In this paper, we investigate code-integrated reasoning (CIR), where models generate code when necessary and integrate feedback by executing it through a code interpreter. To acquire this capability, models must learn when and how to use external code tools effectively, which is supported by tool-augmented reinforcement learning (RL). Despite its benefits, tool-augmented RL can still suffer from potential instability in the learning dynamics. In light of this challenge, we present a systematic approach ETIR (Effective TIR) to improving the training effectiveness and stability of tool-augmented RL for code-integrated reasoning. Specifically, we develop enhanced training strategies that balance exploration and stability, progressively building tool-use capabilities while improving reasoning performance. Through extensive experiments on five mainstream mathematical reasoning benchmarks, our model demonstrates significant performance improvements over multiple competitive baselines. Furthermore, we conduct an in-depth analysis of the mechanism of code-integrated reasoning, revealing several key insights, such as the extension of model’s capability boundaries and the simultaneous improvement of reasoning efficiency through code integration. These findings underscore the potential of code-integrated reasoning as a scalable paradigm for advancing robust and efficient language model reasoning.

Long-context processing remains a significant challenge for large language models (LLMs). Retrieval-augmented generation (RAG) has recently emerged as a promising approach, enabling LLMs to selectively access relevant information from extended contexts to improve efficiency. However, existing RAG approaches often lag behind other efficient long-context processing methods primarily due to inherent limitations on inaccurate retrieval and fragmented contexts. To address these limitations, we propose RetroLM, a novel RAG framework designed for effective long-context processing. Unlike traditional approaches, RetroLM introduces KV-level retrieval augmentation, which partitions the LLM's KV cache into contiguous pages and performs encoding and decoding operations based on the retrieved KV pages. Built upon this framework, we further develop a specialized retriever for precise retrieval of critical pages and conduct unsupervised post-training to optimize the model’s ability to leverage retrieved information. Compared with traditional RAG, the new approach enhances robustness to retrieval inaccuracy, facilitates effective utilization of fragmented contexts, and saves the cost from repeated context-encoding operations. We conduct extensive evaluations across several popular benchmarks, including LongBench, InfiniteBench, and RULER. RetroLM consistently outperforms existing long-LLMs and RAG-based methods, especially in tasks requiring deep reasoning or extreme context lengths.

In this paper, we introduce ReasonEmbed, a novel text embedding model developed for reasoning-intensive document retrieval.Our work includes three key technical contributions.First, we propose ReMixer, a new data synthesis method that overcomes the triviality problem prevalent in previous synthetic datasets, enabling large-scale production of 82K highquality training samples.Second, we design Redapter, a self-adaptive learning algorithm that dynamically adjusts training each sample's weight based on its reasoning intensity.This allows the model to effectively capture the complex semantic relationships between queries and documents.Third, we implement Rea-sonEmbed across multiple backbones of varying sizes, all of which achieve superior performance on reasoning-intensive retrieval tasks.Notably, our ReasonEmbed-Qwen3-8B model offers a record-high nDCG@10 score of 38.1 on the BRIGHT benchmark (SU et al., 2025), which significantly outperforms existing text embedding models.We will fully open-source our created resources in ReasonEmbed to push forward the research advancement in this field 1 .

Kun Luo, Hongjin Qian, Zheng Liu, Ziyi Xia, Shitao Xiao, Zhao Cao, Siqi Bao, Jun Zhao, Kang Liu. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Retrieval-Augmented Generation (RAG) improves the performance of Large Language Models (LLMs) by retrieving and integrating relevant information from external knowledge bases, which helps generate more accurate responses. However, RAG is vulnerable to retrieval poisoning attacks , where attackers can induce LLM to produce inaccurate responses by injecting malicious documents into the retrieval process. In this article, we propose ShieldRAG , a novel defense framework designed to counteract retrieval poisoning attacks by reshaping the retrieval embedding space. ShieldRAG leverages a dual-strategy effect realized via a majority-consensus mechanism: ① Push : Implicitly forces the embedding of a user query away from malicious documents by filtering out their minority signals, reducing their influence. ② Pull : Aligns the embedding of a user query closer to that of benign documents, reinforcing accurate retrieval. These strategies work synergistically to preserve retrieval integrity and enhance the quality of LLM-generated responses. Specifically, ShieldRAG operates through three key steps: Sliding Retrieval Explanation Generation , Keyword Aggregation , and Query Targeting Optimization . These three steps collectively ensure the effective integration of information from benign sources while filtering out malicious interference, thereby significantly enhancing the robustness of RAG systems against retrieval poisoning attacks. We evaluate ShieldRAG on four open-domain Question Answering (QA) datasets: Natural Questions, MS-MARCO, HotpotQA, and 2WikiMultiHopQA, using seven representative LLMs. Extensive experiments demonstrate that ShieldRAG significantly improves response accuracy while mitigating adversarial effects, showcasing strong generalization across multiple datasets and LLM architectures.

As a primary means of information acquisition, information retrieval (IR) systems, such as search engines, have integrated themselves into our daily lives. These systems also serve as components of dialogue, question-answering, and recommender systems. The trajectory of IR has evolved dynamically from its origins in term-based methods to its integration with advanced neural models. While the neural models excel at capturing complex contextual signals and semantic nuances, they still face challenges such as data scarcity, interpretability, and the generation of contextually plausible yet potentially inaccurate responses. This evolution requires a combination of traditional methods (such as term-based sparse retrieval methods with rapid response) and modern neural architectures (such as language models with powerful language understanding capacity). Meanwhile, the emergence of large language models (LLMs) has revolutionized natural language processing due to their remarkable language understanding, generation, and reasoning abilities. Consequently, recent research has sought to leverage LLMs to improve IR systems. Given the rapid evolution of this research trajectory, it is necessary to consolidate existing methodologies and provide nuanced insights through a comprehensive overview. In this survey, we delve into the confluence of LLMs and IR systems, including crucial aspects such as query rewriters, retrievers, rerankers, readers, and search agents.

The pervasive spread of misinformation on social networks highlights the critical necessity for effective fact verification systems. Traditional approaches primarily focus on pairwise correlations between claims and evidence, often neglecting comprehensive multi-hop retrieval and reasoning, which results in suboptimal performance when dealing with complex claims. In this paper, we propose MRR-FV, a generative retrieval-enhanced model designed to address the novel challenge of Multi-hop Retrieval and Reasoning for Fact Verification, which integrates two core modules: Generative Multi-hop Retriever and the Hierarchical Interaction Reasoner. MRR-FV utilizes an autoregressive model for iterative multi-hop evidence retrieval, complemented by a pre-trained compressor to address the challenge of intention shift across retrieval hops. For claim verification, we propose a hierarchical interaction reasoner that conducts intra-sentence reasoning to capture long-term semantic dependencies and inter-sentence reasoning across multi-hop evidence subgraphs to reveal complex evidence interactions. Experimental evaluations on the FEVER and HOVER datasets demonstrate the superior performance of our model in both claim verification and evidence retrieval tasks.

