the final relevance score.Experimental results demonstrate that our HiNT model outperforms existing state-of-the-art retrieval models significantly on benchmark ad-hoc retrieval datasets.

The goal of search result diversification is to select a subset of documents from the candidate set to satisfy as many different subtopics as possible. In general, it is a problem of subset selection and selecting an optimal subset of documents is NP-hard. Existing methods usually formalize the problem as ranking the documents with greedy sequential document selection. At each of the ranking position the document that can provide the largest amount of additional information is selected. It is obvious that the greedy selections inevitably produce suboptimal rankings. In this paper we propose to partially alleviate the problem with a Monte Carlo tree search (MCTS) enhanced Markov decision process (MDP), referred to as M$^2$Div. In M$^2$Div, the construction of diverse ranking is formalized as an MDP process where each action corresponds to selecting a document for one ranking position. Given an MDP state which consists of the query, selected documents, and candidates, a recurrent neural network is utilized to produce the policy function for guiding the document selection and the value function for predicting the whole ranking quality. The produced raw policy and value are then strengthened with MCTS through exploring the possible rankings at the subsequent positions, achieving a better search policy for decision-making. Experimental results based on the TREC benchmarks showed that M$^2$Div can significantly outperform the state-of-the-art baselines based on greedy sequential document selection, indicating the effectiveness of the exploratory decision-making mechanism in M$^2$Div.

Policy gradient, which makes use of Monte Carlo method to get an unbiased estimation of the parameter gradients, has been widely used in reinforcement learning. One key issue in policy gradient is reducing the variance of the estimation. From the viewpoint of statistics, policy gradient with baseline, a successful variance reduction method for policy gradient, directly applies the control variates method, a traditional variance reduction technique used in Monte Carlo, to policy gradient. One problem with control variates method is that the quality of estimation heavily depends on the choice of the control variates. To address the issue and inspired by the antithetic variates method for variance reduction, we propose to combine the antithetic variates method with traditional policy gradient for the multi-armed bandit problem. Furthermore, we achieve a new policy gradient algorithm called Antithetic-Arm Bandit (AAB). In AAB, the gradient is estimated through coordinate ascent where at each iteration gradient of the target arm is estimated through: 1) constructing a sequence of arms which is approximately monotonic in terms of estimated gradients, 2) sampling a pair of antithetic arms over the sequence, and 3) re-estimating the target gradient based on the sampled pair. Theoretical analysis proved that AAB achieved an unbiased and variance reduced estimation. Experimental results based on a multi-armed bandit task showed that AAB can achieve state-of-the-art performances.

A knowledge graph is a graph with entities of different types as nodes and various relations among them as edges. The construction of knowledge graphs in the past decades facilitates many applications, such as link prediction, web search analysis, question answering, and so on. Knowledge graph embedding aims to represent entities and relations in a large-scale knowledge graph as elements in a continuous vector space. Existing methods, for example, TransE, TransH, and TransR, learn the embedding representation by defining a global margin-based loss function over the data. However, the loss function is determined during experiments whose parameters are examined among a closed set of candidates. Moreover, embeddings over two knowledge graphs with different entities and relations share the same set of candidates, ignoring the locality of both graphs. This leads to the limited performance of embedding related applications. In this article, a locally adaptive translation method for knowledge graph embedding, called TransA, is proposed to find the loss function by adaptively determining its margin over different knowledge graphs. Then the convergence of TransA is verified from the aspect of its uniform stability. To make the embedding methods up-to-date when new vertices and edges are added into the knowledge graph, the incremental algorithm for TransA, called iTransA, is proposed by adaptively adjusting the optimal margin over time. Experiments on four benchmark data sets demonstrate the superiority of the proposed method, as compared to the state-of-the-art ones.

Online social media remarkably facilitates the production and delivery of information, intensifying the competition among vast information for users' attention and highlighting the importance of predicting the popularity of information. Existing approaches for popularity prediction fall into two paradigms: feature-based approaches and generative approaches. Feature-based approaches extract various features (e.g., user, content, structural, and temporal features), and predict the future popularity of information by training a regression/classification model. Their predictive performance heavily depends on the quality of hand-crafted features. In contrast, generative approaches devote to characterizing and modeling the process that a piece of information accrues attentions, offering us high ease to understand the underlying mechanisms governing the popularity dynamics of information cascades. But they have less desirable predictive power since they are not optimized for popularity prediction. In this paper, we propose DeepHawkes to combat the defects of existing methods, leveraging end-to-end deep learning to make an analogy to interpretable factors of Hawkes process --- a widely-used generative process to model information cascade. DeepHawkes inherits the high interpretability of Hawkes process and possesses the high predictive power of deep learning methods, bridging the gap between prediction and understanding of information cascades. We verify the effectiveness of DeepHawkes by applying it to predict retweet cascades of Sina Weibo and citation cascades of a longitudinal citation dataset. Experimental results demonstrate that DeepHawkes outperforms both feature-based and generative approaches.

When applying learning to rank algorithms to Web search, a large number of features are usually designed to capture the relevance signals. Most of these features are computed based on the extracted textual elements, link analysis, and user logs. However, Web pages are not solely linked texts, but have structured layout organizing a large variety of elements in different styles. Such layout itself can convey useful visual information, indicating the relevance of a Web page. For example, the query-independent layout (i.e., raw page layout) can help identify the page quality, while the query-dependent layout (i.e., page rendered with matched query words) can further tell rich structural information (e.g., size, position and proximity) of the matching signals. However, such visual information of layout has been seldom utilized in Web search in the past. In this work, we propose to learn rich visual features automatically from the layout of Web pages (i.e., Web page snapshots) for relevance ranking. Both query-independent and query-dependent snapshots are considered as the new inputs. We then propose a novel visual perception model inspired by human's visual search behaviors on page viewing to extract the visual features. This model can be learned end-to-end together with traditional human-crafted features. We also show that such visual features can be efficiently acquired in the online setting with an extended inverted indexing scheme. Experiments on benchmark collections demonstrate that learning visual features from Web page snapshots can significantly improve the performance of relevance ranking in ad-hoc Web retrieval tasks.

Information gathered from multiple sources on the Web often exhibits conflicts. This phenomenon motivates the need of truth discovery, which aims to automatically find the true claim among multiple conflicting claims. Existing truth discovery methods are mainly based on iterative updates or probabilistic models. In particular, iterative methods specify rules that govern how credibility flows from sources to claims and then back to sources. However, these manually-defined rules tend to be ad hoc and are difficult to adapt and analyze. Probabilistic methods model a few latent factors that impact how sources make claims, such as randomly choosing, guessing, or mistaking. However, these manually-defined factors may not well reflect the underlying data distributions. Given these limitations, we propose a new, unsupervised model for truth discovery in this paper. Our model first constructs a heterogenous network that exploits both source-claim and source-source relationships. It then embeds the network into a low dimensional space through a principled algorithm such that trustworthy sources and true claims (meanwhile, unreliable sources and false claims) are close. In this way, truth discovery can be conveniently performed in the embedding space. Compared with existing methods, our model does not need manually-defined rules or factors. Rather, it learns the embeddings automatically from data. Experiments on two real-world datasets demonstrate that our model outperforms existing state-of-the-art methods for truth discovery.

This paper concerns a deep learning approach to relevance ranking in information retrieval (IR). Existing deep IR models such as DSSM and CDSSM directly apply neural networks to generate ranking scores, without explicit understandings of the relevance. According to the human judgement process, a relevance label is generated by the following three steps: 1) relevant locations are detected; 2) local relevances are determined; 3) local relevances are aggregated to output the relevance label. In this paper we propose a new deep learning architecture, namely DeepRank, to simulate the above human judgment process. Firstly, a detection strategy is designed to extract the relevant contexts. Then, a measure network is applied to determine the local relevances by utilizing a convolutional neural network (CNN) or two-dimensional gated recurrent units (2D-GRU). Finally, an aggregation network with sequential integration and term gating mechanism is used to produce a global relevance score. DeepRank well captures important IR characteristics, including exact/semantic matching signals, proximity heuristics, query term importance, and diverse relevance requirement. Experiments on both benchmark LETOR dataset and a large scale clickthrough data show that DeepRank can significantly outperform learning to ranking methods, and existing deep learning methods.

