Legal dispute mediation plays a crucial role in resolving civil disputes, yet its empirical study is limited by privacy constraints and complex multivariate interactions. To address this limitation, we present AgentMediation, the first LLM-based agent framework for simulating dispute mediation. It simulates realistic mediation processes grounded in real-world disputes and enables controlled experimentation on key variables such as disputant strategies, dispute causes, and mediator expertise. Our empirical analysis reveals patterns consistent with sociological theories, including Group Polarization and Surface-level Consensus. As a comprehensive and extensible platform, AgentMediation paves the way for deeper integration of social science and AI in legal research.

The rapid development of large language models (LLMs) has highlighted the need for efficient and reliable methods to evaluate their performance. Traditional evaluation methods often face challenges like high costs, limited task formats, dependence on human references, and systematic biases. To address these limitations, we propose Auto-PRE, an automatic LLM evaluation framework inspired by the peer review process. Unlike previous approaches that rely on human annotations, Auto-PRE automatically selects evaluator LLMs based on three core traits: consistency, pertinence, and self-confidence, which correspond to the instruction, content, and response stages, respectively, and collectively cover the entire evaluation process. Experiments on three representative tasks, including summarization, non-factoid QA, and dialogue generation, demonstrate that Auto-PRE achieves state-of-the-art performance while significantly reducing evaluation costs. Furthermore, the structured and scalable design of our automatic qualification exam framework provides valuable insights into automating the evaluation of LLMs-as-judges, paving the way for more advanced LLM-based evaluation frameworks.

Large Reasoning Models (LRMs) extend large language models with explicit, multi-step reasoning traces to enhance transparency and performance on complex tasks. However, these reasoning traces can be redundant or logically inconsistent, becoming a new and hard-to-detect source of hallucination. Existing hallucination detection methods focus primarily on answer-level uncertainty and often fail to detect hallucinations or logical inconsistencies arising from the model’s reasoning trace. This oversight is particularly problematic for LRMs, where the explicit thinking trace is not only an important support to the model's decision-making process but also a key source of potential hallucination. To this end, we propose RACE (Reasoning and Answer Consistency Evaluation), a novel framework specifically tailored for hallucination detection in LRMs. RACE operates by extracting essential reasoning steps and computing four diagnostic signals: inter-sample consistency of reasoning traces, entropy-based answer uncertainty, semantic alignment between reasoning and answers, and internal coherence of reasoning. This joint analysis enables fine-grained hallucination detection even when the final answer appears correct. Experiments across datasets and different LLMs demonstrate that RACE outperforms existing hallucination detection baselines, offering a robust and generalizable solution for evaluating LRMs.

As large language models (LLMs) increasingly operate as autonomous agents in social contexts, evaluating their capacity for prosocial behavior is both theoretically and practically critical. However, existing research has primarily relied on static, economically framed paradigms, lacking models that capture the dynamic evolution of prosociality and its sensitivity to structural inequities. To address these gaps, we introduce ProSim, a simulation framework for modeling the prosocial behavior in LLM agents across diverse social conditions. We conduct three progressive studies to assess prosocial alignment. First, we demonstrate that LLM agents can exhibit human-like prosocial behavior across a broad range of real-world scenarios and adapt to normative policy interventions. Second, we find that agents engage in fairness-based third-party punishment and respond systematically to variations in inequity magnitude and enforcement cost. Third, we show that policy-induced inequities suppress prosocial behavior, propagate norm erosion through social networks. These findings advance prosocial behavior theory by elucidating how institutional dynamics shape the emergence, decay, and diffusion of prosocial norms in agent-driven societies.

Large language models (LLMs) are largely static and often redo reasoning or repeat mistakes.Prior experience reuse typically relies on external retrieval, which is similarity-based, can introduce noise, and adds latency.We introduce SEAM (Structured Experience Adapter Module), a lightweight, executor-specific plugin that stores experience in its parameters and generates a structured, instance-tailored experience entry in a single forward pass to guide a frozen LLM executor.SEAM is trained for utility via executor rollouts and GRPO while keeping the executor frozen, and can be further improved with logged-success SFT after deployment.Experiments on mathematical reasoning benchmarks show consistent accuracy gains across executors with low overhead.Extensive ablation and analysis further elucidate the mechanisms underlying SEAM's effectiveness and robustness. 1 † Equal contribution.

Automating the drafting of judgment documents is pivotal to judicial efficiency, yet it remains challenging due to the dual requirements of comprehensive retrieval of legal information and rigorous logical reasoning. Existing approaches, typically relying on standard Retrieval-Augmented Generation and Supervised Fine-Tuning, often suffer from insufficient evidence recall, hallucinated statutory references, and logically flawed legal reasoning. To bridge this gap, we propose Judge-R1, a unified framework designed to enhance LLM-based judgment document generation by jointly improving legal information collection and judgment document generation. First, we introduce Agentic Legal Information Collection, which employs a dynamic planning agent to retrieve precise statutes and precedents from multiple sources. Second, we implement Rubric-Guided Optimization, a reinforcement learning phase utilizing Group Relative Policy Optimization (GRPO) with a comprehensive legal reward function to enforce adherence to judicial standards and reasoning logic. Extensive experiments on the JuDGE benchmark demonstrate that Judge-R1 significantly outperforms state-of-the-art baselines in both legal accuracy and generation quality.

The exponential growth of scientific literature has created a pressing need for automated survey generation. Although recent LLM-based agents have shown promise in automating this task, current progress is hindered by the lack of a standardized, scalable evaluation protocol. Existing evaluation methods typically rely on either human evaluation or custom metrics designed to validate specific pipelines, which restricts scalability and hinders fair comparison. To address this, we introduce SurGE, a benchmark and evaluation framework tailored for scientific survey generation. SurGE provides a large-scale retrieval corpus of over one million papers and expert-validated ground-truth surveys. Furthermore, we propose a robust multi-dimensional evaluation protocol that integrates both objective metrics and LLM-based judgments, and empirically verify its high alignment with human experts. Our experiments reveal that while agentic pipelines outperform RAG baselines in fluency and structural quality, they still struggle with citation accuracy, highlighting key directions for future research.

Analytical information needs, such as trend analysis and causal impact assessment, are prevalent across various domains including law, finance, science, and much more. However, existing information retrieval paradigms, whether based on relevance-oriented document ranking or retrieval-augmented generation (RAG) with large language models (LLMs), often struggle to meet the end-to-end requirements of such tasks at the corpus scale. They either emphasize information finding rather than end-to-end problem solving, or simply treat everything as question answering, offering limited control over reasoning, evidence usage, and verifiability. As a result, they struggle to support analytical queries that have diverse utility concepts and high accountability requirements.

Ranking plays a central role in connecting users and providers in Information Retrieval (IR) systems, making provider-side fairness an important challenge. While recent research has begun to address fairness in ranking, most existing approaches adopt an equality-based perspective, aiming to ensure that providers with similar content receive similar exposure. However, it overlooks the diverse needs of real-world providers, whose utility from ranking may depend not only on exposure but also on outcomes like sales or engagement. Consequently, exposure-based fairness may not accurately capture the true utility perceived by different providers with varying priorities. To this end, we introduce an equity-oriented fairness framework that explicitly models each provider's preferences over key outcomes such as exposure and sales, thus evaluating whether a ranking algorithm can fulfill these individualized goals while maintaining overall fairness across providers. Based on this framework, we develop EquityRank, a gradient-based algorithm that jointly optimizes user-side effectiveness and provider-side equity. Extensive offline and online simulations demonstrate that EquityRank offers improved trade-offs between effectiveness and fairness and adapts to heterogeneous provider needs.

Trial-and-error is a fundamental strategy for humans to solve complex problems and a necessary capability for Artificial Intelligence (AI) systems operating in real-world environments. Although several trial-and-error AI techniques have recently been proposed, most of them rely on simple heuristics designed by researchers and achieve limited performance gains. The core issue is the absence of appropriate data: current models cannot learn from detailed records of how humans actually conduct trial-and-error in practice. To address this gap, we introduce a data annotation platform and a corresponding dataset, termed Trial-and-Error Collection (TEC). The platform records users' complete trajectories across multiple trials and collects their reflections after receiving error feedback. Using this platform, we record the problem-solving processes of 46 participants on 58 tasks, resulting in 5,370 trial trajectories along with error reflections across 41,229 webpages. With this dataset, we observe that humans achieve substantially higher accuracy compared to LLMs, which demonstrates that humans are more effective in trial-and-error than LLMs. We believe that the TEC platform and dataset provide a valuable foundation for understanding human trial-and-error behavior and for developing more capable AI systems. Platform and dataset are publicly available. https://github.com/Serendipity0429/TEC.

Efficiently retrieving a concise set of candidates from a large doc- ument corpus remains a pivotal challenge in Information Retrieval (IR). Neural retrieval models, particularly dense retrieval models built with transformers and pretrained language models, have been popular due to their superior performance. However, criticisms have also been raised on their lack of explainability and vulnerability to adversarial attacks. In response to these challenges, we propose to improve the robustness of dense retrieval models by enhancing their sensitivity of fine-grained relevance signals. A model achieving sensitivity in this context should exhibit high variances when doc- uments' key passages determining their relevance to queries have been modified, while maintaining low variances for other changes in irrelevant passages. This sensitivity allows a dense retrieval model to produce robust results with respect to attacks that try to promote documents without actually increasing their relevance. It also makes it possible to analyze which part of a document is actually relevant to a query, and thus improve the explainability of the retrieval model. Motivated by causality and counterfactual analysis, we propose a se- ries of counterfactual regularization methods based on game theory and unsupervised learning with counterfactual passages. Specifically, we first introduce a cooperative game theory-based counterfactual passage extraction method, identifying the key passages that can influence relevance. Then we propose several subsequent unsuper- vised learning tasks, based on these counterfactual passages, serve to regularize the model's learning process to improve the robustness and sensitivity. Experiments show that, our method can extract key passages without reliance on the passage-level relevance annotations. Moreover, the regularized dense retrieval models exhibit heightened robustness against adversarial attacks, surpassing the state-of-the-art anti-attack methods.

Query rewriting is a fundamental technique in information retrieval (IR). It typically employs the retrieval result as relevance feedback to refine the query and thereby addresses the vocabulary mismatch between user queries and relevant documents. Traditional pseudo-relevance feedback (PRF) and its vector-based extension (VPRF) improve retrieval performance by leveraging top-retrieved documents as relevance feedback. However, they are constructed based on two major hypotheses: the relevance assumption (top documents are relevant) and the model assumption (rewriting methods need to be designed specifically for particular model architectures). While recent large language models (LLMs)-based generative relevance feedback (GRF) enables model-free query reformulation, it either suffers from severe LLM hallucination or, again, relies on the relevance assumption to guarantee the effectiveness of rewriting quality. To overcome these limitations, we introduce an assumption-relaxed framework: Generalized Pseudo Relevance Feedback (GPRF), which performs model-free, natural language rewriting based on retrieved documents, not only eliminating the model assumption but also reducing dependence on the relevance assumption. Specifically, we design a utility-oriented training pipeline with reinforcement learning to ensure robustness against noisy feedback. Extensive experiments across multiple benchmarks and retrievers demonstrate that GPRF consistently outperforms strong baselines, establishing it as an effective and generalizable framework for query rewriting.

Short videos have gained immense popularity, necessitating effective recommender systems that cater to individual preferences. The platforms use advanced algorithms to analyze user engagement and provide videos that satisfy users. A critical factor in user satisfaction is immersion , defined as the feeling of being deeply engaged in an activity. However, existing recommendation algorithms in the short video scenario have largely disregarded user immersion. Our study integrates user immersion into recommendation systems, aiming to predict immersion from user interactions and recommend items to enhance the overall viewing experience. Based on the user study of collecting and analyzing user immersion, we integrate immersion into the recommendations for both lab and large-scale scenarios. We adapt user-annotated immersion to large-scale real-world datasets without immersion labels. Specifically, we propose ImmersRec , an immersion-aware recommendation framework with immersion prediction fine-tuning, immersion knowledge alignment, and immersion-enhanced recommendation. Extensive experiments on two short video platforms indicate that our approach achieves significant enhancements among various context-aware recommender backbones. We investigate the predicted immersion and find it impacts not only short-term utility but also long-term user engagement. This research pioneers the incorporation of user immersion in short video recommendation algorithms, emphasizing its potential for improving recommendations with minimal data. The code can be available at https://github.com/hezy18/ImmersRec .