Despite the rapidly growing demand for multimodal retrieval, progress in this field remains severely constrained by a lack of training data. In this paper, we introduce MegaPairs, a novel data synthesis method that leverages vision language models (VLMs) and open-domain images, together with a massive synthetic dataset generated from this method. Our empirical analysis shows that MegaPairs generates high-quality data, enabling the multimodal retriever to significantly outperform the baseline model trained on 70\times more data from existing datasets. Moreover, since MegaPairs solely relies on general image corpora and open-source VLMs, it can be easily scaled up, enabling continuous improvements in retrieval performance. In this stage, we produced more than 26 million training instances and trained several models of varying sizes using this data. These new models achieve state-of-the-art zero-shot performance across 4 popular composed image retrieval (CIR) benchmarks and the highest overall performance on the 36 datasets provided by MMEB. They also demonstrate notable performance improvements with additional downstream fine-tuning. Our code, synthesized dataset, and pre-trained models are publicly available at https://github.com/VectorSpaceLab/MegaPairs.

Jianlyu Chen, Nan Wang, Chaofan Li, Bo Wang, Shitao Xiao, Han Xiao, Hao Liao, Defu Lian, Zheng Liu. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Chaofan Li, Jianlyu Chen, Yingxia Shao, Chaozhuo Li, Quanqing Xu, Defu Lian, Zheng Liu. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Zheng Liu, Ze Liu, Zhengyang Liang, Junjie Zhou, Shitao Xiao, Chao Gao, Chen Jason Zhang, Defu Lian. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Processing long contexts poses a significant challenge for large language models (LLMs) due to their inherent context-window limitations and the computational burden of extensive key-value (KV) activations, which severely impact efficiency.For information-seeking tasks, full context perception is often unnecessary, as a query's information needs can dynamically range from localized details to a global perspective, depending on its complexity.However, existing methods struggle to adapt effectively to these dynamic information needs.In the paper, we propose a method for processing long-context information-seeking tasks via query-guided ACtivation REfilling (ACRE).ACRE constructs a Bi-layer KV Cache for long contexts, where the layer-1 (L1) cache compactly captures global information, and the layer-2 (L2) cache provides detailed and localized information.ACRE establishes a proxying relationship between the two caches, allowing the input query to attend to the L1 cache and dynamically refill it with relevant entries from the L2 cache.This mechanism integrates global understanding with query-specific local details, thus improving answer decoding.Experiments on a variety of long-context informationseeking datasets demonstrate ACRE's effectiveness, achieving improvements in both performance and efficiency.We will release our source codes in this repository.

Retrieval-augmented Large Language Models (RaLLMs) are reshaping knowledge acquisition, offering long-form, knowledge-grounded answers through advanced reasoning and generation capabilities. Despite the emergence of impactful systems like WebGPT and New Bing, the reliability of RaLLMs, especially in complex situations, is under scrutiny. Our study tackles this concern by evaluating RaLLMs’ question-answering performance using a novel benchmark focusing on Correctness and Groundedness. Correctness measures the logical soundness of the responses, and Groundedness checks for support by relevant references. We introduce an automated model-based evaluation pipeline for multi-hop question-answering tasks, revealing RaLLMs’ proneness to generating inaccuracies when dealing with flawed or partial knowledge. To improve accuracy, we introduce two reasoning strategies, Self-Reflection’ and Self-Completion,’ enabling RaLLMs to identify and fill knowledge gaps, significantly improving answer quality without extensive model retraining.

Recent advancements in text-to-image generation, notably the series of Stable Diffusion methods, have enabled the production of diverse, high-quality photo-realistic images. Nevertheless, these techniques still exhibit limitations in terms of knowledge access. Retrieval-augmented image generation is a straightforward way to tackle this problem. Current studies primarily utilize coarse-grained retrievers, employing initial prompts as search queries for knowledge retrieval. This approach, however, is ineffective in accessing valuable knowledge in long-tail text-to-image generation scenarios. To alleviate this problem, we introduce FineRAG, a fine-grained model that systematically breaks down the retrieval-augmented image generation task into four critical stages: query decomposition, candidate selection, retrieval-augmented diffusion, and self-reflection. Experimental results on both general and long-tailed benchmarks show that our proposed method significantly reduces the noise associated with retrieval-augmented image generation and performs better in complex, open-world scenarios.

The evaluation of Long Video Understanding (LVU) performance poses an important but challenging research problem. Despite previous efforts, the existing video understanding benchmarks are severely constrained by several issues, especially the insufficient lengths of videos, a lack of diversity in video types and evaluation tasks, and the inappropriateness for evaluating LVU performances. To address the above problems, we propose a new benchmark called MLVU (Multitask Long Video Understanding Benchmark) for the comprehensive and in-depth evaluation of LVU. MLVU presents the following critical values: 1) The substantial and flexible extension of video lengths, which enables the benchmark to evaluate LVU performance across a wide range of durations. 2) The inclusion of various video genres, such as movies, surveillance, egocentric videos, and cartoons, reflects the models’ LVU performances in different scenarios. 3) The development of diversified evaluation tasks, which enables a comprehensive examination of MLLMs’ key abilities in long-video understanding. The empirical study with 23 latest MLLMs reveals significant room for improvement in today’s technique, as all existing methods struggle with most of the evaluation tasks and exhibit severe performance degradation when handling longer videos. Additionally, it suggests that factors such as context length, image-understanding ability, and the choice of LLM backbone can play critical roles in future advancements. We anticipate that MLVU will advance the research of LVU by providing a comprehensive and in-depth analysis of MLLMs. The code and dataset can be accessed from https://github.com/JUNJIE99/MLVU.

Multimodal Large Language Models (MLLMs) have displayed remarkable performance in multi-modal tasks, particularly in visual comprehension. However, we reveal that MLLMs often generate incorrect answers even when they understand the visual content. To this end, we manually construct a benchmark with 12 categories and design evaluation metrics that assess the degree of error in MLLM responses even when the visual content is seemingly understood. Based on this benchmark, we test 15 leading MLLMs and analyze the distribution of attention maps and logits of some MLLMs. Our investigation identifies two primary issues: 1) most instruction tuning datasets predominantly feature questions that "directly" relate to the visual content, leading to a bias in MLLMs’ responses to other indirect questions, and 2) MLLMs’ attention to visual tokens is notably lower than to system and question tokens. We further observe that attention scores between questions and visual tokens as well as the model’s confidence in the answers are lower in response to misleading questions than to straightforward ones. To address the first challenge, we introduce a paired positive and negative data construction pipeline to diversify the dataset. For the second challenge, we propose to enhance the model’s focus on visual content during decoding by refining the text and visual prompt. For the text prompt, we propose a content guided refinement strategy that performs preliminary visual content analysis to generate structured information before answering the question. Additionally, we employ a visual attention refinement strategy that highlights question-relevant visual tokens to increase the model’s attention to visual content that aligns with the question. Extensive experiments demonstrate that these challenges can be significantly mitigated with our proposed dataset and techniques.

Long video understanding poses a significant challenge for current Multi-modal Large Language Models (MLLMs). Notably, the MLLMs are constrained by their limited context lengths and the substantial costs while processing long videos. Although several existing methods attempt to reduce visual tokens, their strategies encounter severe bottleneck, restricting MLLMs’ ability to perceive fine-grained visual details. In this work, we propose Video-XL, a novel approach that leverages MLLMs’ inherent key-value (KV) sparsification capacity to condense the visual input. Specifically, we introduce a new special token, the Visual Summarization Token (VST), for each interval of the video, which summarizes the visual information within the interval as its associated KV. The VST module is trained by instruction fine-tuning, where two optimizing strategies are offered. 1. Curriculum learning, where VST learns to make small (easy) and large compression (hard) progressively. 2. Composite data curation, which integrates single-image, multi-image, and synthetic data to overcome the scarcity of long-video instruction data. The compression quality is further improved by dynamic compression, which customizes compression granularity based on the information density of different video intervals. Video-XL’s effectiveness is verified from three aspects. First, it achieves a superior long-video understanding capability, outperforming state-of-the-art models of comparable sizes across multiple popular benchmarks. Second, it effectively preserves video information, with minimal compression loss even at 16 × compression ratio. Third, it realizes outstanding cost-effectiveness, enabling high-quality processing of thousands of frames on a single A100 GPU.