Community detection or graph clustering is crucial to understanding the structure of complex networks and extracting relevant knowledge from networked data. Latent factor model, e.g., non-negative matrix factorization and mixed membership block model, is one of the most successful methods for community detection. Latent factor models for community detection aim to find a distributed and generally low-dimensional representation, or coding, that captures the structural regularity of network and reflects the community membership of nodes. Existing latent factor models are mainly based on reconstructing a network from the representation of its nodes, namely network decoder, while constraining the representation to have certain desirable properties. These methods, however, lack an encoder that transforms nodes into their representation. Consequently, they fail to give a clear explanation about the meaning of a community and suffer from undesired computational problems. In this paper, we propose a non-negative symmetric encoder-decoder approach for community detection. By explicitly integrating a decoder and an encoder into a unified loss function, the proposed approach achieves better performance over state-of-the-art latent factor models for community detection task. Moreover, different from existing methods that explicitly impose the sparsity constraint on the representation of nodes, the proposed approach implicitly achieves the sparsity of node representation through its symmetric and non-negative properties, making the optimization much easier than competing methods based on sparse matrix factorization.

One of the central issues in learning to rank for information retrieval is to develop algorithms that construct ranking models by directly optimizing evaluation measures such as normalized discounted cumulative gain~(ND CG). Existing methods usually focus on optimizing a specific evaluation measure calculated at a fixed position, e.g., NDCG calculated at a fixed position K. In information retrieval the evaluation measures, including the widely used NDCG and [email protected], are usually designed to evaluate the document ranking at all of the ranking positions, which provide much richer information than only measuring the document ranking at a single position. Thus, it is interesting to ask if we can devise an algorithm that has the ability of leveraging the measures calculated at all of the ranking postilions, for learning a better ranking model. In this paper, we propose a novel learning to rank model on the basis of Markov decision process (MDP), referred to as MDPRank. In the learning phase of MDPRank, the construction of a document ranking is considered as a sequential decision making, each corresponds to an action of selecting a document for the corresponding position. The policy gradient algorithm of REINFORCE is adopted to train the model parameters. The evaluation measures calculated at every ranking positions are utilized as the immediate rewards to the corresponding actions, which guide the learning algorithm to adjust the model parameters so that the measure is optimized. Experimental results on LETOR benchmark datasets showed that MDPRank can outperform the state-of-the-art baselines.

In this paper we address the issue of learning diverse ranking models for search result diversification. Typical methods treat the problem of constructing a diverse ranking as a process of sequential document selection. At each ranking position, the document that can provide the largest amount of additional information to the users is selected, because the search users usually browse the documents in a top-down manner. Thus, to select an optimal document for a position, it is critical for a diverse ranking model to capture the utility of information the user have perceived from the preceding documents. Existing methods usually calculate the ranking scores (e.g., the marginal relevance) directly based on the query and the selected documents, with heuristic rules or handcrafted features. The utility the user perceived at each of the ranks, however, is not explicitly modeled. In this paper, we present a novel diverse ranking model on the basis of continuous state Markov decision process (MDP) in which the user perceived utility is modeled as a part of the MDP state. Our model, referred to as MDP-DIV, sequentially takes the actions of selecting one document according to current state, and then updates the state for the chosen of the next action. The transition of the states are modeled in a recurrent manner and the model parameters are learned with policy gradient. Experimental results based on the TREC benchmarks showed that MDP-DIV can significantly outperform the state-of-the-art baselines.

Knowledge graph embedding aims to embed entities and relations of knowledge graphs into low-dimensional vector spaces. Translating embedding methods regard relations as the translation from head entities to tail entities, which achieve the state-of-the-art results among knowledge graph embedding methods. However, a major limitation of these methods is the time consuming training process, which may take several days or even weeks for large knowledge graphs, and result in great difficulty in practical applications. In this paper, we propose an efficient parallel framework for translating embedding methods, called ParTrans-X, which enables the methods to be paralleled without locks by utilizing the distinguished structures of knowledge graphs. Experiments on two datasets with three typical translating embedding methods, i.e., TransE [3], TransH [19], and a more efficient variant TransE- AdaGrad [11] validate that ParTrans-X can speed up the training process by more than an order of magnitude.

The queries issued to search engines are often ambiguous or multifaceted, which requires search engines to return diverse results that can fulfill as many different information needs as possible; this is called search result diversification . Recently, the relational learning to rank model, which designs a learnable ranking function following the criterion of maximal marginal relevance, has shown effectiveness in search result diversification [Zhu et al. 2014]. The goodness of a diverse ranking model is usually evaluated with diversity evaluation measures such as α-NDCG [Clarke et al. 2008], ERR-IA [Chapelle et al. 2009], and D#-NDCG [Sakai and Song 2011]. Ideally the learning algorithm would train a ranking model that could directly optimize the diversity evaluation measures with respect to the training data. Existing relational learning to rank algorithms, however, only train the ranking models by optimizing loss functions that loosely relate to the evaluation measures. To deal with the problem, we propose a general framework for learning relational ranking models via directly optimizing any diversity evaluation measure . In learning, the loss function upper-bounding the basic loss function defined on a diverse ranking measure is minimized. We can derive new diverse ranking algorithms under the framework, and several diverse ranking algorithms are created based on different upper bounds over the basic loss function. We conducted comparisons between the proposed algorithms with conventional diverse ranking methods using the TREC benchmark datasets. Experimental results show that the algorithms derived under the diverse learning to rank framework always significantly outperform the state-of-the-art baselines.

Machine learning algorithms have become the key components in many big data applications. However, the full potential of machine learning is still far from been realized because using machine learning algorithms is hard, especially on distributed platforms such as Hadoop and Spark. The key barriers come from not only the implementation of the algorithms themselves, but also the processing for applying them to real applications which often involve multiple steps and different algorithms. In this demo we present a general-purpose dataflow-based system for easing the process of applying machine learning algorithms to real world tasks. In the system, a learning task is formulated as a directed acyclic graph (DAG) in which each node represents an operation (e.g., a machine learning algorithm), and each edge represents the flow of the data from one node to its descendants. Graphical user interface is implemented for making users to create, configure, submit, and monitor a task in a drag-and-drop manner. Advantages of the system include 1) lowering the barriers of defining and executing machine learning tasks; 2) sharing and re-using the implementations of the algorithms, the task dataflow DAGs, and the (intermediate) experimental results; 3) seamlessly integrating the stand-alone algorithms as well as the distributed algorithms in one task. The system has been deployed as a machine learning service and can be access from the Internet.

Search result diversification has attracted considerable attention as a means to tackle the ambiguous or multi-faceted information needs of users. One of the key problems in search result diversification is novelty, that is, how to measure the novelty of a candidate document with respect to other documents. In the heuristic approaches, the predefined document similarity functions are directly utilized for defining the novelty. In the learning approaches, the novelty is characterized based on a set of handcrafted features. Both the similarity functions and the features are difficult to manually design in real world due to the complexity of modeling the document novelty. In this paper, we propose to model the novelty of a document with a neural tensor network. Instead of manually defining the similarity functions or features, the new method automatically learns a nonlinear novelty function based on the preliminary representation of the candidate document and other documents. New diverse learning to rank models can be derived under the relational learning to rank framework. To determine the model parameters, loss functions are constructed and optimized with stochastic gradient descent. Extensive experiments on three public TREC datasets show that the new derived algorithms can significantly outperform the baselines, including the state-of-the-art relational learning to rank models.

Demographic attributes play an important role in retail market to characterize different types of users. Such signals however are often only available for a small fraction of users in practice due to the difficulty in manual collection process by retailers. In this paper, we aim to harness the power of big data to automatically infer users' demographic attributes based on their purchase data. Typically, demographic prediction can be formalized as a multi-task multi-class prediction problem, i.e., multiple demographic attributes (e.g., gender, age and income) are to be inferred for each user where each attribute may belong to one of N possible classes (N-2). Most previous work on this problem explores different types of features and usually predicts different attributes independently. However, modeling the tasks separately may lose the ability to leverage the correlations among different attributes. Meanwhile, manually defined features require professional knowledge and often suffer from under specification. To address these problems, we propose a novel Structured Neural Embedding (SNE) model to automatically learn the representations from users' purchase data for predicting multiple demographic attributes simultaneously. Experiments are conducted on a real-world retail dataset where five attributes (gender, marital status, income, age, and education level) are to be predicted. The empirical results show that our SNE model can improve the performance significantly compared with state-of-the-art baselines.

In social networks, predicting a user’s location mainly depends on those of his/her friends, where the key lies in how to select his/her most influential friends. In this article, we analyze the theoretically maximal accuracy of location prediction based on friends’ locations and compare it with the practical accuracy obtained by the state-of-the-art location prediction methods. Upon observing a big gap between the theoretical and practical accuracy, we propose a new strategy for selecting influential friends in order to improve the practical location prediction accuracy. Specifically, several features are defined to measure the influence of the friends on a user’s location, based on which we put forth a sequential random-walk-with-restart procedure to rank the friends of the user in terms of their influence. By dynamically selecting the top N most influential friends of the user per time slice, we develop a temporal-spatial Bayesian model to characterize the dynamics of friends’ influence for location prediction. Finally, extensive experimental results on datasets of real social networks demonstrate that the proposed influential friend selection method and temporal-spatial Bayesian model can significantly improve the accuracy of location prediction.