Large Language Models (LLMs) like ChatGPT and GPT-4 are versatile and capable of addressing open-domain question-answering(QA) tasks effectively. However, general LLMs, which are developed on open-domain data, may lack the domain-specific knowledge essential for tasks in vertical domains, such as legal, medical, etc. To address this issue, previous approaches either conduct continuous pre-training with domain-specific data or employ retrieval augmentation to support general LLMs in handling QA tasks. Unfortunately, these strategies are either cost-intensive or unreliable in practical applications. To this end, we present a novel framework named BLADE, which enhances Black-box LArge language models with small Domain-spEcific models. BLADE consists of a black-box LLM and a small domain-specific LM. The small LM preserves domain-specific knowledge and offers specialized insights, while the general LLM contributes robust language comprehension and reasoning capabilities. Specifically, our method involves three steps: 1) pre-training the small LM with domain-specific data, 2) fine-tuning this model using knowledge instruction data, and 3) joint Bayesian optimization of the general LLM and the small LM. In our experiments, we verify the effectiveness of BLADE on diverse LLMs and datasets across different domains. This shows the potential of BLADE as an effective and cost-efficient solution in adapting general LLMs for vertical domains.

Recent research demonstrates the effectiveness of using pre-trained language models for legal case retrieval. Most of the existing works focus on improving the representation ability for the contextualized embedding of the [CLS] token and calculate relevance using textual semantic similarity. However, in the legal domain, textual semantic similarity does not always imply that the cases are relevant enough. Instead, relevance in legal cases primarily depends on the similarity of key facts that impact the final judgment. Without proper treatments, the discriminative ability of learned representations could be limited since legal cases are lengthy and contain numerous non-key facts. To this end, we introduce DELTA, a discriminative model designed for legal case retrieval. The basic idea involves pinpointing key facts in legal cases and pulling the contextualized embedding of the [CLS] token closer to the key facts while pushing away from the non-key facts, which can warm up the case embedding space in an unsupervised manner. To be specific, this study brings the word alignment mechanism to the contextual masked auto-encoder. First, we leverage shallow decoders to create information bottlenecks, aiming to enhance the representation ability. Second, we employ the deep decoder to enable ``translation'' between different structures, with the goal of pinpointing key facts to enhance discriminative ability. Comprehensive experiments conducted on publicly available legal benchmarks show that our approach can outperform existing state-of-the-art methods in legal case retrieval. It provides a new perspective on the in-depth understanding and processing of legal case documents.

The use of large language models (LLMs) as automated evaluation tools to assess the quality of generated natural language, known as "LLMsas-Judges", has demonstrated promising capabilities and is rapidly gaining widespread attention.However, when applied to pairwise comparisons of candidate responses, LLM-based evaluators often exhibit selection bias.Specifically, their judgments may become inconsistent when the option positions or ID tokens are swapped, compromising the effectiveness and fairness of the evaluation result.To address this challenge, we introduce CalibraEval, a novel label-free method for mitigating selection bias during inference.Specifically, CalibraEval reformulates debiasing as an optimization task aimed at adjusting observed prediction distributions to align with unbiased prediction distributions.To solve this optimization problem, we propose a non-parametric order-preserving algorithm (NOA).This algorithm leverages the partial order relationships between model prediction distributions, thereby eliminating the need for explicit labels and precise mathematical function modeling.Empirical evaluations of LLMs in multiple representative benchmarks demonstrate that Cal-ibraEval effectively mitigates selection bias and improves performance compared to existing debiasing methods.This work marks a step toward building more robust and unbiased automated evaluation frameworks, paving the way for improved reliability in AI-driven assessments.The code can be found at https: //github.com/CSHaitao/CalibraEval.

Haitao Li, Junjie Chen, Jingli Yang, Qingyao Ai, Wei Jia, Youfeng Liu, Kai Lin, Yueyue Wu, Guozhi Yuan, Yiran Hu, Wuyue Wang, Yiqun Liu, Minlie Huang. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Existing retrieval-augmented code generation (RACG) methods typically use an external retrieval module to fetch semantically similar code snippets used for generating subsequent fragments.However, even for consecutive code fragments, the content often diverges due to logical progression, resulting in a content gap.This gap undermines the performance of current RACG methods, as external retrieval modules based on content matching fail to infer the specific information need of LLMs to generate the next code fragment.Therefore, we propose SelfRACG, a novel paradigm that enables large language models (LLMs) to Self-express their information needs to enhance RACG.Specifically, SelfRACG includes an information need expression module and a two-stage information need-guided training strategy, which encourages LLMs to express their information need.Extensive experiments demonstrate that SelfRACG can retrieve external knowledge that better aligns with the LLM's own information needs, resulting in superior generation performance compared to vanilla RACG.Moreover, both the training and deployment costs for retrieval in our framework are much lower than those of the strongest retrieval model. 1

As Large Language Models (LLMs) demonstrate increasingly strong human-like capabilities, the need to align them with human values has become significant. Recent advanced techniques, such as prompt learning and reinforcement learning, are being employed to bring LLMs closer to aligning with human values. While these techniques address broad ethical and helpfulness concerns, they rarely consider simulating individualized human values. To bridge this gap, we propose SimVBG, a framework that simulates individual values based on individual backstories that reflect their past experience and demographic information. SimVBG transforms structured data on an individual to a backstory and utilizes a multi-module architecture inspired by the Cognitive–Affective Personality System to simulate individual value based on the backstories. We test SimVBG on a self-constructed benchmark derived from the World Values Survey and show that SimVBG improves top-1 accuracy by more than 10% over the retrieval-augmented generation method. Further analysis shows that performance increases as additional interaction user history becomes available, indicating that the model can refine its persona over time. Code, dataset, and complete experimental results are available at https://github.com/bangdedadi/SimVBG.

Multi-agent techniques such as role playing or multi-turn debates have been shown to be effective in improving the performance of large language models (LLMs) in downstream tasks.Despite their differences in workflows, existing multi-agent systems constructed from a single base LLM mostly use natural language for agent communication.While this is appealing for its simplicity and interpretability, it also introduces inevitable information loss as one model must down sample its continuous state vectors to discrete tokens before transferring them to the other model.Such losses are particularly significant when the information to transfer is not simple facts, but reasoning logics or abstractive thoughts.To tackle this problem, we propose a new communication protocol that transfers both natural language tokens and token-wise state transition trajectory from one agent to another.Particularly, compared to the actual state value, we find that the sequence of state changes in LLMs after generating each token can better reflect the information hidden behind the inference process.We propose a State Delta Encoding (SDE) method to represent state transition trajectories.The experimental results show that multi-agent systems with SDE achieve SOTA performance compared to other communication protocols, particularly in tasks that involve complex reasoning.

Knowledge Editing is a technique that updates large language models (LLMs) with new information to maintain their world knowledge.This approach avoids the need to rebuild the model from scratch, thereby addressing the high costs associated with frequent retraining.Among these, the in-context editing paradigm stands out for its effectiveness in integrating new knowledge while preserving the model's original capabilities.Despite its potential, existing in-context knowledge editing methods are often task-specific, focusing primarily on multi-hop QA tasks using structured knowledge triples.Moreover, their reliance on fewshot prompting for task decomposition makes them unstable and less effective in generalizing across diverse tasks.In response to these limitations, we propose EditCoT, a novel knowledge editing framework that flexibly and efficiently updates LLMs across various tasks without retraining.EditCoT works by generating a chain-of-thought (CoT) for a given input and then iteratively refining this CoT process using a CoT editor based on updated knowledge.We evaluate EditCoT across a diverse range of benchmarks, covering multiple languages and tasks.The results demonstrate that our approach achieves state-of-the-art performance while offering superior generalization, effectiveness, and stability compared to existing methods, marking a significant advancement in the field of knowledge updating 1 .

Learning from preference feedback is a common practice for aligning large language models (LLMs) with human value. Conventionally, preference data is learned and encoded into a scalar reward model that connects a value head with an LLM to produce a scalar score as preference. However, scalar models lack interpretability and are known to be susceptible to biases in datasets. This paper investigates leveraging LLM itself to learn from such preference data and serve as a judge to address both limitations in one shot. Specifically, we prompt the pre-trained LLM to generate initial judgment pairs with contrastive preference in natural language form. The self-generated contrastive judgment pairs are used to train the LLM-as-a-Judge with Direct Preference Optimization (DPO) and incentivize its reasoning capability as a judge. This proposal of learning the LLMas-a-Judge using self-generated Contrastive judgments (Con-J) ensures natural interpretability through the generated rationales supporting the judgments, and demonstrates higher robustness against bias compared to scalar models. Experimental results show that Con-J outperforms the scalar reward model trained on the same collection of preference data, and outperforms a series of open-source and closed-source generative LLMs. We open-source the training process and model weights of Con-J at https://github.com/YeZiyi1998/Con-J.

The advent of generative artificial intelligence (AI), driven by advancements in large language models (LLMs), has unlocked transformative possibilities for information retrieval (IR), giving rise to a new wave of interactive and conversational paradigms. This workshop, titled New Interaction Paradigms for Information Retrieval in the Era of Generative AI, aims to serve as a collaborative platform for researchers and practitioners to explore the challenges and opportunities of integrating generative AI into IR systems. By focusing on tasks such as multi-turn conversational search, adaptive retrieval interfaces, and context-aware response generation, this workshop will address key areas including system design, user engagement, and evaluation methodologies. The workshop will also delve into broader concerns such as trust, transparency, and fairness, emphasizing the ethical implications of deploying generative AI in IR systems. Through panel discussions, poster sessions, and interactive roundtables, this workshop will foster critical dialogue and innovation, paving the way for a new era of user-centric, generative AI-powered IR systems.

User-generated content (UGC) communities, especially those featuring multimodal content, improve user experiences by integrating visual and textual information into results (or items).The challenge of improving user experiences in complex systems with search and recommendation (S&R) services has drawn significant attention from both academia and industry these years.However, the lack of high-quality datasets has limited the research progress on multimodal S&R.To address the growing need for developing better S&R services, we present a novel multimodal information retrieval dataset in this paper, namely Qilin.The dataset is collected from Xiaohongshu, a popular social platform with over 300 million monthly active users and an average search penetration rate of over 70%.In contrast to existing datasets, Qilin offers a comprehensive collection of user sessions with heterogeneous results like image-text notes, video notes, commercial notes, and direct answers, facilitating the development of advanced multimodal neural retrieval models across diverse task settings.To better model user satisfaction and support the analysis of heterogeneous user behaviors, we also collect extensive APP-level contextual signals and genuine user feedback.Notably, Qilin contains user-favored answers and their referred results for search requests triggering the

Retrieval-augmented generation (RAG) has proven highly effective in improving large language models (LLMs) across various domains. However, there is no benchmark specifically designed to assess the effectiveness of RAG in the legal domain, which restricts progress in this area. To fill this gap, we propose LexRAG, the first benchmark to evaluate RAG systems for multi-turn legal consultations. LexRAG consists of 1,013 multi-turn dialogue samples and 17,228 candidate legal articles. Each sample is annotated by legal experts and consists of five rounds of progressive questioning. LexRAG includes two key tasks: (1) Conversational knowledge retrieval, requiring accurate retrieval of relevant legal articles based on multi-turn context. (2) Response generation, focusing on producing legally sound answers. To ensure reliable reproducibility, we develop LexiT, a legal RAG toolkit that provides a comprehensive implementation of RAG system components tailored for the legal domain. Additionally, we introduce an LLM-as-a-judge evaluation pipeline to enable detailed and effective assessment. Through experimental analysis of various LLMs and retrieval methods, we reveal the key limitations of existing RAG systems in handling legal consultation conversations. LexRAG establishes a new benchmark for the practical application of RAG systems in the legal domain, with its code and data available at https://github.com/CSHaitao/LexRAG.

Counterfactual learning to rank (CLTR) has attracted extensive attention in the IR community for its ability to leverage massive logged user interaction data to train ranking models. While the CLTR models can be theoretically unbiased when the user behavior assumption is correct and the propensity estimation is accurate, their effectiveness is usually empirically evaluated via simulation-based experiments due to a lack of widely available, large-scale, real click logs. However, many previous simulation-based experiments are somewhat limited because they may have one or more of the following deficiencies: 1) using a weak production ranker to generate initial ranked lists, 2) relying on a simplified user simulation model to simulate user clicks, and 3) generating a fixed number of synthetic click logs. As a result, the robustness of CLTR models in complex and diverse situations is largely unknown and needs further investigation.