The emergence of Large Language Models (LLMs) has unified language generation tasks and revolutionized human-machine interaction. However, in the realm of image generation, a unified model capable of handling various tasks within a single framework remains largely unexplored. In this work, we introduce OmniGen, a new diffusion model for unified image generation. OmniGen is characterized by the following features: 1) Unification: OmniGen not only demonstrates text-to-image generation capabilities but also inherently supports various downstream tasks, such as image editing, subject-driven generation, and visual-conditional generation. 2) Simplicity: The architecture of OmniGen is highly simplified, eliminating the need for additional plugins. Moreover, compared to existing diffusion models, it is more user-friendly and can complete complex tasks end-to-end through instructions without the need for extra intermediate steps, greatly simplifying the image generation workflow. 3) Knowledge Transfer: Benefit from learning in a unified format, OmniGen effectively transfers knowledge across different tasks, manages unseen tasks and domains, and exhibits novel capabilities. We also explore the model’s reasoning capabilities and potential applications of the chain-of-thought mechanism. This work represents the first attempt at a general-purpose image generation model, and we will release our resources at https://github.com/VectorSpaceLab/OmniGen to foster future advancements.

Existing studies have optimized retrievalaugmented generation (RAG) across various sub-tasks, such as query understanding and retrieval refinement, but integrating these optimizations into a unified framework remains challenging.To tackle this problem, this work proposes RoleRAG, a unified RAG framework that achieves efficient multi-task processing through role-specific token optimization.RoleRAG comprises six modules, each handling a specific sub-task within the RAG process.Additionally, we introduce a query graph to represent the decomposition of the query, which can be dynamically resolved according to the decomposing state.All modules are driven by the same underlying LLM, distinguished by task-specific role tokens that are individually optimized.This design allows RoleRAG to dynamically activate different modules within a single LLM instance, thereby streamlining deployment and reducing resource consumption.Experimental results on five open-domain question-answering datasets demonstrate the effectiveness, generalizability, and flexibility of our framework.

Large language models (LLMs) with decoder-only architectures have demonstrated exceptional text-generation capabilities across a variety of tasks. Some researchers have also adapted these models for text representation tasks. However, in text representation tasks, these models often face performance degradation on unseen tasks. In-context learning (ICL), which leverages examples provided in the input context, enables LLMs to handle unseen tasks effectively. Inspired by this, we aim to fully utilize the inherent properties of LLMs to enhance text representation performance across different tasks through the ICL approach. In this paper, we introduce a simple yet effective training strategy, which significantly improves text representation capabilities. Unlike previous models that prepend task instructions to the text, our method randomly samples a varying number of examples during training, endowing the embedding model with in-context learning abilities while maintaining its zero-shot capabilities. This approach does not require additional data construction or modifications to the model architecture. On the contrary, we find that some popular modifications to the model, such as bidirectional attention, can degrade performance, undermining the inherent characteristics of LLMs. We have publicly released our method at this \href{https://github.com/FlagOpen/FlagEmbedding}{repo}.

Long context compression is a critical research problem due to its significance in reducing the high computational and memory costs associated with LLMs. In this paper, we propose Activation Beacon, a plug-in module for transformer-based LLMs that targets effective, efficient, and flexible compression of long contexts. To achieve this, our method introduces the following technical designs. 1) We directly compress the activations (i.e. keys and values at every layer), rather than leveraging soft prompts to relay information (which constitute a major bottleneck to encapsulate the complex information within long contexts). 2) We tailor the compression workflow, where each fine-grained input unit is progressively compressed, enabling high-quality compression and efficient computation during both training and inference. 3) We train the model through compression-based auto-regression, making full use of plain texts and instructional data to optimize the model's compression performance. 4) During training, we randomly sample a compression ratio at each step, teaching the model to support a wide range of compression configurations. Extensive evaluations are conducted on various long-context tasks whose lengths (e.g., 128K) may far exceed the maximum training length (20K), such as document understanding, few-shot learning, and Needle-in-a-Haystack. Whilst existing methods struggle to handle these challenging tasks, Activation Beacon maintains a comparable performance to the uncompressed baseline across various scenarios, achieving a 2x acceleration in inference time and an 8x reduction of memory costs for KV cache.

Most jailbreak methods for large language models (LLMs) focus on superficially improving attack success through manually defined rules. However, they fail to uncover the underlying mechanisms within target LLMs that explain why an attack succeeds or fails. In this paper, we propose investigating the phenomenon of jailbreaks and defenses for LLMs from the perspective of attention distributions within the models. A preliminary experiment reveals that the success of a jailbreak is closely linked to the LLM's attention on sensitive words.Inspired by this interesting finding, we propose incorporating critical signals derived from internal attention distributions within LLMs, namely Attention Intensity on Sensitive Words and Attention Dispersion Entropy, to guide both attacks and defenses. Drawing inspiration from the concept of "Feint and Attack", we introduce an attention-guided jailbreak model, ABA, which redirects the model's attention to benign contexts, and an attention-based defense model, ABD, designed to detect attacks by analyzing internal attention entropy. Experimental results demonstrate the superiority of our proposal when compared to SOTA baselines.

Knowledge-intensive tasks often require complex reasoning and contextual understanding over long contexts. However, the learning and deployment of long-LLMs remains a challenging problem despite recent progresses. In this work, we propose that the short LLMs have great potentiality for solving knowledge-intensive tasks that have long context, i.e. they can be solved by purely working with oracle short-contexts within the input long-context. On top of this argument, we propose a framework called DCISO DynamiC knowledge-Intensive task S>Olver), which enables a short-LLM to address the knowledge-intensive tasks with long context via dynamic context browsing. In our framework, the short-LLM prompts itself to reason for two critical decisions: 1) how to access to the appropriate part of context within the input, 2) how to make effective use of the accessed context. By adaptively accessing and utilizing the context based on the presented tasks, DCISO can serve as a general framework to handle diversified knowledge-intensive long-context problems. We comprehensively evaluate different types of tasks from popular long-context benchmarks, where DCISO is able to achieve a substantially improved performance. Our codes will be released at this repository.

Code embedding models attract increasing attention due to the widespread popularity of retrieval-augmented generation (RAG) in software development. These models are expected to capture the rich semantic relationships inherent to code, which differ significantly from those found in text. However, existing models remain severely limited due to the scarcity of high-quality training data. In this work, we introduce \textbf{CodeR} (\underline{Code} \underline{R}etrieval), a state-of-the-art embedding model for general-purpose code retrieval. The superior performance of CodeR is built upon \textbf{CodeR-Pile}, a large-scale synthetic dataset constructed under the DRU (Diversity, Reliability, Usability) principle via a novel data synthesis pipeline. To optimize training effectiveness, we propose \textbf{Annealing}, a curriculum learning strategy that enables effective knowledge transfer across heterogeneous sources of data. We evaluate CodeR based on 16 diverse code retrieval tasks, where it significantly outperforms existing baselines and exhibits strong out-of-domain generalization performance. We have publicly released our code and the well-trained model to facilitate further research in this critical area\footnote{\url{https://github.com/FlagOpen/FlagEmbedding/tree/master/research/BGE_Coder}}.