There is always a lack of a cluster validity function and optimization strategy to find out clusters and catch the evolution trend of cluster structures on a categorical data stream. Therefore, this paper presents an optimization model for clustering categorical data streams. In the model, a cluster validity function is proposed as the objective function to evaluate the effectiveness of the clustering model while each new input data subset is flowing. It simultaneously considers the certainty of the clustering model and the continuity with the last clustering model in the clustering process. An iterative optimization algorithm is proposed to solve an optimal solution of the objective function with some constraints. Furthermore, we strictly derive a detection index for drifting concepts from the optimization model. We propose a detection method that integrates the detection index and the optimization model to catch the evolution trend of cluster structures on a categorical data stream. The new method can effectively avoid ignoring the effect of the clustering validity on the detection result. Finally, using the experimental studies on several real data sets, we illustrate the effectiveness of the proposed algorithm in clustering categorical data streams, compared with existing data-streams clustering algorithms.

Data sparsity is a long-standing challenge for recommender systems based on collaborative filtering. A promising solution for this problem is multi-context recommendation, i.e., leveraging users' explicit or implicit feedback from multiple contexts. In multi-context recommendation, various types of interactions between entities (users and items) are combined to alleviate data sparsity of a single context in a collective manner. Two issues are crucial for multi-context recommendation: (1) How to differentiate context-specific factors from entity-intrinsic factors shared across contexts? (2) How to capture the salient phenomenon that some entities are insensitive to contexts while others are remarkably context-dependent? Previous methods either do not consider context-specific factors, or assume that a context imposes equal influence on different entities, limiting their capability of combating data sparsity problem by taking full advantage of multiple contexts.

Ranking SVM, which formalizes the problem of learning a ranking model as that of learning a binary SVM on preference pairs of documents, is a state-of-the-art ranking model in information retrieval. The dual form solution of Ranking SVM model can be written as a linear combination of the preference pairs, i.e., w = ∑(i,j) αij (xi - xj), where αij denotes the Lagrange parameters associated with each pair (i,j). It is obvious that there exist significant interactions over the document pairs because two preference pairs could share a same document as their items. Thus it is natural to ask if there also exist interactions over the model parameters αij, which we may leverage to propose better ranking model. This paper aims to answer the question. Firstly, we found that there exists a low-rank structure over the Ranking SVM model parameters αij, which indicates that the interactions do exist. Then, based on the discovery, we made a modification on the original Ranking SVM model by explicitly applying a low-rank constraint to the parameters. Specifically, each parameter αij is decomposed as a product of two low-dimensional vectors, i.e., αij = vi, vj, where vectors vi and vj correspond to document i and j, respectively. The learning process, thus, becomes to optimize the modified dual form objective function with respect to the low-dimensional vectors. Experimental results on three LETOR datasets show that our method, referred to as Factorized Ranking SVM, can outperform state-of-the-art baselines including the conventional Ranking SVM.

Next basket recommendation is a crucial task in market basket analysis. Given a user's purchase history, usually a sequence of transaction data, one attempts to build a recommender that can predict the next few items that the user most probably would like. Ideally, a good recommender should be able to explore the sequential behavior (i.e., buying one item leads to buying another next), as well as account for users' general taste (i.e., what items a user is typically interested in) for recommendation. Moreover, these two factors may interact with each other to influence users' next purchase. To tackle the above problems, in this paper, we introduce a novel recommendation approach, namely hierarchical representation model (HRM). HRM can well capture both sequential behavior and users' general taste by involving transaction and user representations in prediction. Meanwhile, the flexibility of applying different aggregation operations, especially nonlinear operations, on representations allows us to model complicated interactions among different factors. Theoretically, we show that our model subsumes several existing methods when choosing proper aggregation operations. Empirically, we demonstrate that our model can consistently outperform the state-of-the-art baselines under different evaluation metrics on real-world transaction data.

In this paper we address the issue of learning a ranking model for search result diversification. In the task, a model concerns with both query-document relevance and document diversity is automatically created with training data. Ideally a diverse ranking model would be designed to meet the criterion of maximal marginal relevance, for selecting documents that have the least similarity to previously selected documents. Also, an ideal learning algorithm for diverse ranking would train a ranking model that could directly optimize the diversity evaluation measures with respect to the training data. Existing methods, however, either fail to model the marginal relevance, or train ranking models by minimizing loss functions that loosely related to the evaluation measures. To deal with the problem, we propose a novel learning algorithm under the framework of Perceptron, which adopts the ranking model that \emph{maximizes marginal relevance at ranking and can optimize any diversity evaluation measure in training}. The algorithm, referred to as PAMM (Perceptron Algorithm using Measures as Margins), first constructs positive and negative diverse rankings for each training query, and then repeatedly adjusts the model parameters so that the margins between the positive and negative rankings are maximized. Experimental results on three benchmark datasets show that PAMM significantly outperforms the state-of-the-art baseline methods.

Inferring a gold-standard ranking over a set of objects, such as documents or images, is a key task to build test collections for various applications like Web search and recommender systems. Crowdsourcing services provide an efficient and inexpensive way to collect judgments via labeling by sets of annotators. We thus study the problem of finding a consensus ranking from crowdsourced judgments. In contrast to conventional rank aggregation methods which minimize the distance between predicted ranking and input judgments from either pointwise or pairwise perspective, we argue that it is critical to consider the distance in a listwise way to emphasize the position importance in ranking. Therefore, we introduce a new listwise approach in this paper, where ranking measure based objective functions are utilized for optimization. In addition, we also incorporate the annotator quality into our model since the reliability of annotators can vary significantly in crowdsourcing. For optimization, we transform the optimization problem to the Linear Sum Assignment Problem, and then solve it by a very efficient algorithm named CrowdAgg guaranteeing the optimal solution. Experimental results on two benchmark data sets from different crowdsourcing tasks show that our algorithm is much more effective, efficient and robust than traditional methods.

Sentiment classification is a topic-sensitive task, i.e., a classifier trained from one topic will perform worse on another. This is especially a problem for the tweets sentiment analysis. Since the topics in Twitter are very diverse, it is impossible to train a universal classifier for all topics. Moreover, compared to product review, Twitter lacks data labeling and a rating mechanism to acquire sentiment labels. The extremely sparse text of tweets also brings down the performance of a sentiment classifier. In this paper, we propose a semi-supervised topic-adaptive sentiment classification (TASC) model, which starts with a classifier built on common features and mixed labeled data from various topics. It minimizes the hinge loss to adapt to unlabeled data and features including topic-related sentiment words, authors' sentiments and sentiment connections derived from“@” mentions of tweets, named as topic-adaptive features. Text and non-text features are extracted and naturally split into two views for co-training. The TASC learning algorithm updates topic-adaptive features based on the collaborative selection of unlabeled data, which in turn helps to select more reliable tweets to boost the performance. We also design the adapting model along a timeline (TASC-t) for dynamic tweets. An experiment on 6 topics from published tweet corpuses demonstrates that TASC outperforms other well-known supervised and ensemble classifiers. It also beats those semi-supervised learning methods without feature adaption. Meanwhile, TASC-t can also achieve impressive accuracy and F-score. Finally, with timeline visualization of “river” graph, people can intuitively grasp the ups and downs of sentiments' evolvement, and the intensity by color gradation.

With the coming of the era of big data, it is most urgent to establish the knowledge computational engine for the purpose of discovering implicit and valuable knowledge from the huge, rapidly dynamic, and complex network data. In this paper, we first survey the mainstream knowledge computational engines from four aspects and point out their deficiency. To cover these shortages, we propose the open knowledge network (OpenKN), which is a self-adaptive and evolutionable knowledge computational engine for network big data. To the best of our knowledge, this is the first work of designing the end-to-end and holistic knowledge processing pipeline in regard with the network big data. Moreover, to capture the evolutionable computing capability of OpenKN, we present the evolutionable knowledge network for knowledge representation. A case study demonstrates the effectiveness of the evolutionable computing of OpenKN.