Retrieval-Augmented Generation (RAG) has become a foundational paradigm for enhancing large language models (LLMs) with external knowledge, playing an important role in modern information retrieval and knowledge-intensive NLP applications. Standard RAG systems typically adopt a static retrieve-then-generate pipeline and rely on in-context knowledge injection, which can be suboptimal for complex tasks that require multihop reasoning, adaptive information access, and deeper integration of external knowledge. Motivated by these limitations, the research community has moved beyond static retrieval and in-context knowledge injection. Among the emerging directions, this tutorial delves into two rapidly growing and complementary research directions on RAG: Dynamic RAG and Parametric RAG. Dynamic RAG explores how LLMs can actively decide when and what to retrieve during generation, enabling real-time adaptation to evolving information needs. Parametric RAG rethinks how the retrieved knowledge should be incorporated, moving from input-level to parameter-level knowledge injection for improved efficiency and effectiveness. This tutorial offers a comprehensive overview of recent advances in both directions. It provides participants with the theoretical foundations and actionable insights needed to build flexible and scalable RAG systems.

Retrieval-augmented generation (RAG) has emerged as a promising solution to enhance the reliability of large language models (LLMs) with external knowledge. Existing RAG methods share a common strategy for knowledge injection: they place the retrieved documents into the input context of the LLM, which we refer to as the in-context knowledge injection method. While this approach is simple and often effective, it has inherent limitations. Firstly, increasing the context length and number of relevant documents can lead to higher computational overhead and degraded performance, especially in complex reasoning tasks. More importantly, in-context knowledge injection operates primarily at the input level, but LLMs store their internal knowledge in their parameters. This gap fundamentally limits the capacity of in-context methods. To this end, we introduce Parametric RAG, a new RAG paradigm that integrates external knowledge directly into the feed-forward networks of an LLM through document parameterization. This approach not only reduces online computational costs by shortening the input context length, but also deepens the integration of external knowledge by enabling LLMs to utilize it in the same way as internal parametric knowledge. Experimental results demonstrate that Parametric RAG substantially enhances the effectiveness and efficiency of knowledge augmentation in LLMs. Also, it can be combined with in-context RAG methods to achieve even better performance. We have open-sourced all the code, data, and models in the following GitHub link: https://github.com/oneal2000/PRAG

This paper introduces JuDGE (Judgment Document Generation Evaluation), a novel benchmark for evaluating the performance of judgment document generation in the Chinese legal system. We define the task as generating a complete legal judgment document from the given factual description of the case. To facilitate this benchmark, we construct a comprehensive dataset consisting of factual descriptions from real legal cases, paired with their corresponding full judgment documents, which serve as the ground truth for evaluating the quality of generated documents. This dataset is further augmented by two external legal corpora that provide additional legal knowledge for the task: one comprising statutes and regulations, and the other consisting of a large collection of past judgment documents. In collaboration with legal professionals, we establish a comprehensive automated evaluation framework to assess the quality of generated judgment documents across various dimensions. We evaluate various baseline approaches, including few-shot in-context learning, fine-tuning, and a multi-source retrieval-augmented generation (RAG) approach, using both general and legal-domain LLMs. The experimental results demonstrate that, while RAG approaches can effectively improve performance in this task, there is still substantial room for further improvement. All the codes and datasets are available at: https://github.com/oneal2000/JuDGE

Retrieval-augmented generation (RAG) enhances large language models (LLMs) by integrating external knowledge retrieved from a knowledge base. However, its effectiveness is fundamentally constrained by the reliability of both the retriever and the knowledge base (i.e., the retrieval system). In real-world scenarios, imperfections in these components often lead to the retrieval of noisy, irrelevant, or misleading counterfactual information, ultimately undermining the trustworthiness of RAG systems. To address this challenge, we propose Robust Fine-Tuning (RbFT), a method designed to enhance the resilience of LLMs against retrieval defects through two targeted fine-tuning tasks. Experimental results demonstrate that RbFT significantly improves the robustness of RAG systems across diverse retrieval conditions, surpassing existing methods while maintaining high inference efficiency and compatibility with other robustness techniques.

Reconstructing visual images from brain signals is a rapidly evolving research with promising applications in brain-computer interfaces, cognitive neuroscience, and assistive technologies. While visual reconstruction based on functional Magnetic Resonance Imaging (fMRI) has previously achieved notable success, this paper explores cost-effective brain signals, i.e., electroencephalography (EEG) and magnetoencephalography (MEG). These signals are less precise than fMRI, which presents greater challenges for reconstruction. To address this problem, we propose BReAD (Brain Image Reconstruction with Retrieval-Augmented Diffusion), a novel framework that combines EEG/MEG signals with retrieval-augmented diffusion models to improve image reconstruction quality. BReAD utilizes the semantics decoded from brain signals for (1) retrieving semantic priors from a large-scale image database and (2) serving as a conditional constraint during the diffusion process. Extensive experiments demonstrate that BReAD significantly outperforms existing approaches in both qualitative and quantitative evaluations, paving the way for more robust and practical brain-to-image reconstruction systems. Our codes are available at https://github.com/Promise-Z5Q2SQ/BReAD.

Retrieval-Augmented Generation (RAG) systems have become a crucial tool to augment large language models (LLMs) with external knowledge for better task performance.However, existing traditional RAG methods inject knowledge directly into the context, resulting in several limitations.First, these methods highly rely on the in-context learning capability of LLMs, which often leads to excessively long contexts.This is inefficient due to the quadratic complexity of self-attention, leading to significant increase in inference time.Second, the extended context and the nature of self-attention can cause the LLMs to lose important information in the context, thereby degrading the original capabilities of LLMs.Third, the effectiveness of knowledge injection is perturbed by the permutation of knowledge within the extended context, reducing the robustness of existing RAG methods.To tackle the above problems, we propose DecoupledRAG, a method that decouples external knowledge from the context within the RAG framework.Specifically, we introduce a cross-attention based method that injects retrieved knowledge directly into the inference process of LLM on the fly, without modifying its parameters or the input context, so that the external knowledge can be utilized robustly in a permutation-independent manner.To the best of our knowledge, this is the first work that explore how to utilize cross-attention to inject knowledge with low training cost in decoder-only LLM era.By leveraging cross-attention operation, DecoupledRAG enables seamless knowledge aggregation without creating extended context.Experimental results demonstrate that our method could achieve

In legal case retrieval, existing work has shown that human-mediated conversational search can improve users’ search experience. One of the key problems for a practical conversational search system is how to ask high-quality clarifying questions to initiate conversations with users and understand their search intents. Previous works demonstrated that human-annotated external domain knowledge (such as event schemas) can improve the legal utility of clarifying questions generated by large language models. However, these methods are restricted to specific law systems or languages and cannot be generalized to others. To this end, we propose to generate context and domain-specific questions with LLMs without external annotations or knowledge by extracting information from top-retrieved documents given the current conversation context. Specifically, we construct a conversational legal case retrieval system CARQ that iteratively selects neighbor candidate case documents from the retrieved list at each conversation step to ask clarifying questions. We pretrain CARQ to capture the differences between legal cases and employ the reward augmented maximum likelihood to optimize the system directly for retrieval metrics. Extensive automated and human evaluations on three widely adopted legal case retrieval datasets demonstrate the superior effectiveness of our approach as compared with the state-of-the-art baselines.

Legal case retrieval is an important task in information retrieval that aims to retrieve relevant cases for given query cases. Conversational search paradigms have been shown to improve the search experience in legal case retrieval. However, there are two challenges in applying conversational search to legal scenarios. Firstly, legal search conversations often focus on different parts of legal case documents, but existing models struggle to capture the complex structural information and extract accurate relevance signals. Secondly, collecting large-scale conversational search datasets is costly, making it difficult to build reliable conversational legal case retrieval models. To address these challenges, we propose a Structure-Aware Matching Model (SAMM) for conversational legal case retrieval. SAMM extracts matching signals between conversational utterances and segments of the legal cases to incorporate structural information. We decouple the conversational search task into three subtasks and design pre-training tasks to overcome the lack of training data. Additionally, we create ConvLegal, the largest conversational legal case retrieval dataset to the best of our knowledge, for better evaluation of different methods. We train and evaluate SAMM and baselines on both a public dataset (CLCR) and ConvLegal. Experimental results demonstrate that SAMM outperforms existing models in legal case retrieval and conversational search.

Legal case retrieval aims to help legal workers find relevant cases related to their cases at hand, which is important for the guarantee of fairness and justice in legal judgments. While recent advances in neural retrieval methods have significantly improved the performance of open-domain retrieval tasks (e.g., Web search), their advantages haven’t been observed in legal case retrieval due to their thirst for annotated data. As annotating large-scale training data in legal domains is prohibitive due to the need for domain expertise, traditional search techniques based on lexical matching such as TF-IDF, BM25, and Query Likelihood are still prevalent in legal case retrieval systems. While previous studies have designed several pre-training methods for IR models in open-domain tasks, these methods are usually suboptimal in legal case retrieval because they cannot understand and capture the key knowledge and data structures in the legal corpus. To this end, we propose a novel pre-training framework named Caseformer that enables the pre-trained models to learn legal knowledge and domain-specific relevance-matching patterns in legal case retrieval without any human-labeled data. This framework is designed to support both dense retrieval models and neural re-ranking models. Through three unsupervised learning tasks, Caseformer is able to capture the special language, document structure, and relevance-matching patterns of legal case documents, making it a strong backbone for downstream legal case retrieval tasks. Experimental results show that our model has achieved state-of-the-art performance in both zero-shot and fine-tuning settings. Also, experiments on both Chinese and English legal datasets demonstrate that the effectiveness of Caseformer is language-independent in legal case retrieval.

Recently, dense retrieval (DR) models, which represent queries and documents with fixed-width vectors and retrieve relevant ones via nearest neighbor search, have drawn increasing attention from the IR community. However, previous studies have shown that the effectiveness of DR critically relies on sufficient training signals, which leads to severe performance degradation when applied in out-of-domain scenarios, where large-scale training data are usually unavailable. To solve this problem, existing studies adopt a data-augmentation-plus-joint-training paradigm to construct weak/pseudo supervisions on the target domain and combine them with the large-scale human annotated data on the source domain to train the DR models. However, they don't explicitly distinguish the data and the supervision signals in the training process and simply assume that the DR models are mighty enough to capture and memorize different domain knowledge and relevance matching patterns without guidance, which, as shown in this paper, is not true. Based on this observation, we propose a Robust Multi-Supervision Combining strategy (RMSC) that decouples the domain and supervision signals by explicitly telling the DR models how the domain data and supervision signals are combined in the training data with specially designed soft tokens. With the extra soft tokens to store the domain-specific and supervision-specific knowledge, RMSC allows the DR models to conduct retrieval based on human-like relevance matching patterns and target-specific language distribution on the target domain without human annotations. Extensive experiments on zero-shot DR benchmarks show that RMSC significantly improves the ranking performance on the target domain compared to strong DR baselines and domain adaptation methods, while being stable during training and can be combined with query generation or second-stage pre-training.

With the development of deep learning and natural language processing techniques, pre-trained language models have been widely used to solve information retrieval (IR) problems. Benefiting from the pre-training and fine-tuning paradigm, these models achieve state-of-the-art performance. In previous works, plain texts in Wikipedia have been widely used in the pre-training stage. However, the rich structured information in Wikipedia, such as the titles, abstracts, hierarchical heading (multi-level title) structure, relationship between articles, references, hyperlink structures, and the writing organizations, has not been fully explored. In this paper, we devise four pre-training objectives tailored for IR tasks based on the structured knowledge of Wikipedia. Compared to existing pre-training methods, our approach can better capture the semantic knowledge in the training corpus by leveraging the human-edited structured data from Wikipedia. Experimental results on multiple IR benchmark datasets show the superior performance of our model in both zero-shot and fine-tuning settings compared to existing strong retrieval baselines. Besides, experimental results in biomedical and legal domains demonstrate that our approach achieves better performance in vertical domains compared to previous models, especially in scenarios where long text similarity matching is needed. The code is available at https://github.com/oneal2000/Wikiformer.

Dynamic retrieval augmented generation (RAG) paradigm actively decides when and what to retrieve during the text generation process of Large Language Models (LLMs).There are two key elements of this paradigm: identifying the optimal moment to activate the retrieval module (deciding when to retrieve) and crafting the appropriate query once retrieval is triggered (determining what to retrieve).However, current dynamic RAG methods fall short in both aspects.Firstly, the strategies for deciding when to retrieve often rely on static rules.Moreover, the strategies for deciding what to retrieve typically limit themselves to the LLM's most recent sentence or the last few tokens, while the LLM's information needs may span across the entire context.To overcome these limitations, we introduce a new framework, DRAGIN, i.e., Dynamic Retrieval Augmented Generation based on the Information Needs of LLMs.Our framework is specifically designed to make decisions on when and what to retrieve based on the LLM's information needs during the text generation process.We evaluate DRAGIN along with existing methods comprehensively over 4 knowledge-intensive generation datasets.Experimental results show that DRAGIN achieves superior performance on all tasks, demonstrating the effectiveness of our method 1 .