Augmenting large language models (LLMs) with external retrieval has become a standard method to address their inherent knowledge cutoff limitations. However, traditional retrieval-augmented generation methods employ static, pre-inference retrieval strategies, making them inadequate for complex tasks involving ambiguous, multi-step, or evolving information needs. Recent advances in test-time scaling techniques have demonstrated significant potential in enabling LLMs to dynamically interact with external tools, motivating the shift toward adaptive inference-time retrieval. Inspired by Information Foraging Theory (IFT), we propose InForage, a reinforcement learning framework that formalizes retrieval-augmented reasoning as a dynamic information-seeking process. Unlike existing approaches, InForage explicitly rewards intermediate retrieval quality, encouraging LLMs to iteratively gather and integrate information through adaptive search behaviors. To facilitate training, we construct a human-guided dataset capturing iterative search and reasoning trajectories for complex, real-world web tasks. Extensive evaluations across general question answering, multi-hop reasoning tasks, and a newly developed real-time web QA dataset demonstrate InForage's superior performance over baseline methods. These results highlight InForage's effectiveness in building robust, adaptive, and efficient reasoning agents. We provide all codes and datasets in the supplementary materials.

In real-world information-seeking scenarios, users have dynamic and diverse needs, requiring RAG systems to demonstrate adaptable resilience. To comprehensively evaluate the resilience of current RAG methods, we introduce HawkBench, a human-labeled, multi-domain benchmark designed to rigorously assess RAG performance across categorized task types. By stratifying tasks based on information-seeking behaviors, HawkBench provides a systematic evaluation of how well RAG systems adapt to diverse user needs.Unlike existing benchmarks, which focus primarily on specific task types (mostly factoid queries) and rely on varying knowledge bases, HawkBench offers: (1) systematic task stratification to cover a broad range of query types, including both factoid and rationale queries, (2) integration of multi-domain corpora across all task types to mitigate corpus bias, and (3) rigorous annotation for high-quality evaluation.HawkBench includes 1,600 high-quality test samples, evenly distributed across domains and task types. Using this benchmark, we evaluate representative RAG methods, analyzing their performance in terms of answer quality and response latency. Our findings highlight the need for dynamic task strategies that integrate decision-making, query interpretation, and global knowledge understanding to improve RAG generalizability. We believe HawkBench serves as a pivotal benchmark for advancing the resilience of RAG methods and their ability to achieve general-purpose information seeking.

Accurately locating key moments within long videos is crucial for solving long video understanding (LVU) tasks. However, existing benchmarks are either severely limited in terms of video length and task diversity, or they focus solely on the end-to-end LVU performance, making them inappropriate for evaluating whether key moments can be accurately accessed. To address this challenge, we propose MomentSeeker, a novel benchmark for long-video moment retrieval (LVMR), distinguished by the following features. First, it is created based on long and diverse videos, averaging over 1,200 seconds in duration, and collected from various domains, e.g., movie, anomaly, egocentric, and sports. Second, it covers a variety of real-world scenarios in three levels: global-level, event-level, and object-level, covering common tasks like action recognition, object localization, causal reasoning, etc. Third, it incorporates rich forms of queries, including text-only queries, image-conditioned queries, and video-conditioned queries. On top of MomentSeeker, we conduct comprehensive experiments for both generation-based approaches (directly using MLLMs) and retrieval-based approaches (leveraging video retrievers). Our results reveal the significant challenges in long-video moment retrieval in terms of accuracy and efficiency, despite improvements from the latest long-video MLLMs and task-specific fine-tuning. We have publicly released MomentSeeker to facilitate future research in this area.

The existing Retrieval-Augmented Generation (RAG) systems face significant challenges in terms of cost and effectiveness. On one hand, they need to encode the lengthy retrieved contexts before responding to the input tasks, which imposes substantial computational overhead. On the other hand, directly using generic Large Language Models (LLMs) often leads to sub-optimal answers, while task-specific fine-tuning may compromise the LLMs' general capabilities. To address these challenges, we introduce a novel approach called FlexRAG (Flexible Context Adaptation for RAG). In this approach, the retrieved contexts are compressed into compact embeddings before being encoded by the LLMs. Simultaneously, these compressed embeddings are optimized to enhance downstream RAG performance. A key feature of FlexRAG is its flexibility, which enables effective support for diverse compression ratios and selective preservation of important contexts. With these designs, FlexRAG achieves superior generation quality while significantly reducing running costs. The experiments across multiple QA datasets validate our approach as a cost-effective and flexible solution for RAG systems (codebase: https://github.com/wcyno23/FlexRAG).

Large language models (LLMs) provide powerful foundations to perform fine-grained text re-ranking. However, they are often prohibitive in reality due to constraints on computation bandwidth. In this work, we propose a flexible architecture called Matroyshka Re-Ranker, which is designed to facilitate runtime customization of model layers and sequence lengths at each layer based on users' configurations. Consequently, the LLM-based re-rankers can be made applicable across various real-world situations. The increased flexibility may come at the cost of precision loss. To address this problem, we introduce a suite of techniques to optimize the performance. First, we propose cascaded self-distillation, where each sub-architecture learns to preserve a precise re-ranking performance from its super components, whose predictions can be exploited as smooth and informative teacher signals. Second, we design a factorized compensation mechanism, where two collaborative LoRA modules, vertical and horizontal, are jointly employed to compensate for the precision loss resulted from arbitrary combinations of layer and sequence compression. We perform comprehensive experiments using passage and document retrieval datasets from MSMARCO, along with all public datasets from BEIR. In our experiments, Matryoshka Re-Ranker substantially outperforms existing methods, while effectively preserving its superior performance across various compression forms and application scenarios. We have publicly released our method at this https://github.com/FlagOpen/FlagEmbedding repo.

Processing long contexts presents a significant challenge for large language models (LLMs). While recent advancements allow LLMs to handle much longer contexts than before (e.g., 32K or 128K tokens), it is computationally expensive and can still be insufficient for many applications. Retrieval-Augmented Generation (RAG) is considered a promising strategy to address this problem. However, conventional RAG methods face inherent limitations because of two underlying requirements: 1) explicitly stated queries, and 2) well-structured knowledge. These conditions, however, do not hold in general long-context processing tasks.

Yutao Zhu, Peitian Zhang, Chenghao Zhang, Yifei Chen, Binyu Xie, Zheng Liu, Ji-Rong Wen, Zhicheng Dou. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Dense retrieval calls for discriminative embeddings to represent the semantic relationship between query and document.It may benefit from the using of large language models (LLMs), given LLMs' strong capability on semantic understanding.However, the LLMs are learned by auto-regression, whose working mechanism is completely different from representing whole text as one discriminative embedding.Thus, it is imperative to study how to adapt LLMs properly so that they can be effectively initialized as the backbone encoder for dense retrieval.In this paper, we propose a novel approach, called Llama2Vec, which performs unsupervised adaptation of LLM for its dense retrieval application.Llama2Vec consists of two pretext tasks: EBAE (Embedding-Based Auto-Encoding) and EBAR (Embedding-Based Auto-Regression), where the LLM is prompted to reconstruct the input sentence and predict the next sentence based on its text embeddings.Llama2Vec is simple, lightweight, but highly effective.It is used to adapt LLaMA-2-7B on the Wikipedia corpus.With a moderate steps of adaptation, it substantially improves the model's fine-tuned performances on a variety of dense retrieval benchmarks.Notably, it results in the new state-of-the-art performances on popular benchmarks, such as passage and document retrieval on MSMARCO, and zero-shot retrieval on BEIR.The model and source code will be made publicly available to facilitate the future research.Our model is available at https://github.com/FlagOpen/FlagEmbedding.