Populating a knowledge base with new entity mentions extracted from unstructured text can help enhance its coverage and freshness. It naturally consists of two subtasks, namely, fine-grained entity classification and entity linking. Existing studies often focus on one of these two subtasks and they usually populate entity mentions in the same text by implicitly assuming that they are independent. However, these entity mentions are often semantically related to each other and it would be better to populate them into the knowledge base collectively. For solving these problems, in this paper we propose an interdependence graph based and unified collective inference approach, called CIIGA, to populating a knowledge base with collective entities, which can jointly determine the proper locations of all entity mentions in the same text by exploiting their interdependence relationships. Experimental results show that this approach can achieve significant accuracy improvement, as compared to the baseline approach, APOLLO, on the task of knowledge base population with multiple entities.

Sentiment analysis in various languages has been a research hotspot with many applications. However, sentiment resources (e.g., labeled corpora, sentiment lexicons) of different languages are unbalanced in terms of quality and quantity, which arouses interests in cross-lingual sentiment analysis aiming at using the resources in a source language to improve sentiment analysis in a target language. Nevertheless, many existing cross-lingual related works rely on a certain machine translation system to directly adapt the labeled data from the source language to the target language, which usually suffers from inaccurate results generated by the machine translation system. On the other hand, most sentiment analysis studies focus on document-level sentiment classification that cannot solve the aspect dependency problem of sentiment words. For instance, in the reviews on a cell phone, long is positive for the lifespan of its battery, but negative for the response time of its operating system. To solve these problems, this paper develops a novel Cross-Lingual Joint Aspect/Sentiment (CLJAS) model to carry out aspect-specific sentiment analysis in a target language using the knowledge learned from a source language. Specifically, the CLJAS model jointly detects aspects and sentiments of two languages simultaneously by incorporating sentiments into a cross-lingual topic model framework. Extensive experiments on different domains and different languages demonstrate that the proposed model can significantly improve the accuracy of sentiment classification in the target language.

Influence maximization, fundamental for word-of-mouth marketing and viral marketing, aims to find a set of seed nodes maximizing influence spread on social network. Early methods mainly fall into two paradigms with certain benefits and drawbacks: (1) Greedy algorithms, selecting seed nodes one by one, give a guaranteed accuracy relying on the accurate approximation of influence spread with high computational cost; (2) Heuristic algorithms, estimating influence spread using efficient heuristics, have low computational cost but unstable accuracy. We first point out that greedy algorithms are essentially finding a self-consistent ranking, where nodes' ranks are consistent with their ranking-based marginal influence spread. This insight motivates us to develop an iterative ranking framework, i.e., IMRank, to efficiently solve influence maximization problem under independent cascade model. Starting from an initial ranking, e.g., one obtained from efficient heuristic algorithm, IMRank finds a self-consistent ranking by reordering nodes iteratively in terms of their ranking-based marginal influence spread computed according to current ranking. We also prove that IMRank definitely converges to a self-consistent ranking starting from any initial ranking. Furthermore, within this framework, a last-to-first allocating strategy and a generalization of this strategy are proposed to improve the efficiency of estimating ranking-based marginal influence spread for a given ranking. In this way, IMRank achieves both remarkable efficiency and high accuracy by leveraging simultaneously the benefits of greedy algorithms and heuristic algorithms. As demonstrated by extensive experiments on large scale real-world social networks, IMRank always achieves high accuracy comparable to greedy algorithms, while the computational cost is reduced dramatically, about 10-100 times faster than other scalable heuristics.

When applying learning to rank algorithms in real search applications, noise in human labeled training data becomes an inevitable problem which will affect the performance of the algorithms. Previous work mainly focused on studying how noise affects ranking algorithms and how to design robust ranking algorithms. In our work, we investigate what inherent characteristics make training data robust to label noise. The motivation of our work comes from an interesting observation that a same ranking algorithm may show very different sensitivities to label noise over different data sets. We thus investigate the underlying reason for this observation based on two typical kinds of learning to rank algorithms (i.e.~pairwise and listwise methods) and three different public data sets (i.e.~OHSUMED, TD2003 and MSLR-WEB10K). We find that when label noise increases in training data, it is the \emph{document pair noise ratio} (i.e.~\emph{pNoise}) rather than \emph{document noise ratio} (i.e.~\emph{dNoise}) that can well explain the performance degradation of a ranking algorithm.

Search result diversification has gained attention as a way to tackle the ambiguous or multi-faceted information needs of users. Most existing methods on this problem utilize a heuristic predefined ranking function, where limited features can be incorporated and extensive tuning is required for different settings. In this paper, we address search result diversification as a learning problem, and introduce a novel relational learning-to-rank approach to formulate the task. However, the definitions of ranking function and loss function for the diversification problem are challenging. In our work, we firstly show that diverse ranking is in general a sequential selection process from both empirical and theoretical aspects. On this basis, we define ranking function as the combination of relevance score and diversity score between the current document and those previously selected, and loss function as the likelihood loss of ground truth based on Plackett-Luce model, which can naturally model the sequential generation of a diverse ranking list. Stochastic gradient descent is then employed to conduct the unconstrained optimization, and the prediction of a diverse ranking list is provided by a sequential selection process based on the learned ranking function. The experimental results on the public TREC datasets demonstrate the effectiveness and robustness of our approach.

Short texts are popular on today's web, especially with the emergence of social media. Inferring topics from large scale short texts becomes a critical but challenging task for many content analysis tasks. Conventional topic models such as latent Dirichlet allocation (LDA) and probabilistic latent semantic analysis (PLSA) learn topics from document-level word co-occurrences by modeling each document as a mixture of topics, whose inference suffers from the sparsity of word co-occurrence patterns in short texts. In this paper, we propose a novel way for short text topic modeling, referred as biterm topic model (BTM). BTM learns topics by directly modeling the generation of word co-occurrence patterns (i.e., biterms) in the corpus, making the inference effective with the rich corpus-level information. To cope with large scale short text data, we further introduce two online algorithms for BTM for efficient topic learning. Experiments on real-word short text collections show that BTM can discover more prominent and coherent topics, and significantly outperform the state-of-the-art baselines. We also demonstrate the appealing performance of the two online BTM algorithms on both time efficiency and topic learning.

Influence maximization, defined as a problem of finding a set of seed nodes to trigger a maximized spread of influence, is crucial to viral marketing on social networks. For practical viral marketing on large scale social networks, it is required that influence maximization algorithms should have both guaranteed accuracy and high scalability. However, existing algorithms suffer a scalability-accuracy dilemma: conventional greedy algorithms guarantee the accuracy with expensive computation, while the scalable heuristic algorithms suffer from unstable accuracy

Recently,`top-k learning to rank' has attracted much attention in the community of information retrieval. The motivation comes from the difficulty in obtaining a full-order ranking list for training, when employing reliable pairwise preference judgment. Inspired by the observation that users mainly care about top ranked search result, top-k learning to rank proposes to utilize top-k ground-truth for training, where only the total order of top k items are provided, instead of a full-order ranking list. However, it is not clear whether the underlying assumption holds, i.e. top-k ground-truth is sufficient for training. In this paper, we propose to study this problem from both empirical and theoretical aspects. Empirically, our experimental results on benchmark datasets LETOR4.0 show that the test performances of both pairwise and listwise ranking algorithms will quickly increase to a stable value, with the growth of k in the top-k ground-truth. Theoretically, we prove that the losses of these typical ranking algorithms in top-k setting are tighter upper bounds of (1--NDCG@k), compared with that in full-order setting. Therefore, our studies reveal that learning on top-k ground-truth is surely sufficient for ranking, which lay a foundation for the new learning to rank framework.

Sentiment classification is an important problem in tweets mining. There lack labeled data and rating mechanism for generating them in Twitter service. And topics in Twitter are more diverse while sentiment classifiers always dedicate themselves to a specific domain or topic. Thus it is a challenge to make sentiment classification adaptive to diverse topics without sufficient labeled data. Therefore we formally propose an adaptive multiclass SVM model which transfers an initial common sentiment classifier to a topic-adaptive one. To tackle the tweet sparsity, non-text features are explored besides the conventional text features, which are intuitively split into two views. An iterative algorithm is proposed for solving this model by alternating among three steps: optimization, unlabeled data selection and adaptive feature expansion steps. The algorithm alternatively minimizes the margins of two independent objectives on different views to learn coefficient matrices, which are collaboratively used for unlabeled tweets selection from the topic that the algorithm is adapting to. And then topic-adaptive sentiment words are expended based on the above selection, in turn to help the first two steps find more confident and unlabeled tweets and boost the final performance. Comparing with the well-known supervised sentiment classifiers and semi-supervised approaches, our algorithm achieves promising increases in accuracy averagely on the 6 topics from public tweet corpus.