Jingtao Zhan, Qingyao Ai, Yiqun Liu, Yingwei Pan, Ting Yao, Jiaxin Mao, Shaoping Ma, Tao Mei. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

The impressive performance of large language models (LLMs) has attracted considerable attention from the academic and industrial communities. Besides how to construct and train LLMs, how to effectively evaluate and compare the capacity of LLMs has also been well recognized as an important yet difficult problem. Existing paradigms rely on either human annotators or model-based evaluators to evaluate the performance of LLMs on different tasks. However, these paradigms often suffer from high cost, low generalizability, and inherited biases in practice, which make them incapable of supporting the sustainable development of LLMs in the long term. In order to address these issues, inspired by the peer review systems widely used in the academic publication process, we propose a novel framework that can automatically evaluate LLMs through a peer-review process. Specifically, for the evaluation of a specific task, we first construct a small qualification exam to select "reviewers'' from a couple of powerful LLMs. Then, to actually evaluate the "submissions" written by different candidate LLMs, i.e., the evaluatees, we use the reviewer LLMs to rate or compare the submissions. The final ranking of evaluatee LLMs is generated based on the results provided by all reviewers. We conducted extensive experiments on both text summarization and non-factoid question-answering tasks with eleven LLMs including GPT-4. The results demonstrate the existence of biasness when evaluating using a single LLM. Also, our PRE model outperforms all the baselines, illustrating the effectiveness of the peer review mechanism.

Legal question answering based on case documents is a pivotal legal AI application and helps extract key elements from the legal case documents to promote downstream tasks. Intuitively, the form of this task is similar to legal machine reading comprehension. However, in existing legal machine reading comprehension datasets, the background information is much shorter than the legal case documents, and the questions are not designed from the perspective of legal knowledge. In this paper, we present LeDQA, the first Chinese legal case document-based question answering dataset to our best knowledge. Specifically, we build a comprehensive question schema (including 48 element-based questions) for the Chinese civil law by legal professionals. And considering the cost of human annotations are too expensive, we use one of the SOTA LLMs (i.e., GPT-4) to annotate the relevant sentences to these questions in each case document. The constructed dataset originates from Chinese civil cases and contains 100 case documents, 4,800 case-question pairs and 132,048 sentence-level relevance annotations. We implement several text matching algorithms for relevant sentence selection and various Large Language Models(LLMs) for legal question answering on LeDQA. The experimental results indicate that incorporating relevant sentences can benefit the performance of question answering models, but further efforts are still required to address the remaining challenges such as retrieving irrelevant sentences and incorrect reasoning between retrieved sentences.

In the information retrieval scenario, query augmentation is an essential technique to refine semantically imprecise queries to align more closely with users' actual information needs. Traditional methods typically rely on extracting signals from user interactions such as browsing or clicking behaviors to augment the queries, which may not accurately reflect the actual user intent due to inherent noise and the dependency on initial user interactions. To overcome these limitations, we introduce Brain-Aug, a novel approach that decodes semantic information directly from brain signals of users to augment query representation. Brain-Aug builds on three techniques: (i) Structurally, an adapter network is utilized to project brain signals into the embedding space of a language model, allowing query augmentation conditioned on both the users' initial query and their brain signals. (ii) During training, we use a next token prediction task for query augmentation and adopt prompt tuning to efficiently train the brain adapter. (iii)At the inference stage, a ranking-oriented decoding strategy is implemented, enabling Brain-Aug to generate augmentations that improve ranking performance. We evaluate our approach on multiple functional magnetic resonance imaging (fMRI) datasets, demonstrating that Brain-Aug not only produces semantically richer queries but also significantly improves document ranking accuracy, particularly for ambiguous queries. These results validate the effectiveness of Brain-Aug, and reveal the potential of using internal cognitive states to understand and augment text-based queries. Supplementary materials and code are available at https://github.com/YeZiyi1998/Brain-Query-Augmentation.

Large language models (LLMs) have made significant progress in natural language processing tasks and demonstrate considerable potential in the legal domain. However, legal applications demand high standards of accuracy, reliability, and fairness. Applying existing LLMs to legal systems without careful evaluation of their potential and limitations could pose significant risks in legal practice.To this end, we introduce a standardized comprehensive Chinese legal benchmark LexEval.This benchmark is notable in the following three aspects: (1) Ability Modeling: We propose a new taxonomy of legal cognitive abilities to organize different tasks. (2) Scale: To our knowledge, LexEval is currently the largest Chinese legal evaluation dataset, comprising 23 tasks and 14,150 questions. (3) Data: we utilize formatted existing datasets, exam datasets and newly annotated datasets by legal experts to comprehensively evaluate the various capabilities of LLMs. LexEval not only focuses on the ability of LLMs to apply fundamental legal knowledge but also dedicates efforts to examining the ethical issues involved in their application.We evaluated 38 open-source and commercial LLMs and obtained some interesting findings. The experiments and findings offer valuable insights into the challenges and potential solutions for developing Chinese legal systems and LLM evaluation pipelines. The LexEval dataset and leaderboard are publicly available at https://github.com/CSHaitao/LexEval and will be continuously updated.

Large language models (LLMs) have demonstrated remarkable capabilities across various research domains, including the field of Information Retrieval (IR). However, the responses generated by off-the-shelf LLMs tend to be generic, i.e., cannot capture the distinctiveness of each document with similar content. This limits the performance of LLMs in IR because finding and distinguishing relevant documents from substantial similar documents is a typical problem in many IR tasks. To address this issue, we propose an unsupervised alignment method, namely Reinforcement Learning from Contrastive Feedback (RLCF), empowering LLMs to generate both high-quality and context-specific responses. Our approach constructs unsupervised contrastive feedback signals based on similar document groups, and adopts a reward function, named group-wise reciprocal rank, to optimize LLMs. We conduct extensive experiments to evaluate the effectiveness of RLCF.

Scaling laws have been observed in a wide range of tasks, particularly in language generation. Previous studies have found that the performance of large language models adheres to predictable patterns with respect to the size of models and datasets. This helps us design training strategies effectively and efficiently, especially as large-scale training becomes increasingly resource-intensive. Yet, in dense retrieval, such scaling law has not been fully explored. In this study, we investigate how scaling affects the performance of dense retrieval models. We implement dense retrieval models with different numbers of parameters, and train them with various amounts of annotated data. We propose to use the contrastive entropy as the evaluation metric, which is continuous compared with discrete ranking metrics and thus can accurately reflect model performance. Results indicate that the performance of dense retrieval models follows a precise power-law scaling related to the model size and the number of annotations across different datasets and annotation methods. Additionally, we show that the scaling laws help optimize the training process, such as resolving the resource allocation problem under a budget constraint. We believe that these findings significantly contribute to understanding the scaling effect of dense retrieval models and offer meaningful guidance for future research.

As an important component of intelligent legal systems, legal case retrieval plays a critical role in ensuring judicial justice and fairness. However, the development of legal case retrieval technologies in the Chinese legal system is restricted by three problems in existing datasets: limited data size, narrow definitions of legal relevance, and naive candidate pooling strategies used in data sampling.

The logs of the use of a search engine provide sufficient data to train a better ranker. However, it is well known that such implicit feedback reflects biases, and in particular a presentation bias that favors higher-ranked results. Unbiased Learning-to-Rank (ULTR) methods attempt to optimize performance by jointly modeling this bias along with the ranker so that the bias can be removed. Such methods have been shown to provide theoretical soundness, and promise superior performance and low deployment costs. However, existing ULTR methods don't recognize that query-document relevance is a confounder -- it affects both the likelihood of a result being clicked because of relevance and the likelihood of the result being ranked high by the base ranker. Moreover, the performance guarantees of existing ULTR methods assume the use of a weak ranker -- one that does a poor job of ranking documents based on relevance to a query. In practice, of course, commercial search engines use highly tuned rankers, and desire to improve upon them using the implicit judgments in search logs. This results in a significant correlation between position and relevance, which leads existing ULTR methods to overestimate click propensities in highly ranked results, reducing ULTR's effectiveness. This paper is the first to demonstrate the problem of propensity overestimation by ULTR algorithms, based on a causal analysis. We develop a new learning objective based on a backdoor adjustment. In addition, we introduce the Logging-Policy-aware Propensity (LPP) model that can jointly learn LPP and a more accurate ranker. We extensively test our approach on two public benchmark tasks and show that our proposal is effective, practical and significantly outperforms the state of the art.

Sequential recommender systems predict items that may interest users by modeling their preferences based on historical interactions. Traditional sequential recommendation methods rely on capturing implicit collaborative filtering signals among items. Recent relation-aware sequential recommendation models have achieved promising performance by explicitly incorporating item relations into the modeling of user historical sequences, where most relations are extracted from knowledge graphs. However, existing methods rely on manually predefined relations and suffer the sparsity issue, limiting the generalization ability in diverse scenarios with varied item relations.

Text-to-image generation systems have emerged as revolutionary tools in the realm of artistic creation, offering unprecedented ease in transforming textual prompts into visual art. However, the efficacy of these systems is intricately linked to the quality of user-provided prompts, which often poses a challenge to users unfamiliar with prompt crafting. This paper addresses this challenge by leveraging user reformulation data from interaction logs to develop an automatic prompt reformulation model. Our in-depth analysis of these logs reveals that user prompt reformulation is heavily dependent on the individual user's capability, resulting in significant variance in the quality of reformulation pairs. To effectively use this data for training, we introduce the Capability-aware Prompt Reformulation (CAPR) framework. CAPR innovatively integrates user capability into the reformulation process through two key components: the Conditional Reformulation Model (CRM) and Configurable Capability Features (CCF). CRM reformulates prompts according to a specified user capability, as represented by CCF. The CCF, in turn, offers the flexibility to tune and guide the CRM's behavior. This enables CAPR to effectively learn diverse reformulation strategies across various user capacities and to simulate high-capability user reformulation during inference. Extensive experiments demonstrate CAPR's superior performance over existing baselines and its remarkable robustness on unseen systems. Furthermore, comprehensive analyses validate the effectiveness of different components. CAPR can facilitate user-friendly interaction with text-to-image systems and make advanced artistic creation more achievable for a broader range of users.

In recent years, short video platforms have gained widespread popularity, making the quality of video recommendations crucial for retaining users. Existing recommendation systems primarily rely on behavioral data, which faces limitations when inferring user preferences due to issues such as data sparsity and noise from accidental interactions or personal habits. To address these challenges and provide a more comprehensive understanding of user affective experience and cognitive activity, we propose EEG-SVRec, the first EEG dataset with User Multidimensional Affective Engagement Labels in Short Video Recommendation. The study involves 30 participants and collects 3,657 interactions, offering a rich dataset that can be used for a deeper exploration of user preference and cognitive activity. By incorporating self-assessment techniques and real-time, low-cost EEG signals, we offer a more detailed understanding user affective experiences (valence, arousal, immersion, interest, visual and auditory) and the cognitive mechanisms behind their behavior. We establish benchmarks for rating prediction by the recommendation algorithm, showing significant improvement with the inclusion of EEG signals. Furthermore, we demonstrate the potential of this dataset in gaining insights into the affective experience and cognitive activity behind user behaviors in recommender systems. This work presents a novel perspective for enhancing short video recommendation by leveraging the rich information contained in EEG signals and multidimensional affective engagement scores, paving the way for future research in short video recommendation systems.

Ranking is at the core of many artificial intelligence (AI) applications, including search engines, recommender systems, etc. Modern ranking systems are often constructed with learning-to-rank (LTR) models built from user behavior signals. While previous studies have demonstrated the effectiveness of using user behavior signals (e.g., clicks) as both features and labels of LTR algorithms, we argue that existing LTR algorithms that indiscriminately treat behavior and non-behavior signals in input features could lead to suboptimal performance in practice. Because user behavior signals often have strong correlations with the ranking objective and can only be collected on items that have already been shown to users, directly using behavior signals in LTR could create an exploitation bias that hurts the system performance in the long run.

Conversational search has seen increased recent attention in both the IR and NLP communities. It seeks to clarify and solve users' search needs through multi-turn natural language interactions. However, most existing systems are trained and demonstrated with recorded or artificial conversation logs. Eventually, conversational search systems should be trained, evaluated, and deployed in an open-ended setting with unseen conversation trajectories. A key challenge is that training and evaluating such systems both require a human-in-the-loop, which is expensive and does not scale. One strategy is to simulate users, thereby reducing the scaling costs. However, current user simulators are either limited to only responding to yes-no questions from the conversational search system or unable to produce high-quality responses in general.