Retrieval augmentation is a promising approach to handle long-context language modeling.However, the existing retrieval methods usually work with the chunked context, which is prone to inferior quality of semantic representation and incomplete retrieval of useful information.In this work, we propose a new method for the retrieval augmentation of longcontext language modeling, called Landmark Embedding.Our method is characterized by threefold technical contributions.Firstly, we introduce a chunking-free architecture, which keeps the long context coherent such that highquality embeddings can be generated for the fine-grained units within the context.Secondly, we present a position-aware objective function, which prioritizes the ultimate boundary for a consecutive span of information.By learning to discriminate such a special position, the useful information can be comprehensively retrieved for the query.Thirdly, we design a novel multistage learning algorithm, which makes the best use of readily available data and synthetic data for cost-effective training of the landmark embedding.In our experimental study, landmark embedding is able to substantially improve the performance for both LLaMA-2 and ChatGPT in a variety of long-context tasks; meanwhile, it also outperforms the existing retrieval methods with a notable advantage.Our model and code will be made publicly available 1 .

This paper presents a novel Chunking-Free In-Context (CFIC) retrieval approach, specifically tailored for Retrieval-Augmented Generation (RAG) systems.Traditional RAG systems often struggle with grounding responses using precise evidence text due to the challenges of processing lengthy documents and filtering out irrelevant content.Commonly employed solutions, such as document chunking and adapting language models to handle longer contexts, have their limitations.These methods either disrupt the semantic coherence of the text or fail to effectively address the issues of noise and inaccuracy in evidence retrieval.CFIC addresses these challenges by circumventing the conventional chunking process.It utilizes the encoded hidden states of documents for in-context retrieval, employing autoaggressive decoding to accurately identify the specific evidence text required for user queries, eliminating the need for chunking.CFIC is further enhanced by incorporating two decoding strategies, namely Constrained Sentence Prefix Decoding and Skip Decoding.These strategies not only improve the efficiency of the retrieval process but also ensure that the fidelity of the generated grounding text evidence is maintained.Our evaluations of CFIC on a range of open QA datasets demonstrate its superiority in retrieving relevant and accurate evidence, offering a significant improvement over traditional methods.By doing away with the need for document chunking, CFIC presents a more streamlined, effective, and efficient retrieval solution, making it a valuable advancement in the field of RAG systems.The codes will be released in this repository.

LLMs confront inherent limitations in terms of its knowledge, memory, and action.The retrieval augmentation stands as a vital mechanism to address these limitations, which brings in useful information from external sources to augment the LLM.However, existing retrieval methods encounter two pressing issues.On one hand, the general retrievers are not properly optimized for retrieval augmentation hence exhibit limited effectiveness; on the other hand, the task-specific retrievers excel in the targeted retrieval augmentation scenario, while lack the versatility to handle diverse scenarios.In this work, we propose LLM-Embedder for the unified support of diverse retrieval augmentation scenarios.Our method presents three technical contributions.Firstly, we introduce a new reward formulation, namely rank-aware reward.It exploits the ranking position of the desired output among N sampled outputs from the LLM, which leads to fine-grained and robust computation of reward from the LLM's feedback.Secondly, we design a novel distillation objective, called graded distillation.It incorporates both the absolute value and the relative order of the reward for more sufficient utilization of the LLM's feedback.Thirdly, we systematically optimize the multi-task learning, which effectively unifies the multiple retrieval functionalities into one model.In our experiment, LLM-Embedder notably improves the LLM's performances in various downstream tasks, and outperforms both general and task-specific retrievers with a substantial advantage.Our data, code, and model have been released at https: //github.com/FlagOpen/FlagEmbedding.

Multi-modal retrieval becomes increasingly popular in practice.However, the existing retrievers are mostly text-oriented, which lack the capability to process visual information.Despite the presence of vision-language models like CLIP, the current methods are severely limited in representing the text-only and imageonly data.In this work, we present a new embedding model VISTA for universal multimodal retrieval.Our work brings forth threefold technical contributions.Firstly, we introduce a flexible architecture which extends a powerful text encoder with the image understanding capability by introducing visual token embeddings.Secondly, we develop two data generation strategies, which bring highquality composed image-text to facilitate the training of the embedding model.Thirdly, we introduce a multi-stage training algorithm, which first aligns the visual token embedding with the text encoder using massive weakly labeled data, and then develops multi-modal representation capability using the generated composed image-text data.In our experiments, VISTA achieves superior performances across a variety of multi-modal retrieval tasks in both zero-shot and supervised settings.Our model, data, and source code are available at https://github.com/FlagOpen/FlagEmbedding.

Pre-trained language models like BERT and T5 serve as crucial backbone encoders for dense retrieval.However, these models often exhibit limited generalization capabilities and face challenges in improving in-domain accuracy.Recent research has explored using large language models (LLMs) as retrievers, achieving state-of-the-art performance across various tasks.Despite these advancements, the specific benefits of LLMs over traditional retrievers and the impact of different LLM configurations-such as parameter sizes, pre-training duration, and alignment processes-on retrieval tasks remain unclear.In this work, we conduct a comprehensive empirical study on six key dimensions of dense retrieval capabilities, including in-domain accuracy, data efficiency, zero-shot generalization, lengthy retrieval, instruction-based retrieval, and multi-task learning.We evaluate over 15 different backbone LLMs and non-LLMs.Our findings reveal that larger models and extensive pre-training consistently enhance in-domain accuracy and data efficiency.Additionally, larger models demonstrate significant potential in zero-shot generalization, lengthy retrieval, instruction-based retrieval, and multi-task learning.These results underscore the advantages of LLMs as versatile and effective backbone encoders in dense retrieval, providing valuable insights for future research and development in this field.

Conversational search requires accurate interpretation of user intent from complex multiturn contexts.This paper presents ChatRetriever, which inherits the strong generalization capability of large language models to robustly represent complex conversational sessions for dense retrieval.To achieve this, we propose a simple and effective dual-learning approach that adapts LLM for retrieval via contrastive learning while enhancing the complex session understanding through masked instruction tuning on high-quality conversational instruction tuning data.Extensive experiments on five conversational search benchmarks demonstrate that ChatRetriever substantially outperforms existing conversational dense retrievers, achieving state-of-the-art performance on par with LLM-based rewriting approaches.Furthermore, ChatRetriever exhibits superior robustness in handling diverse conversational contexts.Our work highlights the potential of adapting LLMs for retrieval with complex inputs like conversational search sessions and proposes an effective approach to advance this research direction.