Topic-focused multi-document summarization aims to produce a summary over a set of documents and conveys the most important aspects of a given topic. Most existing extractive methods view the task as a multi-criteria ranking problem over sentences, where relevance, salience and diversity are three typical requirements. However, diversity is a challenging problem as it involves modeling the relationship between sentences during ranking, where traditional methods usually tackle it in a heuristic or implicit way. In this paper, we propose a novel relational learning-to-rank approach (R-LTR) to solve this problem. Relational learning-to-rank is a new learning framework which further incorporates relationships into traditional learning-to-rank in an elegant way. Specifically, the ranking function is defined as the combination of content-based score of individual sentence, and relation-based score between the current sentence and those already selected. On this basis, we propose to learn the ranking function by minimizing the likelihood loss based on Plackett-Luce model, which can naturally model the sequential ranking procedure of candidate sentences. Stochastic gradient descent is then employed to conduct the learning process, and the summary is predicted by the greedy selection procedure based on the learned ranking function. Finally, we conduct extensive experiments on benchmark data sets TAC2008 and TAC2009. Experimental results show that our approach can significantly outperform the state-of-the-art methods from both quantitative and qualitative aspects.

Nowadays, short texts are very prevalent in various web applications, such as microblogs, instant messages. The severe sparsity of short texts hinders existing topic models to learn reliable topics. In this paper, we propose a novel way to tackle this problem. The key idea is to learn topics by exploring term correlation data, rather than the high-dimensional and sparse term occurrence information in documents. Such term correlation data is less sparse and more stable with the increase of the collection size, and can well capture the necessary information for topic learning. To obtain reliable topics from term correlation data, we first introduce a novel way to compute term correlation in short texts by representing each term with its co-occurred terms. Then we formulated the topic learning problem as symmetric non-negative matrix factorization on the term correlation matrix. After learning the topics, we can easily infer the topics of documents. Experimental results on three data sets show that our method provides substantially better performance than the baseline methods.

Recommender system has become an effective tool for information filtering, which usually provides the most useful items to users by a top-k ranking list. Traditional recommendation techniques such as Nearest Neighbors (NN) and Matrix Factorization (MF) have been widely used in real recommender systems. However, neither approaches can well accomplish recommendation task since that: (1) most NN methods leverage the neighbor's behaviors for prediction, which may suffer the severe data sparsity problem; (2) MF methods are less sensitive to sparsity, but neighbors' influences on latent factors are not fully explored, since the latent factors are often used independently. To overcome the above problems, we propose a new framework for recommender systems, called collaborative factorization. It expresses the user as the combination of his own factors and those of the neighbors', called collaborative latent factors, and a ranking loss is then utilized for optimization. The advantage of our approach is that it can both enjoy the merits of NN and MF methods. In this paper, we take the logistic loss in RankNet and the likelihood loss in ListMLE as examples, and the corresponding collaborative factorization methods are called CoF-Net and CoF-MLE. Our experimental results on three benchmark datasets show that they are more effective than several state-of-the-art recommendation methods.

It is well recognized that users rely on social media (e.g. Twitter or Digg) to fulfill two common needs (i.e. social need and informational need) that is to keep in touch with their friends in the real world and to have access to information they are interested in. Traditional friend recommendation methods in social media mainly focus on a user's social need, but seldom address their informational need (i.e. suggesting friends that can provide information one may be interested in but have not been able to obtain so far). In this paper, we propose to recommend friends according to the informational utility, which stands for the degree to which a friend satisfies the target user's unfulfilled informational need, called informational friend recommendation. In order to capture users' informational need, we view a post in social media as an item and utilize collaborative filtering techniques to predict the rating for each post. The candidate friends are then ranked according to their informational utility for recommendation. In addition, we also show how to further consider diversity in such recommendations. Experiments on benchmark datasets demonstrate that our approach can significantly outperform the traditional friend recommendation methods under informational evaluation measures.

Social tagging has grown in popularity on the web due to its effectiveness in organizing and accessing web pages. This short paper addresses the problem of automated social tagging, which aims to predict tags for web pages automatically and help with future navigation, filtering or search. We explore and find three foundations of the collaborative tags in social tagging services, that are consistency, sharability and stability. The complementary advantages are studied among three well-known methods, i.e. TF-weighted keyword extraction, collaborative filtering approach, and Corr-LDA (correspondence latent Dirichlet allocation) topic model. We then propose a blending model for automated social tagging to emphasize all the foundations, which linearly combines those tags generated by the three methods, and a permutation probability model is built to learn the linear blending. With the experiments on 50,000 training and 10,000 testing web pages from Delicious database, the results show that our blending method outperforms the four baselines. Furthermore, compared with both topic models, Corr-LDA and mixed membership LDA, our approach results in 14.2% and 25.6% of NDCG10improvement separately.

Learning low dimensional representations of text corpora is critical in many content analysis and data mining applications. It is even more desired and challenging to learn a sparse representation in practice for large scale text modeling. However, traditional probabilistic topic models (PTM) lack a mechanism to directly control the posterior sparsity of the inferred representations; While the emerged non-probabilistic models (NPM) can explicitly control sparsity using sparse constraint like l_1 norm, they convey different limitations in latent representations. To address the existing problems, we propose a novel non-probabilistic topic model for discovering sparse latent representations of large text corpora, referred as group sparse topical coding (GSTC). Our model enjoys both the merits of the PTMs and NPMs. On one hand, GSTC can naturally derive document-level admixture proportions in topic simplex like PTMs, which is useful for semantic analysis, classification or retrieval. On the other hand, GSTC can directly control the sparsity of the inferred representations with group lasso by relaxing the normalization constraint. Moreover, the relaxed non-probabilistic GSTC can be effectively learned using coordinate descent method. Experimental results on benchmark datasets show that GSTC can discover meaningful compact latent representations of documents, and improve the document classification accuracy and time efficiency.

Uncovering the topics within short texts, such as tweets and instant messages, has become an important task for many content analysis applications. However, directly applying conventional topic models (e.g. LDA and PLSA) on such short texts may not work well. The fundamental reason lies in that conventional topic models implicitly capture the document-level word co-occurrence patterns to reveal topics, and thus suffer from the severe data sparsity in short documents. In this paper, we propose a novel way for modeling topics in short texts, referred as biterm topic model (BTM). Specifically, in BTM we learn the topics by directly modeling the generation of word co-occurrence patterns (i.e. biterms) in the whole corpus. The major advantages of BTM are that 1) BTM explicitly models the word co-occurrence patterns to enhance the topic learning; and 2) BTM uses the aggregated patterns in the whole corpus for learning topics to solve the problem of sparse word co-occurrence patterns at document-level. We carry out extensive experiments on real-world short text collections. The results demonstrate that our approach can discover more prominent and coherent topics, and significantly outperform baseline methods on several evaluation metrics. Furthermore, we find that BTM can outperform LDA even on normal texts, showing the potential generality and wider usage of the new topic model.

Ranking is an important problem in various applications, such as Information Retrieval (IR), natural language processing, computational biology, and social sciences. Many ranking approaches have been proposed to rank objects according to their degrees of relevance or importance. Beyond these two goals, diversity has also been recognized as a crucial criterion in ranking. Top ranked results are expected to convey as little redundant information as possible, and cover as many aspects as possible. However, existing ranking approaches either take no account of diversity, or handle it separately with some heuristics. In this paper, we introduce a novel approach, Manifold Ranking with Sink Points (MRSPs), to address diversity as well as relevance and importance in ranking. Specifically, our approach uses a manifold ranking process over the data manifold, which can naturally find the most relevant and important data objects. Meanwhile, by turning ranked objects into sink points on data manifold, we can effectively prevent redundant objects from receiving a high rank. MRSP not only shows a nice convergence property, but also has an interesting and satisfying optimization explanation. We applied MRSP on two application tasks, update summarization and query recommendation, where diversity is of great concern in ranking. Experimental results on both tasks present a strong empirical performance of MRSP as compared to existing ranking approaches.

Bilingual sentiment lexicon is fundamental resource for cross-language sentiment analysis but its compilation remains a major bottleneck in computational linguistics. Traditional word alignment algorithm faces with the status of large alignment space, which may introduce redundant computations as well as alignment errors. In this paper, we use collocation alignment to extract bilingual sentiment lexicon overcoming the drawbacks of word alignment. The idea of collocation alignment is inspired by the strong cohesion between feature words and opinion words in sentiment corpus. Experimental results show that our approach not only decreases the computing time dramatically but also improves the precision of extracted bilingual word pairs due to the smaller alignment space.