Abstract How to collect relevance judgment has long been an important problem in Information Retrieval (IR). A popular method is to collect relevance judgment in a point‐wise manner, in which assessors examine and give an absolute relevance score for each item independently of the others. As an alternative, pair‐wise relevance judgment, also named preference judgment, allows an assessor to compare two items side‐by‐side and express their preference for one over the other. Previous work has explored the differences between these two paradigms of relevance judgments from many different aspects. Most of these works are conducted through explicit/implicit feedback. However, few works investigate the underlying neurological mechanisms of the two paradigms. In this paper, we conduct a lab study to investigate and compare point‐wise and pair‐wise relevance judgment in image search scenarios. We study the neurological mechanisms of the two paradigms through an event‐related potential (ERP) analysis of the users' brain signals while viewing images during a search process. We have obtained several observations, such as search engine users tend to pay more attention to preferred items in the point‐wise paradigm but unpreferred items in the pair‐wise paradigm. Furthermore, we test the adoption of brain signals as implicit feedback for predicting pair‐wise relevance judgment, highlighting the feasibility of leveraging brain signals to understand users' relevance judgments.

Legal case retrieval has received increasing attention in recent years. However, compared to ad hoc retrieval tasks, legal case retrieval has its unique challenges. First, case documents are rather lengthy and contain complex legal structures. Therefore, it is difficult for most existing dense retrieval models to encode an entire document and capture its inherent complex structure information. Most existing methods simply truncate part of the document content to meet the input length limit of PLMs, which will lead to information loss. Additionally, the definition of relevance in the legal domain differs from that in the general domain. Previous semantic-based or lexical-based methods fail to provide a comprehensive understanding of the relevance of legal cases. In this article, we propose a Structured Legal case Retrieval (SLR) framework, which incorporates internal and external structural information to address the above two challenges. Specifically, to avoid the truncation of long legal documents, the internal structural information, which is the organization pattern of legal documents, can be utilized to split a case document into segments. By dividing the document-level semantic matching task into segment-level subtasks, SLR can separately process segments using different methods based on the characteristic of each segment. In this way, the key elements of a case document can be highlighted without losing other content information. Second, toward a better understanding of relevance in the legal domain, we investigate the connections between criminal charges appearing in large-scale case corpus to generate a chargewise relation graph. Then, the similarity between criminal charges can be pre-computed as the external structural information to enhance the recognition of relevant cases. Finally, a learning-to-rank algorithm integrates the features collected from internal and external structures to output the final retrieval results. Experimental results on public legal case retrieval benchmarks demonstrate the superior effectiveness of SLR over existing state-of-the-art baselines, including traditional bag-of-words and neural-based methods. Furthermore, we conduct a case study to visualize how the proposed model focuses on key elements and improves retrieval performance.

Legal case retrieval is a special Information Retrieval (IR) task focusing on legal case documents. Depending on the downstream tasks of the retrieved case documents, users’ information needs in legal case retrieval could be significantly different from those in Web search and traditional ad hoc retrieval tasks. While there are several studies that retrieve legal cases based on text similarity, the underlying search intents of legal retrieval users, as shown in this article, are more complicated than that yet mostly unexplored. To this end, we present a novel hierarchical intent taxonomy of legal case retrieval. It consists of five intent types categorized by three criteria, i.e., search for Particular Case(s) , Characterization , Penalty , Procedure , and Interest . The taxonomy was constructed transparently and evaluated extensively through interviews, editorial user studies, and query log analysis. Through a laboratory user study, we reveal significant differences in user behavior and satisfaction under different search intents in legal case retrieval. Furthermore, we apply the proposed taxonomy to various downstream legal retrieval tasks, e.g., result ranking and satisfaction prediction, and demonstrate its effectiveness. Our work provides important insights into the understanding of user intents in legal case retrieval and potentially leads to better retrieval techniques in the legal domain, such as intent-aware ranking strategies and evaluation methodologies.

State-of-the-art industrial-level recommender system applications mostly adopt complicated model structures such as deep neural networks. While this helps with the model performance, the lack of system explainability caused by these nearly blackbox models also raises concerns and potentially weakens the users’ trust in the system. Existing work on explainable recommendation mostly focuses on designing interpretable model structures to generate model-intrinsic explanations. However, most of them have complex structures, and it is difficult to directly apply these designs onto existing recommendation applications due to the effectiveness and efficiency concerns. However, while there have been some studies on explaining recommendation models without knowing their internal structures (i.e., model-agnostic explanations), these methods have been criticized for not reflecting the actual reasoning process of the recommendation model or, in other words,faithfulness. How to develop model-agnostic explanation methods and evaluate them in terms of faithfulness is mostly unknown. In this work, we propose a reusable evaluation pipeline for model-agnostic explainable recommendation. Our pipeline evaluates the quality of model-agnostic explanation from the perspectives of faithfulness and scrutability. We further propose a model-agnostic explanation framework for recommendation and verify it with the proposed evaluation pipeline. Extensive experiments on public datasets demonstrate that our model-agnostic framework is able to generate explanations that are faithful to the recommendation model. We additionally provide quantitative and qualitative study to show that our explanation framework could enhance the scrutability of blackbox recommendation model. With proper modification, our evaluation pipeline and model-agnostic explanation framework could be easily migrated to existing applications. Through this work, we hope to encourage the community to focus more on faithfulness evaluation of explainable recommender systems.

The Relevance Feedback (RF) process relies on accurate and real-time relevance estimation of feedback documents to improve retrieval performance. Since collecting explicit relevance annotations imposes an extra burden on the user, extensive studies have explored using pseudo-relevance signals and implicit feedback signals as substitutes. However, such signals are indirect indicators of relevance and suffer from complex search scenarios where user interactions are absent or biased. Recently, the advances in portable and high-precision brain-computer interface (BCI) devices have shown the possibility to monitor user’s brain activities during search process. Brain signals can directly reflect user’s psychological responses to search results and thus it can act as additional and unbiased RF signals. To explore the effectiveness of brain signals in the context of RF, we propose a novel RF framework that combines BCI-based RF with pseudo-relevance signals and implicit signals to improve the performance of document re-ranking. The experimental results on the user study dataset show that incorporating brain signals leads to significant performance improvement in our RF framework. Besides, we observe that brain signals perform particularly well in several hard search scenarios, especially when implicit signals as feedback are missing or noisy. This reveals when and how to exploit brain signals in the context of RF.

Ranking systems are the key components of modern Information Retrieval (IR) applications, such as search engines and recommender systems. Besides the ranking relevance to users, the exposure fairness to item providers has also been considered an important factor in ranking optimization. Many fair ranking algorithms have been proposed to jointly optimize both ranking relevance and fairness. However, we find that most existing fair ranking methods adopt greedy algorithms that only optimize rankings for the next immediate session or request. As shown in this paper, such a myopic paradigm could limit the upper bound of ranking optimization and lead to suboptimal performance in the long term.

Passage retrieval is a fundamental task in many information systems, such as web search and question answering, where both efficiency and effectiveness are critical concerns. In recent years, neural retrievers based on pre-trained language models (PLM), such as dual-encoders, have achieved huge success. Yet, studies have found that the performance of dual-encoders are often limited due to the neglecting of the interaction information between queries and candidate passages. Therefore, various interaction paradigms have been proposed to improve the performance of vanilla dual-encoders. Particularly, recent state-of-the-art methods often introduce late-interaction during the model inference process. However, such late-interaction based methods usually bring extensive computation and storage cost on large corpus. Despite their effectiveness, the concern of efficiency and space footprint is still an important factor that limits the application of interaction-based neural retrieval models. To tackle this issue, we Incorporate Implicit Interaction into dual-encoders, and propose I3 retriever. In particular, our implicit interaction paradigm leverages generated pseudo-queries to simulate query-passage interaction, which jointly optimizes with query and passage encoders in an end-to-end manner. It can be fully pre-computed and cached, and its inference process only involves simple dot product operation of the query vector and passage vector, which makes it as efficient as the vanilla dual encoders. We conduct comprehensive experiments on MSMARCO and TREC2019 Deep Learning Datasets, demonstrating the I3 retriever's superiority in terms of both effectiveness and efficiency. Moreover, the proposed implicit interaction is compatible with special pre-training and knowledge distillation for passage retrieval, which brings a new state-of-the-art performance. The codes are available at https://github.com/Deriq-Qian-Dong/III-Retriever.

Legal case retrieval is a special IR task aiming to retrieve supporting cases for a given query case. Existing works have shown that conversational search paradigm can improve users' search experience in legal case retrieval. One of the keys to a practical conversational search system is how to ask high-quality clarifying questions to initiate conversations with users and understand their search intents. Recently, Large Language Models, such as ChatGPT and GPT-4, have shown superior ability in both open-domain QA and conversations with human. Thus it is natural to believe that they could be applied to legal conversational search as well. However, our preliminary study has shown that generating clarifying questions in legal conversational search with SOTA LLMs (e.g., GPT-4) often suffers from several problems such as duplication and low-utility contents. To address these problems, we propose LeClari, which leverages legal event schema as external knowledge to instruct LLMs to generate effective clarifying questions for legal conversational search. LeClari is constructed with a prompt module and a novel legal event selection module. The former defines a prompt with legal events for clarifying question generation and the latter selects potential event types by modeling the relationships of legal event types, conversational context, and candidate cases. We also propose ranking-oriented rewards and employ the reward augmented maximum likelihood (RAML) method to optimize LeClari directly based on the final retrieval performance of the conversational legal search system. Empirical results over two widely adopted legal case retrieval datasets demonstrate the effectiveness of our approach as compared with the state-of-the-art baselines.

Legal case retrieval is a critical process for modern legal information systems.While recent studies have utilized pre-trained language models (PLMs) based on the general domain selfsupervised pre-training paradigm to build models for legal case retrieval, there are limitations in using general domain PLMs as backbones.Specifically, these models may not fully capture the underlying legal features in legal case documents.To address this issue, we propose CaseEncoder, a legal document encoder that leverages fine-grained legal knowledge in both the data sampling and pre-training phases.In the data sampling phase, we enhance the quality of the training data by utilizing fine-grained law article information to guide the selection of positive and negative examples.In the pretraining phase, we design legal-specific pretraining tasks that align with the judging criteria of relevant legal cases.Based on these tasks, we introduce an innovative loss function called Biased Circle Loss to enhance the model's ability to recognize case relevance in fine grains.Experimental results on multiple benchmarks

Ranking is at the core of Information Retrieval. Classic ranking optimization studies often treat ranking as a sorting problem with the assumption that the best performance of ranking would be achieved if we rank items according to their individual utility. Accordingly, considerable ranking metrics have been developed and learning-to-rank algorithms that have been designed to optimize these simple performance metrics have been widely used in modern IR systems. As applications evolve, however, people's need for information retrieval have shifted from simply retrieving relevant documents to more advanced information services that satisfy their complex working and entertainment needs. Thus, more complicated and user-centric objectives such as user satisfaction and engagement have been adopted to evaluate modern IR systems today. Those objectives, unfortunately, are difficult to be optimized under existing learning-to-rank frameworks as they are subject to great variance and complicated structures that cannot be explicitly explained or formulated with math equations like those simple performance metrics. This leads to the following research question -- how to optimize result ranking for complex ranking metrics without knowing their internal structures? To address this question, we conduct formal analysis on the limitation of existing ranking optimization techniques and describe three research tasks in Metric-agnostic Ranking Optimization: (1) develop surrogate metric models to simulate complex online ranking metrics on offline data; (2) develop differentiable ranking optimization frameworks for list or session level performance metrics without fine-grained supervision signals; and (3) develop efficient parameter exploration and exploitation techniques for ranking optimization in metric-agnostic scenarios. Through the discussion of potential solutions to these tasks, we hope to encourage more people to look into the problem of ranking optimization in complex search and recommendation scenarios.

With the increasing popularity of location-based services, the point-of-interest (POI) search has received considerable attention in recent years. Existing studies on POI search mostly focus on how to construct better retrieval models to retrieve the relevant POI based on query-POI matching. However, user behavior in POI search, i.e., how users examine the search engine result page (SERP), is mostly underexplored. A good understanding of user behavior is well-recognized as a key to develop effective user models and retrieval models to improve the search quality. Therefore, in this paper, we propose to investigate user behavior in POI search with a lab study in which users' eye movements and their implicit feedback on the SERP are collected. Based on the collected data, we analyze (1) query-level user behavior patterns in POI search, i.e., examination and interactions on SERP; (2) session-level user behavior patterns in POI search, i.e., query reformulation, termination of search, etc. Our work sheds light on user behavior in POI search and could potentially benefit future studies on related research topics.