The emergence of Large Language Models (LLMs) has significantly advanced natural language processing, but these models often generate factually incorrect information, known as "hallucination." Initial retrieval-augmented generation (RAG) methods like the "Retrieve-Read" framework was inadequate for complex reasoning tasks. Subsequent prompt-based RAG strategies and Supervised Fine-Tuning (SFT) methods improved performance but required frequent retraining and risked altering foundational LLM capabilities. To cope with these challenges, we propose Assistant-based Retrieval-Augmented Generation (AssistRAG), integrating an intelligent information assistant within LLMs. This assistant manages memory and knowledge through tool usage, action execution, memory building, and plan specification. Using a two-phase training approach—Curriculum Assistant Learning and Reinforced Preference Optimization—AssistRAG enhances information retrieval and decision-making. Experiments show AssistRAG significantly outperforms benchmarks, especially benefiting less advanced LLMs, by providing superior reasoning capabilities and accurate responses.

We introduce C-Pack, a package of resources that significantly advances the field of general text embeddings for Chinese. C-Pack includes three critical resources. 1) C-MTP is a massive training dataset for text embedding, which is based on the curation of vast unlabeled corpora and the integration of high-quality labeled corpora. 2) C-MTEB is a comprehensive benchmark for Chinese text embeddings covering 6 tasks and 35 datasets. 3) BGE is a family of embedding models covering multiple sizes. Our models outperform all prior Chinese text embeddings on C-MTEB by more than +10% upon the time of the release. We also integrate and optimize the entire suite of training methods for BGE. Along with our resources on general Chinese embedding, we release our data and models for English text embeddings. The English models also achieve state-of-the-art performance on the MTEB benchmark; meanwhile, our released English data is 2 times larger than the Chinese data. Both Chinese and English datasets are the largest public release of training data for text embeddings. All these resources are made publicly available at https://github.com/FlagOpen/FlagEmbedding.

Recently, generative retrieval has emerged as a promising alternative to the traditional retrieval paradigms. It assigns each document a unique identifier, known as the DocID, and employs a generative model to directly generate the relevant DocID for the input query. A common choice for the DocID is one or several natural language sequences, e.g. the title, synthetic queries, or n-grams, so that the pre-trained knowledge of the generative model can be effectively utilized. However, a sequence is generated token by token, where only the most likely candidates are kept and the rest are pruned at each decoding step, thus, retrieval fails if any token within the relevant DocID is falsely pruned. What's worse, during decoding, the model can only perceive preceding tokens in the DocID while being blind to subsequent ones, hence is prone to make such errors. To address this problem, we present a novel framework for generative retrieval, dubbed Term-Set Generation (TSGen). Instead of sequences, we use a set of terms as the DocID. The terms are selected based on learned weights from relevance signals, so that they concisely summarize the document's semantics and distinguish it from others. On top of the term-set DocID, we propose a permutation-invariant decoding algorithm, with which the term set can be generated in any permutation yet will always lead to the corresponding document. Remarkably, TSGen perceives all valid terms rather than only the preceding ones at each decoding step. Given the constant decoding space, it can make more reliable decisions due to the broader perspective. TSGen is also resilient to errors: the relevant DocID will not be falsely pruned as long as the decoded term belongs to it. Moreover, TSGen can explore the optimal decoding permutation of the term set on its own, which further improves the likelihood of generating the relevant DocID. Lastly, we design an iterative optimization procedure to incentivize the model to generate the relevant term set in its favorable permutation. We conduct extensive experiments on popular benchmarks of generative retrieval, which validate the effectiveness, the generalizability, the scalability, and the efficiency of TSGen.

Retrieval-augmented generation have become central in natural language processing due to their efficacy in generating factual content. While traditional methods employ single-time retrieval, more recent approaches have shifted towards multi-time retrieval for multi-hop reasoning tasks. However, these strategies are bound by predefined reasoning steps, potentially leading to inaccuracies in response generation. This paper introduces MetaRAG, an approach that combines the retrieval-augmented generation process with metacognition. Drawing from cognitive psychology, metacognition allows an entity to self-reflect and critically evaluate its cognitive processes. By integrating this, MetaRAG enables the model to monitor, evaluate, and plan its response strategies, enhancing its introspective reasoning abilities. Through a three-step metacognitive regulation pipeline, the model can identify inadequacies in initial cognitive responses and fixes them. Empirical evaluations show that MetaRAG significantly outperforms existing methods.

Abstract The advent of large language models marks a revolutionary breakthrough in artificial intelligence. With the unprecedented scale of training and model parameters, the capability of large language models has been dramatically improved, leading to human-like performances in understanding, language synthesizing, common-sense reasoning, etc. Such a major leap forward in general AI capacity will fundamentally change the pattern of how personalization is conducted. For one thing, it will reform the way of interaction between humans and personalization systems. Instead of being a passive medium of information filtering, like conventional recommender systems and search engines, large language models present the foundation for active user engagement. On top of such a new foundation, users’ requests can be proactively explored, and users’ required information can be delivered in a natural, interactable, and explainable way. For another thing, it will also considerably expand the scope of personalization, making it grow from the sole function of collecting personalized information to the compound function of providing personalized services. By leveraging large language models as a general-purpose interface, the personalization systems may compile user’s requests into plans, calls the functions of external tools (e.g., search engines, calculators, service APIs, etc.) to execute the plans, and integrate the tools’ outputs to complete the end-to-end personalization tasks. Today, large language models are still being rapidly developed, whereas the application in personalization is largely unexplored. Therefore, we consider it to be right the time to review the challenges in personalization and the opportunities to address them with large language models. In particular, we dedicate this perspective paper to the discussion of the following aspects: the development and challenges for the existing personalization system, the newly emerged capabilities of large language models, and the potential ways of making use of large language models for personalization.

To better support information retrieval tasks such as web search and open-domain question answering, growing effort is made to develop retrieval-oriented language models, e.g., RetroMAE (Xiao et al., 2022b) and many others (Gao and Callan, 2021; Wang et al., 2021a).Most of the existing works focus on improving the semantic representation capability for the contextualized embedding of the [CLS] token.However, recent study shows that the ordinary tokens besides [CLS] may provide extra information, which help to produce a better representation effect (Lin et al., 2022).As such, it's necessary to extend the current methods where all contextualized embeddings can be jointly pre-trained for the retrieval tasks.In this work, we propose a novel pre-training method called Duplex Masked Auto-Encoder, a.k.a.DupMAE.It is designed to improve the quality of semantic representation where all contextualized embeddings of the pre-trained model can be leveraged.It takes advantage of two complementary auto-encoding tasks: one reconstructs the input sentence with the [CLS] embedding; the other one predicts the bagof-words feature of the input sentence with the ordinary tokens' embeddings.The two tasks are jointly conducted to train a unified encoder, where the whole contextualized embeddings are aggregated in a compact way to produce the final semantic representation.DupMAE is simple but empirically competitive: it substantially improves the pre-trained model's representation capability and transferability, where superior retrieval performances can be achieved on popular benchmarks, like MS MARCO and BEIR.Our

Pre-trained language models (PLMs) have achieved the preeminent position in dense retrieval due to their powerful capacity in modeling intrinsic semantics.However, most existing PLM-based retrieval models encounter substantial computational costs and are infeasible for processing long documents.In this paper, a novel retrieval model Longtriever is proposed to embrace three core challenges of long document retrieval: substantial computational cost, incomprehensive document understanding, and scarce annotations.Longtriever splits long documents into short blocks and then efficiently models the local semantics within a block and the global context semantics across blocks in a tightly-coupled manner.A pretraining phase is further proposed to empower Longtriever to achieve a better understanding of underlying semantic correlations.Experimental results on two popular benchmark datasets demonstrate the superiority of our proposal.The source code is released at https: //github.com/SamuelYang1/Longtriever .