This paper is concerned with top-k ranking problem, which reflects the fact that people pay more attention to the top ranked objects in real ranking application like information retrieval. A popular approach to top-k ranking problem is based on probabilistic models, such as Luce model and Mallows model. However, whether the sequential generative process described in these models is a suitable way for top-k ranking remains a question. According to the riffled independence factorization proposed in recent literature, which is a natural structural assumption on top-k ranking, we propose a new generative process of top-k ranking data. Our approach decomposes distributions over the top-k ranking into two layers: the first layer describes the relative ordering between the top k objects and the rest n-k objects, and the second layer describes the full ordering on the top k objects. On this basis, we propose a new probabilistic model for top-k ranking problem, called hierarchical ordering model. Specifically, we use three different probabilistic models to describe different generative processes of the first layer, and Luce model to describe the sequential generative process of the second layer, thus we obtain three different specific hierarchical ordering models. We also conduct extensive experiments on benchmark datasets to show that our proposed models can outperform previous models significantly.

In this paper, we aim to jointly extract aspects and aspect-specific sentiment knowledge from online reviews, where the sentiment knowledge refers to the aspect-specific opinion words along with their aspect-aware sentiment polarities. To this end, we propose a Joint Aspect/Sentiment model (JAS). JAS detects aspect-specific opinion words by integrating opinion word lexicon knowledge to explicitly separate opinion words from factual words. More importantly, JAS exploits sentiment prior and aspect-contextual sentence-level co-occurrences of opinion words in reviews to further identify aspect-aware sentiment polarities for the opinion words. We apply the learned aspect-specific sentiment knowledge to practical aspect-level sentiment analysis tasks. Experimental results show the effectiveness of JAS in learning aspect-specific sentiment knowledge and the practical value of this knowledge when applied to aspect-level sentiment classification.

Non-negative matrix factorization (NMF) has been successfully applied in document clustering. However, experiments on short texts, such as microblogs, Q&A documents and news titles, suggest unsatisfactory performance of NMF. An major reason is that the traditional term weighting schemes, like binary weight and tfidf, cannot well capture the terms' discriminative power and importance in short texts, due to the sparsity of data. To tackle this problem, we proposed a novel term weighting scheme for NMF, derived from the Normalized Cut (Ncut) problem on the term affinity graph. Different from idf, which emphasizes discriminability on document level, the Ncut weighting measures terms' discriminability on term level. Experiments on two data sets show our weighting scheme significantly boosts NMF's performance on short text clustering.

Query recommendation plays a critical role in helping users' search. Most existing approaches on query recommendation aim to recommend relevant queries. However, the ultimate goal of query recommendation is to assist users to reformulate queries so that they can accomplish their search task successfully and quickly. Only considering relevance in query recommendation is apparently not directly toward this goal. In this paper, we argue that it is more important to directly recommend queries with high utility, i.e., queries that can better satisfy users' information needs. For this purpose, we propose a novel generative model, referred to as Query Utility Model (QUM), to capture query utility by simultaneously modeling users' reformulation and click behaviors. The experimental results on a publicly released query log show that, our approach is more effective in helping users find relevant search results and thus satisfying their information needs.

In this paper, we propose a novel top-k learning to rank framework, which involves labeling strategy, ranking model and evaluation measure. The motivation comes from the difficulty in obtaining reliable relevance judgments from human assessors when applying learning to rank in real search systems. The traditional absolute relevance judgment method is difficult in both gradation specification and human assessing, resulting in high level of disagreement on judgments. While the pairwise preference judgment, as a good alternative, is often criticized for increasing the complexity of judgment from O(n) to (n log n). Considering the fact that users mainly care about top ranked search results, we propose a novel top-k labeling strategy which adopts the pairwise preference judgment to generate the top k ordering items from n documents (i.e. top-k ground-truth) in a manner similar to that of HeapSort. As a result, the complexity of judgment is reduced to O(n log k). With the top-k ground-truth, traditional ranking models (e.g. pairwise or listwise models) and evaluation measures (e.g. NDCG) no longer fit the data set. Therefore, we introduce a new ranking model, namely FocusedRank, which fully captures the characteristics of the top-k ground-truth. We also extend the widely used evaluation measures NDCG and ERR to be applicable to the top-k ground-truth, referred as κ-NDCG and κ-ERR, respectively. Finally, we conduct extensive experiments on benchmark data collections to demonstrate the efficiency and effectiveness of our top-k labeling strategy and ranking models.

The explosion of multilingual opinion data has made essential the need for automatic tools to analyze and understand people's opinions in multilingual reviews. In cross-language opinion mining, bilingual opinion lexicon plays a central role. Most of the bilingual lexicon extraction methods are based on EM algorithm. However, these methods are faced with two major problems: high complexity and unsatisfying precision. In this paper, we propose a novel approach to extract bilingual opinion lexicon using collocation alignment, where a collocation is a combination of a feature word and an opinion word. There are tight association between feature words and opinion words, which can be helpful to reduce the computation space and alignment errors. Experimental results demonstrate that our solution is effective and competitive.

Word-of-mouth has proven an effective strategy for promoting products through social relations. Particularly, existing studies have convincingly demonstrated that word-of-mouth recommendations can boost users' prior expectation and hence encourage them to adopt a certain innovation, such as buying a book or watching a movie. However, less attention has been paid to studying the posterior effect of word-of-mouth recommendations, i.e., whether or not word-of-mouth recommendations can influence users' posterior evaluation on the products or services recommended to them, the answer to which is critical to estimating user satisfaction when proposing a word-of-mouth marketing strategy. In order to fill this gap, in this paper we empirically study the above issue and verify that word-of-mouth recommendations are strongly associated with users' posterior evaluation. Through elaborately designed statistical hypothesis tests we prove the causality that word-of-mouth recommendations directly prompt the posterior evaluation of receivers. Finally, we propose a method for investigating users' social influence, namely, their ability to affect followers' posterior evaluation via word-of-mouth recommendations, by examining the number of their followers and their sensitivity of discovering good items. The experimental results on real datasets show that our method can successfully identify 78% influential friends with strong social influence.

This paper aims to find blog feeds having a principal inclination towards making opinionated comments on the given topic, so that we can subscribe to them to track influential and interesting opinions in the blogosphere. One major challenge is assigning topic-related opinion scores to blog feeds, which is embodied in two aspects. Firstly, we should identify whether the blog feed has a principal on-topic opinionated inclination. This inclination should be collectively revealed by all posts of the feed. We should fully consider evidences from all the posts of the feed to identify salient information among many posts of the feed. Secondly, we should capture topic-related opinions in the blog feed while ignoring irrelevant opinions.

Query similarity calculation is an important problem and has a wide range of applications in IR, including query recommendation, query expansion, and even advertisement matching. Existing work on query similarity aims to provide a single similarity measure without considering the fact that queries are ambiguous and usually have multiple search intents. In this paper, we argue that query similarity should be defined upon search intents, so-called intent-aware query similarity. By introducing search intents into the calculation of query similarity, we can obtain more accurate and also informative similarity measures on queries and thus help a variety of applications, especially those related to diversification. Specifically, we first identify the potential search intents of queries, and then measure query similarity under different intents using intent-aware representations. A regularized topic model is employed to automatically learn the potential intents of queries by using both the words from search result snippets and the regularization from query co-clicks. Experimental results confirm the effectiveness of intent-aware query similarity on ambiguous queries which can provide significantly better similarity scores over the traditional approaches. We also experimentally verified the utility of intent-aware similarity in the application of query recommendation, which can suggest diverse queries in a structured way to search users.

Many methods for cross-lingual processing tasks are resource-dependent, which will not work without machine translation system or bilingual lexicon. In this paper, we propose a novel approach for multilingual sentiment classification just by few seed words. For a given language, the proposed approach learns a sentiment classifier from the initial seed words instead of any labeled data. We employ our method both in supervised learning and unsupervised learning. Experimental results demonstrate that our method relies less on external resource but performs as well as or better than the baseline.

Query recommendation has been widely used in modern search engines. Recently, several context-aware methods have been proposed to improve the accuracy of recommendation by mining query sequence patterns from query sessions. However, the existing methods usually do not address the ambiguity of queries explicitly and often suffer from the sparsity of the training data. In this paper, we propose a novel context-aware query recommendation approach by modeling the high-order relation between queries and clicks in query log, which captures users' latent search intents. Empirical experiment results demonstrate that our approach outperforms the baseline methods in providing high quality recommendations for ambiguous queries.