Legal case retrieval, which aims to find relevant cases for a query case, plays a core role in the intelligent legal system. Despite the success that pre-training has achieved in ad-hoc retrieval tasks, effective pre-training strategies for legal case retrieval remain to be explored. Compared with general documents, legal case documents are typically long text sequences with intrinsic logical structures. However, most existing language models have difficulty understanding the long-distance dependencies between different structures. Moreover, in contrast to the general retrieval, the relevance in the legal domain is sensitive to key legal elements. Even subtle differences in key legal elements can significantly affect the judgement of relevance. However, existing pre-trained language models designed for general purposes have not been equipped to handle legal elements.

Recent studies have shown that Dense Retrieval (DR) techniques can significantly improve the performance of first-stage retrieval in IR systems. Despite its empirical effectiveness, the application of DR is still limited. In contrast to statistic retrieval models that rely on highly efficient inverted index solutions, DR models build dense embeddings that are difficult to be pre-processed with most existing search indexing systems. To avoid the expensive cost of brute-force search, the Approximate Nearest Neighbor (ANN) algorithm and corresponding indexes are widely applied to speed up the inference process of DR models. Unfortunately, while ANN can improve the efficiency of DR models, it usually comes with a significant price on retrieval performance.

Unbiased Learning to Rank(ULTR), i.e., learning to rank documents with biased user feedback data, is a well-known challenge in information retrieval. Existing methods in unbiased learning to rank typically rely on click modeling or inverse propensity weighting(IPW). Unfortunately, search engines face the issue of a severe long-tail query distribution, which neither click modeling nor IPW handles well. Click modeling usually requires that the same query-document pair appears multiple times for reliable inference, which makes it fall short for tail queries; IPW suffers from high variance since it is highly sensitive to small propensity score values. Therefore, a general debiasing framework that works well under tail queries is sorely needed. To address this problem, we propose a model-based unbiased learning-to-rank framework. Specifically, we develop a general context-aware user simulator to generate pseudo clicks for unobserved ranked lists to train rankers, which addresses the data sparsity problem. In addition, considering the discrepancy between pseudo clicks and actual clicks, we take the observation of a ranked list as the treatment variable and further incorporate inverse propensity weighting with pseudo labels in a doubly robust way. The derived bias and variance indicate that the proposed model-based method is more robust than existing methods. Extensive experiments on benchmark datasets, including simulated datasets and real click logs, demonstrate that the proposed model-based method consistently outperforms state-of-the-art methods in various scenarios. The code is available at https://github.com/rowedenny/MULTR.

Ranking systems are ubiquitous in modern Internet services, including online marketplaces, social media, and search engines. Traditionally, ranking systems only focus on how to get better relevance estimation. When relevance estimation is available, they usually adopt a user-centric optimization strategy where ranked lists are generated by sorting items according to their estimated relevance. However, such user-centric optimization ignores the fact that item providers also draw utility from ranking systems. It has been shown in existing research that such user-centric optimization will cause much unfairness to item providers, followed by unfair opportunities and unfair economic gains for item providers.

A long-standing challenge for search and conversational assistants is query intention detection in ambiguous queries. Asking clarifying questions in conversational search has been widely studied and considered an effective solution to resolve query ambiguity. Existing work have explored various approaches for clarifying question ranking and generation. However, due to the lack of real conversational search data, they have to use artificial datasets for training, which limits their generalizability to real-world search scenarios. As a result, the industry has shown reluctance to implement them in reality, further suspending the availability of real conversational search interaction data. The above dilemma can be formulated as a cold start problem of clarifying question generation and conversational search in general. Furthermore, even if we do have large-scale conversational logs, it is not realistic to gather training data that can comprehensively cover all possible queries and topics in open-domain search scenarios. The risk of fitting bias when training a clarifying question retrieval/generation model on incomprehensive dataset is thus another important challenge.

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.

The truncation of ranking lists predicted by retrieval models is vital to ensure users' search experience. Particularly, in specific vertical domains where documents are usually complicated and extensive (e.g., legal cases), the cost of browsing results is much higher than traditional IR tasks (e.g., Web search) and setting a reasonable cut-off position is quite necessary. While it is straightforward to apply existing result list truncation approaches to legal case retrieval, the effectiveness of these methods is limited because they only focus on simple document statistics and usually fail to capture the context information of documents in the ranking list. These existing efforts also treat result list truncation as an isolated task instead of a component in the entire ranking process, limiting the usage of truncation in practical systems. To tackle these limitations, we propose LeCut, a ranking list truncation model for legal case retrieval. LeCut utilizes contextual features of the retrieval task to capture the semantic-level similarity between documents and decides the best cut-off position with attention mechanisms. We further propose a Joint Optimization of Truncation and Reranking (JOTR) framework based on LeCut to improve the performance of truncation and retrieval tasks simultaneously. Comparison against competitive baselines on public benchmark datasets demonstrates the effectiveness of LeCut and JOTR. A case study is conducted to visualize the cut-off positions of LeCut and the process of how JOTR improves both retrieval and truncation tasks.

Using implicit feedback collected from user clicks as training labels for learning-to-rank algorithms is a well-developed paradigm that has been extensively studied and used in modern IR systems. Using user clicks as ranking features, on the other hand, has not been fully explored in existing literature. Despite its potential in improving short-term system performance, whether the incorporation of user clicks as ranking features is beneficial for learning-to-rank systems in the long term is still questionable. Two of the most important problems are (1) the explicit bias introduced by noisy user behavior, and (2) the implicit bias, which we refer to as the exploitation bias, introduced by the dynamic training and serving of learning-to-rank systems with behavior features. In this paper, we explore the possibility of incorporating user clicks as both training labels and ranking features for learning to rank. We formally investigate the problems in feature collection and model training, and propose a counterfactual feature projection function and a novel uncertainty-aware learning to rank framework. Experiments on public datasets show that ranking models learned with the proposed framework can significantly outperform models built with raw click features and algorithms that rank items without considering model uncertainty.

In conversational search, agents can interact with users by asking clarifying questions to increase their chance of finding better results. Many recent works and shared tasks in both natural language processing and information retrieval communities have focused on identifying the need to ask clarifying questions and methodologies of generating them. These works assume that asking a clarifying question is a safe alternative to retrieving results. As existing conversational search models are far from perfect, it is possible and common that they could retrieve/generate bad clarifying questions. Asking too many clarifying questions can also drain a user’s patience when the user prefers searching efficiency over correctness. Hence, these models can backfire and harm a user’s search experience due to these risks from asking clarifying questions. In this work, we propose a simulation framework to simulate the risk of asking questions in conversational search and further revise a risk-aware conversational search model to control the risk. We show the model’s robustness and effectiveness through extensive experiments on three conversational datasets — MSDialog, Ubuntu Dialog Corpus, and Opendialkg — in which we compare it with multiple baselines. We show that the risk-control module can work with two different re-ranker models and outperform all of the baselines in most of our experiments.

Product retrieval systems have served as the main entry for customers to discover and purchase products online. With increasing concerns on the transparency and accountability of AI systems, studies on explainable information retrieval has received more and more attention in the research community. Interestingly, in the domain of e-commerce, despite the extensive studies on explainable product recommendation, the studies of explainable product search is still in an early stage. In this paper, we study how to construct effective explainable product search by comparing model-agnostic explanation paradigms with model-intrinsic paradigms and analyzing the important factors that determine the performance of product search explanations. We propose an explainable product search model with model-intrinsic interpretability and conduct crowdsourcing to compare it with the state-of-the-art explainable product search model with model-agnostic interpretability. We observe that both paradigms have their own advantages and the effectiveness of search explanations on different properties are affected by different factors. For example, explanation fidelity is more important for user's overall satisfaction on the system while explanation novelty may be more useful in attracting user purchases. These findings could have important implications for the future studies and design of explainable product search engines.

Learning to rank system has become an important aspect of our daily life. However, the implicit user feedback that is used to train many learning to rank models is usually noisy and suffers from user bias (i.e., position bias). Thus, obtaining unbiased model using biased feedback has become an important research field for IR. Existing studies on unbiased learning to rank (ULTR) can be generalized into two families-algorithms that attain unbiasness with logged data, offline learning, and algorithms that achieve unbiasness by estimating unbiased parameters with real-time user interactions, namely online learning. While there exist many algorithms from both families, there lacks a unified way to compare and benchmark them. As a result, it can be challenging for researchers to choose the right technique for their problems or for people who are new to the field to learn and understand existing algorithms. To solve this problem, we introduced ULTRA, which is a flexible, extensible, and easily configure ULTR toolbox. Its key features include support for multiple ULTR algorithms with configurable hyper parameters, a variety of built-in click models that can be used separately to simulate clicks, different ranking model architectures and evaluation metrics, and simple learning to rank pipeline creation. In this paper, we discuss the general framework of ULTR, briefly describe the algorithms in ULTRA, detail the structure, and pipeline of the toolbox. We experimented on all the algorithms supported by ULTRA and showed that the toolbox performance is reasonable. Our toolbox is an important resource for researchers to conduct experiments on ULTR algorithms with different configurations as well as testing their own algorithms with the supported features.

Recently, we have witnessed that deep learning-based approaches has been widely applied to empower many internet-scale applications. However, the data in these internet-scale applications are high dimensional and extremely sparse, which makes it different from those applications with dense data processing, such as image classification and speech recognition, where deep learning-based approaches have been extensively studied. One of the main applications is the user-centric platform that consists of great deal of users, items and user generated tabular data which are quite high-dimensional. The characteristics of such data pose unique challenges to the adoption of deep learning in these applications, including modeling, training, and online serving, etc. More and more communities from both academia and industry have initiated the endeavors to solve these challenges. This workshop will provide a venue for both the research and engineering communities to discuss and formulate the challenges, utilize opportunities, and propose new ideas in the practice of deep learning on high-dimensional sparse data.

Product search has been a crucial entry point to serve people shopping online. Most existing personalized product models follow the paradigm of representing and matching user intents and items in the semantic space, where finer-grained matching is totally discarded and the ranking of an item cannot be explained further than just user/item level similarity. In addition, while some models in existing studies have created dynamic user representations based on search context, their representations for items are static across all search sessions. This makes every piece of information about the item always equally important in representing the item during matching with various user intents. Aware of the above limitations, we propose a review-based transformer model (RTM) for personalized product search, which encodes the sequence of query, user reviews, and item reviews with a transformer architecture. RTM conducts review-level matching between the user and item, where each review has a dynamic effect according to the context in the sequence. This makes it possible to identify useful reviews to explain the scoring. Experimental results show that RTM significantly outperforms state-of-the-art personalized product search baselines.

Probability Ranking Principle (PRP)[31], which assumes that each document has a unique and independent probability to satisfy a particular information need, is one of the fundamental principles for ranking. Traditionally, heuristic ranking features and well-known learning-to-rank approaches have been designed by following the PRP principle. Recently, neural IR models, which adopt deep learning to enhance the ranking performances, also obey the PRP principle. Though it has been widely used for nearly five decades, in-depth analysis shows that PRP is not an optimal principle for ranking, due to its independent assumption that each document should be independent of the rest candidates. Counter examples include pseudo relevance feedback[24], interactive information retrieval[46], search result diversification[10] etc. To solve the problem, researchers recently proposed to model the dependencies among the documents during the designing of ranking models. A number of ranking models have been proposed and state-of-the-art ranking performances have been achieved. This tutorial aims to give a comprehensive survey on these recently developed ranking models that go beyond the PRP principle. The tutorial tries to categorize these models based on their intrinsic assumptions: assuming that the documents are independent, sequentially dependent, or globally dependent. In this way, we expect the researchers focusing on ranking in search and recommendation can have a novel angle of view on the designing of ranking models, and therefore can stimulate new ideas on developing novel ranking models.

Probability Ranking Principle (PRP), which assumes that each document has a unique and independent probability to satisfy a particular information need, is one of the fundamental principles for ranking. Traditionally, heuristic ranking features and well-known learning-to-rank approaches have been designed by following the PRP principle. Recently, neural IR models, which adopt deep learning to enhance the ranking performances, also obey the PRP principle. Though it has been widely used for nearly five decades, in-depth analysis shows that PRP is not an optimal principle for ranking, due to its independent assumption that each document should be independent of the rest candidates. Counter examples include pseudo relevance feedback, interactive information retrieval, search result diversification, etc. To solve the problem, researchers recently proposed to model the dependencies among the documents during the designing of ranking models. A number of ranking models have been proposed and state-of-the-art ranking performances have been achieved. This tutorial aims to give a comprehensive survey on these recently developed ranking models that go beyond the PRP principle. The tutorial tries to categorize these models based on their intrinsic assumptions: assuming that the documents are independent, sequentially dependent, or globally dependent. In this way, we expect the researchers focusing on ranking in search and recommendation can have a novel angle of view on the designing of ranking models, and therefore can stimulate new ideas on developing novel ranking models. The material of this tutorial can be found in https://github.com/pl8787/wsdm2021-beyond-prp-tutorial.