Inverted file structure is a common technique for accelerating dense retrieval.It clusters documents based on their embeddings; during searching, it probes nearby clusters w.r.t. an input query and only evaluates documents within them by subsequent codecs, thus avoiding the expensive cost of exhaustive traversal.However, the clustering is always lossy, which results in the miss of relevant documents in the probed clusters and hence degrades retrieval quality.In contrast, lexical matching, such as overlaps of salient terms, tends to be strong feature for identifying relevant documents.In this work, we present the Hybrid Inverted Index (HI 2 ), where the embedding clusters and salient terms work collaboratively to accelerate dense retrieval.To make best of both effectiveness and efficiency, we devise a cluster selector and a term selector, to construct compact inverted lists and efficiently searching through them.Moreover, we leverage simple unsupervised algorithms as well as end-to-end knowledge distillation to learn these two modules, with the latter further boosting the effectiveness.Based on comprehensive experiments on popular retrieval benchmarks, we verify that clusters and terms indeed complement each other, enabling HI 2 to achieve lossless retrieval quality with competitive efficiency across various index settings.

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.

Vector quantization is one of the critical techniques which enables dense retrieval for realtime applications. The recent study shows that vanilla vector quantization methods, like those implemented by FAISS [8], are lossy and prone to limited retrieval performances when large acceleration ratios are needed [14, 16, 18]. Besides, there have also been multiple algorithms which make the retriever and VQ better collaborated to alleviate such a loss. On top of these progresses, we develop LibVQ, which optimizes vector quantization for efficient dense retrieval. Our toolkit is highlighted for three advantages. 1. Effectiveness. The retrieval quality can be substantially improved over the vanilla implementations of VQ. 2. Simplicity. The optimization can be conducted in a lowcode fashion, and the optimization results can be easily loaded to ANN indexes to support downstream applications. 3. Universality. The optimization is agnostic to the embedding's learning process, and may accommodate different input conditions and ANN back-ends with little modification of the workflow. LibVQ may also support rich applications beyond dense retrieval, e.g., embedding compression, topic modeling, and de-duplication. In this demo, we provide comprehensive hand-on examples and evaluations for LibVQ. The toolkit is publicly released at: https://github.com/staoxiao/LibVQ/tree/demo.

A dozen recommendation libraries have recently been developed to accommodate popular recommendation algorithms for reproducibility. However, they are almost simply a collection of algorithms, overlooking the modularization of recommendation algorithms and their usage in practical scenarios. Algorithmic modularization has the following advantages: 1) helps to understand the effectiveness of each algorithm; 2) easily assembles new algorithms with well-performed modules by either drag-and-drop programming or automatic machine learning; 3) enables reinforcement between algorithms since one algorithm may act as a module of another algorithm. To this end, we develop a highly-modularized recommender system -- RecStudio, in which any recommendation algorithm is categorized into either a ranker or a retriever. In the RecStudio library, we implement 90 recommendation algorithms with the pure Pytorch, covering both common algorithms in other libraries and complex algorithms involving multiple recommendation models. RecStudio is featured from several perspectives, such as index-supported efficient recommendation and evaluation, GPU-accelerated negative sampling, hyperparameter learning on the validation, and cooperation between the retriever and ranker. RecStudio is also equipped with a web service, where the recommendation pipeline can be quickly established and visually evaluated on selected datasets, and the evaluation results are automatically archived and visualized in a leaderboard. The project and documents are released at http://recstudio.org.cn.

Deep recommender systems (DRS) are intensively applied in modern web services. To deal with the massive web contents, DRS employs a two-stage workflow: retrieval and ranking, to generate its recommendation results. The retriever aims to select a small set of relevant candidates from the entire items with high efficiency; while the ranker, usually more precise but time-consuming, is supposed to further refine the best items from the retrieved candidates. Traditionally, the two components are trained either independently or within a simple cascading pipeline, which is prone to poor collaboration effect. Though some latest works suggested to train retriever and ranker jointly, there still exist many severe limitations: item distribution shift between training and inference, false negative, and misalignment of ranking order. As such, it remains to explore effective collaborations between retriever and ranker.

Deep semantic matching aims at discriminating the relationship between documents based on deep neural networks. In recent years, it becomes increasingly popular to organize documents with a graph structure, then leverage both the intrinsic document features and the extrinsic neighbor features to derive discrimination. Most of the existing works mainly care about how to utilize the presented neighbors, whereas limited effort is made to filter appropriate neighbors. We argue that the neighbor features could be highly noisy and partially useful. Thus, a lack of effective neighbor selection will not only incur a great deal of unnecessary computation cost but also restrict the matching accuracy severely. In this work, we propose a novel framework, C ascaded D eep S emantic M atching ( CDSM ), for accurate and efficient semantic matching on textual graphs. CDSM is highlighted for its two-stage workflow. In the first stage, a lightweight CNN-based ad-hod neighbor selector is deployed to filter useful neighbors for the matching task with a small computation cost. We design both one-step and multi-step selection methods. In the second stage, a high-capacity graph-based matching network is employed to compute fine-grained relevance scores based on the well-selected neighbors. It is worth noting that CDSM is a generic framework which accommodates most of the mainstream graph-based semantic matching networks. The major challenge is how the selector can learn to discriminate the neighbors’ usefulness which has no explicit labels. To cope with this problem, we design a weak-supervision strategy for optimization, where we train the graph-based matching network at first and then the ad-hoc neighbor selector is learned on top of the annotations from the matching network. We conduct extensive experiments with three large-scale datasets, showing that CDSM notably improves the semantic matching accuracy and efficiency thanks to the selection of high-quality neighbors. The source code is released at https://github.com/jingjyyao/CDSM.

User identity linkage, which aims to link identities of a natural person across different social platforms, has attracted increasing research interest recently. Existing approaches usually first embed the identities as deterministic vectors in a shared latent space, and then learn a classifier based on the available annotations. However, the formation and characteristics of real-world social platforms are full of uncertainties, which makes these deterministic embedding based methods sub-optimal. Besides, semi-supervised models utilize the unlabeled data to help capture the intrinsic data distribution. However, the existing semi-supervised linkage methods heavily rely on the heuristically defined similarity measurements to incorporate the innate closeness between labeled and unlabeled samples. Such manually designed assumptions may not be consistent with the actual linkage signals and further introduce the noises. To address the mentioned limitations, in this paper we propose a novel Noise-aware Semi-supervised Variational User Identity Linkage (NSVUIL) model. Specifically, we first propose a novel supervised linkage module to incorporate the available annotations. Each social identity is represented by a Gaussian distribution in the Wasserstein space to simultaneously preserve the fine-grained social profiles and model the uncertainty of identities. Then, a noise-aware self-learning module is designed to faithfully augment the few available annotations, which is capable of filtering noises from the pseudo-labels generated by the supervised module. The filtered reliable candidates are added into the labeled set to provide enhanced training guidance for the next training iteration. Empirically, we evaluate the NSVUIL model over multiple real-world datasets, and the experimental results demonstrate its superiority.