Topic-focused multi-document summarization aims to produce a summary given a specific topic description and a set of related documents. It has become a crucial text processing task in many real applications that can help users consume the massive information. This paper presents a novel extractive approach based on supervised lazy random walk (Super Lazy). This approach naturally combines the rich features of sentences with the intrinsic sentence graph structure in a principled way, and thus enjoys the advantages of both the existing supervised and unsupervised approaches. Moreover, our approach can achieve the three major goals of topic-focused multi-document summarization (i.e. relevance, salience and diversity) simultaneously with a unified ranking process. Experiments on the benchmark dataset TAC2008 and TAC2009 are performed and the ROUGE evaluation results demonstrate that our approach can significantly outperform both the state-of-the-art supervised and unsupervised methods.

Many network-based ranking approaches have been proposed to rank objects according to different criteria, including relevance, prestige and diversity. However, existing approaches either only aim at one or two of the criteria, or handle them with additional heuristics in multiple steps. Inspired by DivRank, we propose a unified ranking model, Decayed DivRank (DDRank), to meet the three criteria simultaneously. Empirical experiments on paper citation network show that DDRank can outperform existing algorithms in capturing relevance, diversity and prestige simultaneously in ranking.

Related entity finding (REF) is a promising application, which aims to return a list of related entities given a query that describes the source entity, the specific type of target entities, and the relation between the source entity and target entities. One typical entity ranking strategy is to rank the candidate entities based on the co-occurrence between the candidate entities and the given query. However, such a strategy is inadequate to rank entities properly especially for those related but unpopular entities. In this paper, we propose a bipartite graph based entity ranking method, which leverage the Co-List relationship between candidate entities (i.e., entities co-occurring in the same structured/unstructured lists) to help improve the entity ranking. Specifically, we first estimate the initial relevance scores for the candidate entities based on a generative probabilistic model. We then construct a bipartite graph based on Co-List relation between candidate entities, and apply an iterative refinement process analogous to heat diffusion on the graph to propagate the relevance scores over entities. Finally the candidate entities are ranked according to their refined scores. We further develop an optimization framework for the iterative refinement process in our ranking method. Experimental results on the data collection from the TREC 2010 Entity Track show the effectiveness of our proposed method.

Query recommendation has been considered as an effective way to help search users in their information seeking activities. Traditional approaches mainly focused on recommending alternative queries with close search intent to the original query. However, to only take relevance into account may generate redundant recommendations to users. It is better to provide diverse as well as relevant query recommendations, so that we can cover multiple potential search intents of users and minimize the risk that users will not be satisfied. Besides, previous query recommendation approaches mostly relied on measuring the relevance or similarity between queries in the Euclidean space. However, there is no convincing evidence that the query space is Euclidean. It is more natural and reasonable to assume that the query space is a manifold. In this paper, therefore, we aim to recommend diverse and relevant queries based on the intrinsic query manifold. We propose a unified model, named manifold ranking with stop points, for query recommendation. By turning ranked queries into stop points on the query manifold, our approach can generate query recommendations by simultaneously considering both diversity and relevance in a unified way. Empirical experimental results on a large scale query log of a commercial search engine show that our approach can effectively generate highly diverse as well as closely related query recommendations.

Social media, e.g. Weblog and Internet forum, generate rich historical textual datasets which record lots of valuable events. Automatic event detection tries to discover important and interesting events and their related documents. Existing solutions to event detection, however, are mostly proposed for high quality news stories and may not work well when they are applied to noisy social media datasets, where content quality varies drastically from informative to trivial or even spamming. In this paper, an event detection framework, which directly utilizes burst property of events to filter out noise, is proposed. Experimental results on real dataset from Tencent Internet forum, a popular forum in China, demonstrate the effectiveness of the proposed framework.

Update summarization aims to create a summary over a topic-related multi-document dataset based on the assumption that the user has already read a set of earlier documents of the same topic. Beyond the problems (i.e., topic relevance, salience, and diversity in extracted information) tackled by topic-focused multi-document summarization, the update summarization must address the novelty problem as well. In this paper, we propose a novel extractive approach based on manifold ranking with sink points for update summarization. Specifically, our approach leverages a manifold ranking process over the sentence manifold to find topic relevant and salient sentences. More important, by introducing the sink points into sentence manifold, the ranking process can further capture the novelty and diversity based on the intrinsic sentence manifold. Therefore, we are able to address the four challenging problems above for update summarization in a unified way. Experiments on benchmarks of TAC are performed and the evaluation results show that our approach can achieve comparative performance to the existing best performing systems in TAC tasks.

Query recommendation has been recognized as an important mean to help users search and also improve the usability of search engines. Existing approaches mainly focus on helping users refine their search queries and the recommendations typically stick to users' search intent, named search interests in this paper. However, users may also have some vague or delitescent interests which they are unaware of until they are faced with one, named exploratory interests. These interests may be provoked within a search session when users read a web page from search results or even follow links on the page. By considering exploratory interests in query recommendation, we attract more user clicks on recommendations. This type of query recommendation has not been explicitly addressed in previous work. In this paper, we propose to recommend queries in a structured way for better satisfying both search and exploratory interests of users. Specifically, we construct a query relation graph from query logs and social annotation data which capture two types of interests respectively. Based on the query relation graph, we employ hitting time to rank possible recommendations, leverage a modularity based approach to group top recommendations into clusters, and label each cluster with social tags. Empirical experimental results indicate that our structured approach to query recommendation with social annotation data can better satisfy users' interests and significantly enhance users' click behavior on recommendations.

To improve web search effectiveness and help personalized search applications, it is important to understand users' search process, especially the underlying information goal transitions and satisfaction judgment on result pages. Unlike previous work modeling the two types of hidden information separately, the paper proposes to simultaneously model them based on users' full search process, including both queries and clicks. Thus, a full model can be built up and the dependences between them can be leveraged. Specially, we employ a hierarchical conditional random field (HCRF) for learning and prediction, with fruitful search activity features proposed and leveraged. Experimental results show that our approach reaches a high overall precision(87%) and significantly outperforms the baseline methods. Moreover, our model is applied in a re-ranking application and shows that it can benefit personalized web search.

Domain-oriented sentiment lexicons are widely used for fine-grained sentiment analysis on reviews; therefore, the automatic construction of domain-oriented sentiment lexicon is a fundamental and important task for sentiment analysis research. Most of existing construction approaches take only the kind of relationships between words into account, which makes them have a lot of room for improvement. This paper proposes an adapted information bottleneck method for the construction of domain-oriented sentiment lexicon. This approach can naturally make full use of the mutual reinforcement between documents and words by fusing three kinds of relationships either from words to documents or from words to words; either homogeneous or heterogeneous; either within-domain or cross-domain. The experimental results demonstrate that proposed method could dramatically improve the accuracy of the baseline approach on the construction of out-of-domain sentiment lexicon.

This paper addresses the problem of Named Entity Recognition in Query (NERQ), which involves detection of the named entity in a given query and classification of the named entity into predefined classes. NERQ is potentially useful in many applications in web search. The paper proposes taking a probabilistic approach to the task using query log data and Latent Dirichlet Allocation. We consider contexts of a named entity (i.e., the remainders of the named entity in queries) as words of a document, and classes of the named entity as topics. The topic model is constructed by a novel and general learning method referred to as WS-LDA (Weakly Supervised Latent Dirichlet Allocation), which employs weakly supervised learning (rather than unsupervised learning) using partially labeled seed entities. Experimental results show that the proposed method based on WS-LDA can accurately perform NERQ, and outperform the baseline methods.

Sentiment classification is attracting more and more attention because of its great benefits to social and human life. Usually supervised classification approaches perform well in sentiment classification, but the performance decreases sharply when transferred from one domain to another domain. In this paper, we propose an approach, SentiRank, which integrates the sentiment orientations of the documents into the graph-ranking algorithm for cross-domain sentiment classification. We apply the graph-ranking algorithm using the accurate labels of old-domain documents as well as the “pseudo” labels of new-domain documents, and investigate their relative importance for cross-domain sentiment classification. The experiment results indicate that the proposed algorithm could improve the performance of cross-domain sentiment classification dramatically.

This paper addresses the problem of query classification (QC), which aims to classify Web search queries into one or more predefined categories. The state-of-the-art solution for QC is to employ a bridging classifier via an intermediate taxonomy. In this paper, we advanced the bridging method by leveraging probabilistic topic models. The topic model, referred as RCTM (Regularized Correlated Topic Model), is an extension of the conventional CTM (Correlated Topic Model). RCTM learns a topic model by leveraging weak supervision from existing annotated data rather than in an unsupervised fashion, and thus it can effectively address the problem in topic modeling while the topics are predefined. The experimental evaluations show that our QC approach outperforms other baseline methods.