Users often formulate their search queries and questions with immature language without well-developed keywords and complete structures. Such queries are likely to fail to express their true information needs and raise ambiguity as fragmental language often yield various interpretations and aspects. This gives search engines a hard time processing and understanding the query, and eventually leads to unsatisfactory retrieval results. An alternative approach to direct answer while facing an ambiguous query is to proactively ask clarifying questions to the user. Recent years have seen many works and shared tasks from both NLP and IR community about identifying the need for asking clarifying question and methodology to generate them. An often neglected fact by these works is that although sometimes the need for clarifying questions is correctly recognized, the clarifying questions these system generate are still off-topic and dissatisfaction provoking to users and may just cause users to leave the conversation.

Rankings, especially those in search and recommendation systems, often determine how people access information and how information is exposed to people. Therefore, how to balance the relevance and fairness of information exposure is considered as one of the key problems for modern IR systems. As conventional ranking frameworks that myopically sorts documents with their relevance will inevitably introduce unfair result exposure, recent studies on ranking fairness mostly focus on dynamic ranking paradigms where result rankings can be adapted in real-time to support fairness in groups (i.e., races, genders, etc.). Existing studies on fairness in dynamic learning to rank, however, often achieve the overall fairness of document exposure in ranked lists by significantly sacrificing the performance of result relevance and fairness on the top results. To address this problem, we propose a fair and unbiased ranking method named Maximal Marginal Fairness (MMF). The algorithm integrates unbiased estimators for both relevance and merit-based fairness while providing an explicit controller that balances the selection of documents to maximize the marginal relevance and fairness in top-k results. Theoretical and empirical analysis shows that, with small compromises on long list fairness, our method achieves superior efficiency and effectiveness comparing to the state-of-the-art algorithms in both relevance and fairness for top-k rankings.

How to obtain an unbiased ranking model by learning to rank with biased user feedback is an important research question for IR. Existing work on unbiased learning to rank (ULTR) can be broadly categorized into two groups—the studies on unbiased learning algorithms with logged data, namely, the offline unbiased learning, and the studies on unbiased parameters estimation with real-time user interactions, namely, the online learning to rank. While their definitions of unbiasness are different, these two types of ULTR algorithms share the same goal—to find the best models that rank documents based on their intrinsic relevance or utility. However, most studies on offline and online unbiased learning to rank are carried in parallel without detailed comparisons on their background theories and empirical performance. In this article, we formalize the task of unbiased learning to rank and show that existing algorithms for offline unbiased learning and online learning to rank are just the two sides of the same coin. We evaluate eight state-of-the-art ULTR algorithms and find that many of them can be used in both offline settings and online environments with or without minor modifications. Further, we analyze how different offline and online learning paradigms would affect the theoretical foundation and empirical effectiveness of each algorithm on both synthetic and real search data. Our findings provide important insights and guidelines for choosing and deploying ULTR algorithms in practice.

Different from shopping at retail stores, consumers on e-commerce platforms usually cannot touch or try products before purchasing, which means that they have to make decisions when they are uncertain about the outcome (e.g., satisfaction level) of purchasing a product. To study people's preferences with regard to choices that have uncertain outcomes, economics researchers have proposed the hypothesis of Expected Utility (EU) that models the subject value associated with an individual's choice as the statistical expectations of that individual's valuations of the outcomes of this choice. Despite its success in studies of game theory and decision theory, the effectiveness of EU, however, is mostly unknown in e-commerce recommendation systems. Previous research on e-commerce recommendation interprets the utility of purchase decisions either as a function of the consumed quantity of the product or as the gain of sellers/buyers in the monetary sense. As most consumers just purchase one unit of a product at a time and most alternatives have similar prices, such modeling of purchase utility is likely to be inaccurate in practice. In this paper, we interpret purchase utility as the satisfaction level a consumer gets from a product and propose a recommendation framework using EU to model consumers' behavioral patterns. We assume that consumer estimates the expected utilities of all the alternatives and choose products with maximum expected utility for each purchase. To deal with the potential psychological biases of each consumer, we introduce the usage of Probability Weight Function (PWF) and design our algorithm based on Weighted Expected Utility (WEU). Empirical study on real-world e-commerce datasets shows that our proposed ranking-based recommendation framework achieves statistically significant improvement against both classical Collaborative Filtering/Latent Factor Models and state-of-the-art deep models in top-K recommendation.

Leveraging biased click data for optimizing learning to rank systems has been a popular approach in information retrieval. Because click data is often noisy and biased, a variety of methods have been proposed to construct unbiased learning to rank (ULTR) algorithms for the learning of unbiased ranking models. Among them, automatic unbiased learning to rank (AutoULTR) algorithms that jointly learn user bias models (i.e., propensity models) with unbiased rankers have received a lot of attention due to their superior performance and low deployment cost in practice. Despite their differences in theories and algorithm design, existing studies on ULTR usually use uni-variate ranking functions to score each document or result independently. On the other hand, recent advances in context-aware learning-to-rank models have shown that multivariate scoring functions, which read multiple documents together and predict their ranking scores jointly, are more powerful than uni-variate ranking functions in ranking tasks with human-annotated relevance labels. Whether such superior performance would hold in ULTR with noisy data, however, is mostly unknown. In this paper, we investigate existing multivariate scoring functions and AutoULTR algorithms in theory and prove that permutation invariance is a crucial factor that determines whether a context-aware learning-to-rank model could be applied to existing AutoULTR framework. Our experiments with synthetic clicks on two large-scale benchmark datasets show that AutoULTR models with permutation-invariant multivariate scoring functions significantly outperform those with uni-variate scoring functions and permutation-variant multivariate scoring functions.

Product search is an important way for people to browse and purchase items on E-commerce platforms. While customers tend to make choices based on their personal tastes and preferences, analysis of commercial product search logs has shown that personalization does not always improve product search quality. Most existing product search techniques, however, conduct undifferentiated personalization across search sessions. They either use a fixed coefficient to control the influence of personalization or let personalization take effect all the time with an attention mechanism. The only notable exception is the recently proposed zero-attention model (ZAM) that can adaptively adjust the effect of personalization by allowing the query to attend to a zero vector. Nonetheless, in ZAM, personalization can act at most as equally important as the query and the representations of items are static across the collection regardless of the items co-occurring in the user's historical purchases. Aware of these limitations, we propose a transformer-based embedding model (TEM) for personalized product search, which could dynamically control the influence of personalization by encoding the sequence of query and user's purchase history with a transformer architecture. Personalization could have a dominant impact when necessary and interactions between items can be taken into consideration when computing attention weights. Experimental results show that TEM outperforms state-of-the-art personalization product retrieval models significantly.

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.

Product search is one of the most popular methods for customers to discover products online. Most existing studies on product search focus on developing effective retrieval models that rank items by their likelihood to be purchased. However, they ignore the problem that there is a gap between how systems and customers perceive the relevance of items. Without explanations, users may not understand why product search engines retrieve certain items for them, which consequentially leads to imperfect user experience and suboptimal system performance in practice. In this work, we tackle this problem by constructing explainable retrieval models for product search. Specifically, we propose to model the “search and purchase” behavior as a dynamic relation between users and items, and create a dynamic knowledge graph based on both the multi-relational product data and the context of the search session. Ranking is conducted based on the relationship between users and items in the latent space, and explanations are generated with logic inferences and entity soft matching on the knowledge graph. Empirical experiments show that our model, which we refer to as the Dynamic Relation Embedding Model (DREM), significantly outperforms the state-of-the-art baselines and has the ability to produce reasonable explanations for search results.

Product search is one of the most popular methods for people to discover and purchase products on e-commerce websites. Because personal preferences often have an important influence on the purchase decision of each customer, it is intuitive that personalization should be beneficial for product search engines. While synthetic experiments from previous studies show that purchase histories are useful for identifying the individual intent of each product search session, the effect of personalization on product search in practice, however, remains mostly unknown. In this paper, we formulate the problem of personalized product search and conduct large-scale experiments with search logs sampled from a commercial e-commerce search engine. Results from our preliminary analysis show that the potential of personalization depends on query characteristics, interactions between queries, and user purchase histories. Based on these observations, we propose a Zero Attention Model for product search that automatically determines when and how to personalize a user-query pair via a novel attention mechanism. Empirical results on commercial product search logs show that the proposed model not only significantly outperforms state-of-the-art personalized product retrieval models, but also provides important information on the potential of personalization in each product search session.

Intelligent assistants change the way people interact with computers and make it possible for people to search for products through conversations when they have purchase needs. During the interactions, the system could ask questions on certain aspects of the ideal products to clarify the users' needs. For example, previous work proposed to ask users the exact characteristics of their ideal items before showing results. However, users may not have clear ideas about what an ideal item looks like, especially when they have not seen any item. So it is more feasible to facilitate the conversational search by showing example items and asking for feedback instead. In addition, when the users provide negative feedback for the presented items, it is easier to collect their detailed feedback on certain properties (aspect-value pairs) of the non-relevant items. By breaking down the item-level negative feedback to fine-grained feedback on aspect-value pairs, more information is available to help clarify users' intents. So in this paper, we propose a conversational paradigm for product search driven by non-relevant items, based on which fine-grained feedback is collected and utilized to show better results in the next iteration. We then propose an aspect-value likelihood model to incorporate both positive and negative feedback on fine-grained aspect-value pairs of the non-relevant items. Experimental results show that our model is significantly better than state-of-the-art product search baselines without using feedback and those baselines using item-level negative feedback.

Users are known to interact more with fresh content in certain temporally associated domains such as news search or job seeking, leading to an uneven distribution of interactions over items of different degrees of freshness. Data collected under such an "aging effect'' is usually used unconditionally on all sort of recommendation tasks, and as a result more recently published content may be over-represented during model training and evaluation. In this study, we characterize this temporal influence as a recency bias, and present an analysis in the domain of job recommendation. We show that, by correcting for recency bias using an unbiased learning to rank approach, one can improve the quality of recommendation significantly over a recent neural collaborative filtering model on RecSys Challenge 2017 data.

In web image search, items users search for are images instead of Web pages or online services. Web image search constitutes a very important part of web search. Re-ranking is a trusted technique to improve retrieval effectiveness in web search. Previous work on re-ranking web image search results mainly focuses on intra-query information (e.g., human interactions with the initial list of the current query). Contextual information such as the query sequence and implicit user feedback provided during a search session prior to the current query is known to improve the performance of general web search but has so far not been used in web image search. The differences in result placement and interaction mechanisms of image search make the search process rather different from general Web search engines. Because of these differences, context-aware re-ranking models that have originally been developed for general web search cannot simply be applied to web image search. We propose CARM, a context-aware re-ranking model, a neural network-based framework to re-rank web image search results for a query based on previous interaction behavior in the search session in which the query was submitted. Specifically, we explore a hybrid encoder with an attention mechanism to model intra-query and inter-query user preferences for image results in a two-stage structure. We train context-aware re-ranking model (CARM) to jointly learn query and image representations so as to be able to deal with the multimodal characteristics of web image search. Extensive experiments are carried out on a commercial web image search dataset. The results show that CARM outperforms state-of-the-art baseline models in terms of personalized evaluation metrics. Also, CARM combines the original ranking can improve the original ranking on personalized ranking and relevance estimation. We make the implementation of CARM and relevant datasets publicly available to facilitate future studies.

Explainable recommendation and search attempt to develop models or methods that not only generate high-quality recommendation or search results, but also intuitive explanations of the results for users or system designers, which can help to improve the system transparency, persuasiveness, trustworthiness, and effectiveness, etc. This is even more important in personalized search and recommendation scenarios, where users would like to know why a particular product, web page, news report, or friend suggestion exists in his or her own search and recommendation lists. The tutorial focuses on the research and application of explainable recommendation and search algorithms, as well as their application in real-world systems such as search engine, e-commerce and social networks. The tutorial aims at introducing and communicating explainable recommendation and search methods to the community, as well as gathering researchers and practitioners interested in this research direction for discussions, idea communications, and research promotions.

Matching between two information objects is the core of many different information retrieval (IR) applications including Web search, question answering, and recommendation. Recently, deep learning methods have yielded immense success in speech recognition, computer vision, and natural language processing, significantly advancing state-of-the-art of these areas. In the IR community, deep learning has also attracted much attention, and researchers have proposed a large number of deep matching models to tackle the matching problem for different IR applications. Despite the fact that deep matching models have gained significant progress in these areas, there are still many challenges to be addressed when applying these models to real IR scenarios. In this workshop, we focus on the applicability of deep matching models to practical applications. We aim to discuss the issues of applying deep matching models to production systems, as well as to shed some light on the fundamental characteristics of different matching tasks in IR. website : https://wsdm2019-dapa.github.io/index.html

Implicit feedback (e.g., user clicks) is an important source of data for modern search engines. While heavily biased [8, 9, 11, 27], it is cheap to collect and particularly useful for user-centric retrieval applications such as search ranking. To develop an unbiased learning-to-rank system with biased feedback, previous studies have focused on constructing probabilistic graphical models (e.g., click models) with user behavior hypothesis to extract and train ranking systems with unbiased relevance signals. Recently, a novel counterfactual learning framework that estimates and adopts examination propensity for unbiased learning to rank has attracted much attention. Despite its popularity, there is no systematic comparison of the unbiased learning-to-rank frameworks based on counterfactual learning and graphical models. In this tutorial, we aim to provide an overview of the fundamental mechanism for unbiased learning to rank. We will describe the theory behind existing frameworks, and give detailed instructions on how to conduct unbiased learning to rank in practice.

Conversational search and recommendation based on user-system dialogs exhibit major differences from conventional search and recommendation tasks in that 1) the user and system can interact for multiple semantically coherent rounds on a task through natural language dialog, and 2) it becomes possible for the system to understand the user needs or to help users clarify their needs by asking appropriate questions from the users directly. We believe the ability to ask questions so as to actively clarify the user needs is one of the most important advantages of conversational search and recommendation. In this paper, we propose and evaluate a unified conversational search/recommendation framework, in an attempt to make the research problem doable under a standard formalization. Specifically, we propose a System Ask -- User Respond (SAUR) paradigm for conversational search, define the major components of the paradigm, and design a unified implementation of the framework for product search and recommendation in e-commerce. To accomplish this, we propose the Multi-Memory Network (MMN) architecture, which can be trained based on large-scale collections of user reviews in e-commerce. The system is capable of asking aspect-based questions in the right order so as to understand the user needs, while (personalized) search is conducted during the conversation, and results are provided when the system feels confident. Experiments on real-world user purchasing data verified the advantages of conversational search and recommendation against conventional search and recommendation algorithms in terms of standard evaluation measures such as NDCG.

Learning to rank has been intensively studied and widely applied in information retrieval. Typically, a global ranking function is learned from a set of labeled data, which can achieve good performance on average but may be suboptimal for individual queries by ignoring the fact that relevant documents for different queries may have different distributions in the feature space. Inspired by the idea of pseudo relevance feedback where top ranked documents, which we refer as the local ranking context, can provide important information about the query's characteristics, we propose to use the inherent feature distributions of the top results to learn a Deep Listwise Context Model that helps us fine tune the initial ranked list. Specifically, we employ a recurrent neural network to sequentially encode the top results using their feature vectors, learn a local context model and use it to re-rank the top results. There are three merits with our model: (1) Our model can capture the local ranking context based on the complex interactions between top results using a deep neural network; (2) Our model can be built upon existing learning-to-rank methods by directly using their extracted feature vectors; (3) Our model is trained with an attention-based loss function, which is more effective and efficient than many existing listwise methods. Experimental results show that the proposed model can significantly improve the state-of-the-art learning to rank methods on benchmark retrieval corpora.

Learning to rank with biased click data is a well-known challenge. A variety of methods has been explored to debias click data for learning to rank such as click models, result interleaving and, more recently, the unbiased learning-to-rank framework based on inverse propensity weighting. Despite their differences, most existing studies separate the estimation of click bias (namely the propensity model ) from the learning of ranking algorithms. To estimate click propensities, they either conduct online result randomization, which can negatively affect the user experience, or offline parameter estimation, which has special requirements for click data and is optimized for objectives (e.g. click likelihood) that are not directly related to the ranking performance of the system. In this work, we address those problems by unifying the learning of propensity models and ranking models. We find that the problem of estimating a propensity model from click data is a dual problem of unbiased learning to rank. Based on this observation, we propose a Dual Learning Algorithm (DLA) that jointly learns an unbiased ranker and an unbiased propensity model. DLA is an automatic unbiased learning-to-rank framework as it directly learns unbiased ranking models from biased click data without any preprocessing. It can adapt to the change of bias distributions and is applicable to online learning. Our empirical experiments with synthetic and real-world data show that the models trained with DLA significantly outperformed the unbiased learning-to-rank algorithms based on result randomization and the models trained with relevance signals extracted by click models.

Email continues to be an important form of communication as well as a way to manage tasks and archive personal information. As the volume of email grows, organizing and finding relevant email remains challenging. In this paper, we present a large-scale log analysis of the activities that people perform on email mes-sages (accessing external information via links or attachments, responding to messages, and organizing messages), their search behavior, and their organizational practices in a popular web email client.

The Web has accumulated a rich source of information, such as text, image, rating, etc, which represent different aspects of user preferences. However, the heterogeneous nature of this information makes it difficult for recommender systems to leverage in a unified framework to boost the performance. Recently, the rapid development of representation learning techniques provides an approach to this problem. By translating the various information sources into a unified representation space, it becomes possible to integrate heterogeneous information for informed recommendation.

Product search is an important part of online shopping. In contrast to many search tasks, the objectives of product search are not confined to retrieving relevant products. Instead, it focuses on finding items that satisfy the needs of individuals and lead to a user purchase. The unique characteristics of product search make search personalization essential for both customers and e-shopping companies. Purchase behavior is highly personal in online shopping and users often provide rich feedback about their decisions (e.g. product reviews). However, the severe mismatch found in the language of queries, products and users make traditional retrieval models based on bag-of-words assumptions less suitable for personalization in product search. In this paper, we propose a hierarchical embedding model to learn semantic representations for entities (i.e. words, products, users and queries) from different levels with their associated language data. Our contributions are three-fold: (1) our work is one of the initial studies on personalized product search; (2) our hierarchical embedding model is the first latent space model that jointly learns distributed representations for queries, products and users with a deep neural network; (3) each component of our network is designed as a generative model so that the whole structure is explainable and extendable. Following the methodology of previous studies, we constructed personalized product search benchmarks with Amazon product data. Experiments show that our hierarchical embedding model significantly outperforms existing product search baselines on multiple benchmark datasets.

Key frames are playing a very important role for many video applications, such as on-line movie preview and video information retrieval. Although a number of key frame selection methods have been proposed in the past, existing technologies mainly focus on how to precisely summarize the video content, but seldom take the user preferences into consideration. However, in real scenarios, people may cast diverse interests on the contents even for the same video, and thus they may be attracted by quite different key frames, which makes the selection of key frames an inherently personalized process. In this paper, we propose and investigate the problem of personalized key frame recommendation to bridge the above gap. To do so, we make use of video images and user time-synchronized comments to design a novel key frame recommender that can simultaneously model visual and textual features in a unified framework. By user personalization based on her/his previously reviewed frames and posted comments, we are able to encode different user interests in a unified multi-modal space, and can thus select key frames in a personalized manner, which, to the best of our knowledge, is the first time in the research field of video content analysis. Experimental results show that our method performs better than its competitors on various measures.

As the number of email users and messages continues to grow, search is becoming more important for finding information in personal archives. In spite of its importance, email search is much less studied than web search, particularly using large-scale behavioral log analysis. In this paper we report the results of a large-scale log analysis of email search and complement this with a survey to better understand email search intent and success. We characterize email search behaviors and highlight differences from web search. When searching for email, people know many attributes about what they are looking for; they often look for specific known items; their queries are shorter and they click on fewer items than in web search. Although repeat queries are common in both email and web search, repeat visits to the same search result are much less common in email search suggesting that the same query is used for different search intents over time. We consider search intent from multiple angles. In email search logs, we find that people use email search not just to find information but also to perform tasks such as cleanup or organization, and that the distribution of actions they perform depends on the type of query. In our survey, people reported that they looked for specific information in both email search and web search, but they were much less likely to search for general information on a topic in email. The differences in overall behavior, re-finding patterns and search intents we observed between email and web search have important implications for the design of email search algorithms and interfaces.

In recent years, deep neural networks have led to exciting breakthroughs in speech recognition, computer vision, and natural language processing (NLP) tasks. However, there have been few positive results of deep models on ad-hoc retrieval tasks. This is partially due to the fact that many important characteristics of the ad-hoc retrieval task have not been well addressed in deep models yet. Typically, the ad-hoc retrieval task is formalized as a matching problem between two pieces of text in existing work using deep models, and treated equivalent to many NLP tasks such as paraphrase identification, question answering and automatic conversation. However, we argue that the ad-hoc retrieval task is mainly about relevance matching while most NLP matching tasks concern semantic matching, and there are some fundamental differences between these two matching tasks. Successful relevance matching requires proper handling of the exact matching signals, query term importance, and diverse matching requirements. In this paper, we propose a novel deep relevance matching model (DRMM) for ad-hoc retrieval. Specifically, our model employs a joint deep architecture at the query term level for relevance matching. By using matching histogram mapping, a feed forward matching network, and a term gating network, we can effectively deal with the three relevance matching factors mentioned above. Experimental results on two representative benchmark collections show that our model can significantly outperform some well-known retrieval models as well as state-of-the-art deep matching models.

A common limitation of many information retrieval (IR) models is that relevance scores are solely based on exact (i.e., syntactic) matching of words in queries and documents under the simple Bag-of-Words (BoW) representation. This not only leads to the well-known vocabulary mismatch problem, but also does not allow semantically related words to contribute to the relevance score. Recent advances in word embedding have shown that semantic representations for words can be efficiently learned by distributional models. A natural generalization is then to represent both queries and documents as Bag-of-Word-Embeddings (BoWE), which provides a better foundation for semantic matching than BoW. Based on this representation, we introduce a novel retrieval model by viewing the matching between queries and documents as a non-linear word transportation (NWT) problem. With this formulation, we define the capacity and profit of a transportation model designed for the IR task. We show that this transportation problem can be efficiently solved via pruning and indexing strategies. Experimental results on several representative benchmark datasets show that our model can outperform many state-of-the-art retrieval models as well as recently introduced word embedding-based models. We also conducted extensive experiments to analyze the effect of different settings on our semantic matching model.

As an alternative to question answering methods based on feature engineering, deep learning approaches such as convolutional neural networks (CNNs) and Long Short-Term Memory Models (LSTMs) have recently been proposed for semantic matching of questions and answers. To achieve good results, however, these models have been combined with additional features such as word overlap or BM25 scores. Without this combination, these models perform significantly worse than methods based on linguistic feature engineering. In this paper, we propose an attention based neural matching model for ranking short answer text. We adopt value-shared weighting scheme instead of position-shared weighting scheme for combining different matching signals and incorporate question term importance learning using question attention network. Using the popular benchmark TREC QA data, we show that the relatively simple aNMM model can significantly outperform other neural network models that have been used for the question answering task, and is competitive with models that are combined with additional features. When aNMM is combined with additional features, it outperforms all baselines.

Incorporating topic level estimation into language models has been shown to be beneficial for information retrieval (IR) models such as cluster-based retrieval and LDA-based document representation. Neural embedding models, such as paragraph vector (PV) models, on the other hand have shown their effectiveness and efficiency in learning semantic representations of documents and words in multiple Natural Language Processing (NLP) tasks. However, their effectiveness in information retrieval is mostly unknown. In this paper, we study how to effectively use the PV model to improve ad-hoc retrieval. We propose three major improvements over the original PV model to adapt it for the IR scenario: (1) we use a document frequency-based rather than the corpus frequency-based negative sampling strategy so that the importance of frequent words will not be suppressed excessively; (2) we introduce regularization over the document representation to prevent the model overfitting short documents along with the learning iterations; and (3) we employ a joint learning objective which considers both the document-word and word-context associations to produce better word probability estimation. By incorporating this enhanced PV model into the language modeling framework, we show that it can significantly outperform the state-of-the-art topic enhanced language models.

Developing effective methods for fusing multiple ranked lists of documents is crucial to many applications. Federated web search, for instance, has become a common practice where a query is issued to different verticals and a single ranked list of blended results is created. While federated search is regarded as collection fusion, data fusion techniques aim at improving search coverage and precision by combining multiple search runs on a single document collection. In this paper, we study in depth and extend a neural network-based approach, LambdaMerge, for merging results of ranked lists drawn from one (i.e., data fusion) or more (i.e., collection fusion) verticals. The proposed model considers the impact of the quality of documents, ranked lists and verticals for producing the final merged result in an optimization framework. We further investigate the potential of incorporating deep structures into the model with an aim of determining better combinations of different evidence. In the experiments on collection fusion and data fusion, the proposed approach significantly outperforms several standard baselines and state-of-the-art learning-based approaches.