Topic representation mismatch is a key problem in topic-oriented summarization for the specified topic is usually too short to understand/interpret. This paper proposes a novel adaptive model for summarization, AdaSum, under the assumption that the summary and the topic representation can be mutually boosted. AdaSum aims to simultaneously optimize the topic representation and extract effective summaries. This model employs a mutual boosting process to minimize the topic representation mismatch for base summarizers. Furthermore, a linear combination of base summarizers is proposed to further reduce the topic representation mismatch from the diversity of base summarizers with a general learning framework. We prove that the training process of AdaSum can enhance the performance measure used. Experimental results on DUC 2007 dataset show that AdaSum significantly outperforms the baseline methods for summarization (e.g. MRP, LexRank, and GSPS).

This paper addresses the issue of query refinement, which involves reformulating ill-formed search queries in order to enhance relevance of search results. Query refinement typically includes a number of tasks such as spelling error correction, word splitting, word merging, phrase segmentation, word stemming, and acronym expansion. In previous research, such tasks were addressed separately or through employing generative models. This paper proposes employing a unified and discriminative model for query refinement. Specifically, it proposes a Conditional Random Field (CRF) model suitable for the problem, referred to as Conditional Random Field for Query Refinement (CRF-QR). Given a sequence of query words, CRF-QR predicts a sequence of refined query words as well as corresponding refinement operations. In that sense, CRF-QR differs greatly from conventional CRF models. Two types of CRF-QR models, namely a basic model and an extended model are introduced. One merit of employing CRF-QR is that different refinement tasks can be performed simultaneously and thus the accuracy of refinement can be enhanced. Furthermore, the advantages of discriminative models over generative models can be fully leveraged. Experimental results demonstrate that CRF-QR can significantly outperform baseline methods. Furthermore, when CRF-QR is used in web search, a significant improvement of relevance can be obtained.

In this work, we propose a novel scheme for sentiment classification (without labeled examples) which combines the strengths of both "learn-based" and "lexicon-based" approaches as follows: we first use a lexicon-based technique to label a portion of informative examples from given task (or domain); then learn a new supervised classifier based on these labeled ones; finally apply this classifier to the task. The experimental results indicate that proposed scheme could dramatically outperform "learn-based" and "lexicon-based" techniques.

Language model is widely used in many retrieval systems. Its document representation is based on the bag of words assumption. Hence, each term in document is treated as an equal object and only the term frequency is considered as the evidence of the importance of term. In this paper, we study the problem of cognition attention attenuation in processing documents and present a cognition attention attenuation based language model. This model estimates the document model by attenuation process of term in document. Compared with the classical language model, the advantage of this model is considering about the document structure which is often used in text summarization. From the experiments results, our novel cognition attention attenuation based language model outperformed the classical language model with Dirichlet smoothing in blog page and Web page.

Web forum has become an important resource on the Web due to its rich information contributed by millions of Internet users every day. Meanwhile, thousands of junk or valueless messages exist in Web forum. Recognizing high-quality topics should be fundamental tasks in search engine and Web mining systems. However, it is not a trivial problem to quantify high-quality topics on web forum. Users face a daunting challenge in identifying a small subset of topics worthy of their attention. In this paper, we present several characteristics to measure high-quality topic, based on these characteristics, we propose a novel model to recognize high-quality topics on Web forum. Our model consists of three steps. First, time series signals which contain distinctive characteristics between high-quality topics and non-high-quality topics are extracted from topics. Second, features are obtained from signals by using wavelet packet transform (WPT). Third, upon the features, high-quality topics are recognized by using backpropagation neural network. Conducting experiments on Tencent Message Boards which have 2,710,994 messages and 189,962 authors ranging from Jan 1, 2005 to Nov 12, 2007, we demonstrate the efficiency of our model, showing that the average accuracy rate of high-quality topic recognition is 95% and nearly 50,000 topics can be recognized in one second.

Finding the core members of a virtual community is an important problem in community analysis. Here we presented an simulated annealing algorithm to solve this problem by optimizing the user interests concentration ratio in user groups. As an example, we test this algorithm on a virtual community site and evaluate its results using human "gold standard" method.

Most template detection methods process web pages in batches that a newly crawled page can not be processed until enough pages have been collected. This results in large storage consumption and a huge delay of data refreshing. In this paper, we present an incremental framework to detect templates in which a page is processed as soon as it has been crawled. In this framework, we don't need to cache any web page. Experiments show that our framework consumes less than 7% storage than traditional methods. And also the speed of data refreshing is accelerated because of the incremental manner.

Previous on-line index maintenance strategies are mainly designed for document insertions without considering document deletions. In a truly dynamic search environment, however, documents may be added to and removed from the collection at any point in time. In this paper, we examine issues of on-line index maintenance with support for instantaneous document deletions and insertions. We present a DBT Merge strategy that can dynamically adjust the sequence of sub-index merge operations during index construction, and offers better query processing performance than previous methods, while providing an equivalent level of index maintenance performance when document insertions and deletions exist in parallel. Using experiments on 426 GB of web data we demonstrate the efficiency of our method in practice, showing that on-line index construction for dynamic text collections can be performed efficiently and almost as fast as for growing text collections.

Inspired by how search behavior works in human society, we propose CTO, a self-organized semantic overlay based on concept tree for P2P IR infrastructure, which is efficient for full text search in pure P2P environment without any central control or powerful peer as hub node. Especially, CTO performs very well on searching the unpopular resources shared by a few peers. In our experiment, while searching for the scarce documents shared by the peers, CTO achieves about 80% recall rate when the search covers less than 5% peers in the overlay. The search latency of CTO is also very low, which is controlled in the range about 5~12 hops.

In this work, we attempt to tackle domain-transfer problem by combining old-domain labeled examples with new-domain unlabeled ones. The basic idea is to use old-domain-trained classifier to label some informative unlabeled examples in new domain, and retrain the base classifier over these selected examples. The experimental results demonstrate that proposed scheme can significantly boost the accuracy of the base sentiment classifier on new domain.

In World Wide Web, contents of web documents play important roles in the evolution process because of their effects on linking preference. A majority of topological properties are content-related, and among them the clustering features are sensitive to contents of Web documents. In this paper, we first observe the impacts of content similarity on web links by introducing a metric called Linkage Probability. Then we investigate how contents influence the formation mechanism of the most basic cluster, triangle, with a metric named Triangularization Probability. Experimental results indicate that content similarity has a positive function in the process of cluster formation in theWeb. Theoretical analysis predicts the contents influence on the clustering features in the Web very well.

In the evolution process of World Wide Web, contents of web pages play important roles because of their direct effect on linking preference. In this paper, we propose a model which combines vertex connectivity and content similarity in a proportional manner. Analytical solutions indicate that our model exhibits a power-law degree distribution with variable exponent determined by the weight of content similarity. Distribution of content similarity on connected vertex pairs shows content similar web pages trend to be linked together. Simulation results show our model yields remarkably agreements of both degree and content similarity distributions with real network.

With the aim to reduce the dimensionality without sacrificing classification performance, the author gains insights from attribute reduction based on discernibility matrix in rough-set theory and proposes two text feature selection algorithms, i.e., DB1 and DB2. The experimental results indicate that DB2 not only yields much higher accuracy than information gain when the number of features is smaller than 6000, but also incurs much smaller CPU time than information gain.

In this paper we present a novel strategy, DragPushing, for improving the performance of text classifiers. The strategy is generic and takes advantage of training errors to successively refine the classification model of a base classifier. We describe how it is applied to generate two new classification algorithms; a Refined Centroid Classifier and a Refined Naïve Bayes Classifier. We present an extensive experimental evaluation of both algorithms on three English collections and one Chinese corpus. The results indicate that in each case, the refined classifiers achieve significant performance improvement over the base classifiers used. Furthermore, the performance of the Refined Centroid Classifier implemented is comparable, if not better, to that of state-of-the-art support vector machine (SVM)-based classifier, but offers a much lower computational cost.

We present a novel algorithm, DragPushing, for automatic text classification. Using a training data set, the algorithm first calculates the prototype vectors, or centroids, for each of the available document classes. Using misclassified examples, it then iteratively refines these centroids; by dragging the centroid of a correct class towards a misclassified example and in the same time pushing the centroid of an incorrect class away from the misclassified example. The algorithm is simple to implement and is computationally very efficient. Evaluation experiments conducted on two benchmark collections show that its classification accuracy is comparable to that of more complex methods, such as support vector machines (SVM).

This paper proposes an example-based phrase translation method in a Chinese to English cross-language information retrieval (CLIR) system. The method can generate much more accurate query translations than dictionary-based and common MT-based methods, and then improves the retrieval performance of our CLIR system.