We focus on Maximum Inner Product Search (MIPS), which is an essential problem in many machine learning communities. Given a query, MIPS finds the most similar items with the maximum inner products. Methods for Nearest Neighbor Search (NNS) which is usually defined on metric space do not exhibit the satisfactory performance for MIPS problem since inner product is a non-metric function. However, inner products exhibit many good properties compared with metric functions, such as avoiding vanishing and exploding gradients. As a result, inner product is widely used in many recommendation systems, which makes efficient Maximum Inner Product Search a key for speeding up many recommendation systems. Graph-based methods for NNS problem show the superiorities compared with other class methods. Each data point of the database is mapped to a node of the proximity graph. Nearest neighbor search in the database can be converted to route on the proximity graph to find the nearest neighbor for the query. This technique can be used to solve MIPS problem. Instead of searching the nearest neighbor for the query, we search the item with a maximum inner product with query on the proximity graph. In this article, we propose a reinforcement model to train an agent to search on the proximity graph automatically for MIPS problem if we lack the ground truths of training queries. If we know the ground truths of some training queries, our model can also utilize these ground truths by imitation learning to improve the agent’s searchability. By experiments, we can see that our proposed mode which combines reinforcement learning with imitation learning shows the superiorities over the state-of-the-art methods.

Graph neural networks (GNNs) have been widely applied in the recommendation tasks and have achieved very appealing performance. However, most GNN-based recommendation methods suffer from the problem of data sparsity in practice. Meanwhile, pre-training techniques have achieved great success in mitigating data sparsity in various domains such as natural language processing (NLP) and computer vision (CV) . Thus, graph pre-training has the great potential to alleviate data sparsity in GNN-based recommendations. However, pre-training GNNs for recommendations faces unique challenges. For example, user-item interaction graphs in different recommendation tasks have distinct sets of users and items, and they often present different properties. Therefore, the successful mechanisms commonly used in NLP and CV to transfer knowledge from pre-training tasks to downstream tasks such as sharing learned embeddings or feature extractors are not directly applicable to existing GNN-based recommendations models. To tackle these challenges, we delicately design an adaptive graph pre-training framework for localized collaborative filtering (ADAPT) . It does not require transferring user/item embeddings, and is able to capture both the common knowledge across different graphs and the uniqueness for each graph simultaneously. Extensive experimental results have demonstrated the effectiveness and superiority of ADAPT.

Maximum Inner Product Search (MIPS) plays an important role in many applications ranging from information retrieval, recommender systems to natural language processing and machine learning. However, exhaustive MIPS is often expensive and impractical when there are a large number of candidate items. The state-of-the-art approximated MIPS is product quantization with a score-aware loss, which weighs more heavily on items with larger inner product scores. However, it is challenging to extend the score-aware loss for additive quantization due to parallel-orthogonal decomposition of residual error. Learning additive quantization with respect to this loss is important since additive quantization can achieve a lower approximation error than product quantization. To this end, we propose a quantization method called Anisotropic Additive Quantization to combine the score-aware anisotropic loss and additive quantization. To efficiently update the codebooks in this algorithm, we develop a new alternating optimization algorithm. The proposed algorithm is extensively evaluated on three real-world datasets. The experimental results show that it outperforms the state-of-the-art baselines with respect to approximate search accuracy while guaranteeing a similar retrieval efficiency.

Generalized text representations are the foundation of many natural language understanding tasks. To fully utilize the different corpus, it is inevitable that models need to understand the relevance among them. However, many methods ignore the relevance and adopt a single-channel model (a coarse paradigm) directly for all tasks, which lacks enough rationality and interpretation. In addition, some existing works learn downstream tasks by stitches skill block (a fine paradigm), which might cause irrational results due to its redundancy and noise. In this work, we first analyze the task correlation through three different perspectives, , data property, manual design, and model-based relevance, based on which the similar tasks are grouped together. Then, we propose a hierarchical framework with a coarse-to-fine paradigm, with the bottom level shared to all the tasks, the mid-level divided to different groups, and the top-level assigned to each of the tasks. This allows our model to learn basic language properties from all tasks, boost performance on relevant tasks, and reduce the negative impact from irrelevant tasks. Our experiments on 13 benchmark datasets across five natural language understanding tasks demonstrate the superiority of our method.

Despite pre-training's progress in many important NLP tasks, it remains to explore effective pre-training strategies for dense retrieval.In this paper, we propose RetroMAE, a new retrieval oriented pre-training paradigm based on Masked Auto-Encoder (MAE).RetroMAE is highlighted by three critical designs.1) A novel MAE workflow, where the input sentence is polluted for encoder and decoder with different masks.The sentence embedding is generated from the encoder's masked input; then, the original sentence is recovered based on the sentence embedding and the decoder's masked input via masked language modeling.2) Asymmetric model structure, with a full-scale BERT like transformer as encoder, and a one-layer transformer as decoder.3) Asymmetric masking ratios, with a moderate ratio for encoder: 15∼30%, and an aggressive ratio for decoder: 50∼70%.Our framework is simple to realize and empirically competitive: the pre-trained models dramatically improve the SOTA performances on a wide range of dense retrieval benchmarks, like BEIR and MS MARCO.The source code and pre-trained models are made publicly available at https://github.com/staoxiao/RetroMAEso as to inspire more interesting research.

News recommendation calls for deep insights of news articles' underlying semantics. Therefore, pretrained language models (PLMs), like BERT and RoBERTa, may substantially contribute to the recommendation quality. However, it's extremely challenging to have news recommenders trained together with such big models: the learning of news recommenders requires intensive news encoding operations, whose cost is prohibitive if PLMs are used as the news encoder. In this paper, we propose a novel framework, SpeedyFeed, which efficiently trains PLMs-based news recommenders of superior quality. SpeedyFeed is highlighted for its light-weight encoding pipeline, which gives rise to three major advantages. Firstly, it makes the intermediate results fully reusable for the training workflow, which removes most of the repetitive but redundant encoding operations. Secondly, it improves the data efficiency of the training workflow, where non-informative data can be eliminated from encoding. Thirdly, it further saves the cost by leveraging simplified news encoding and compact news representation.

Embedding based retrieval (EBR) is a fundamental building block in many web applications. However, EBR in sponsored search is distinguished from other generic scenarios and technically challenging due to the need of serving multiple retrieval purposes: firstly, it has to retrieve high-relevance ads, which may exactly serve user's search intent; secondly, it needs to retrieve high-CTR ads so as to maximize the overall user clicks. In this paper, we present a novel representation learning framework Uni-Retriever developed for Bing Search, which unifies two different training modes knowledge distillation and contrastive learning to realize both required objectives. On one hand, the capability of making high-relevance retrieval is established by distilling knowledge from the "relevance teacher model''. On the other hand, the capability of making high-CTR retrieval is optimized by learning to discriminate user's clicked ads from the entire corpus. The two training modes are jointly performed as a multi-objective learning process, such that the ads of high relevance and CTR can be favored by the generated embeddings. Besides the learning strategy, we also elaborate our solution for EBR serving pipeline built upon the substantially optimized DiskANN, where massive-scale EBR can be performed with competitive time and memory efficiency, and accomplished in high-quality. We make comprehensive offline and online experiments to evaluate the proposed techniques, whose findings may provide useful insights for the future development of EBR systems. Uni-Retriever has been mainstreamed as the major retrieval path in Bing's production thanks to the notable improvements on the representation and EBR serving quality.

Recommender systems (RS) have to select the top-N items from a massive item set. For the sake of efficient recommendation, RS usually represents user and item as latent embeddings, and relies on approximate nearest neighbour search (ANNs) to retrieve the recommendation result. Despite the reduction of running time, the representation learning is independent of ANNs index construction; thus, the two operations can be incompatible, which results in potential loss of recommendation accuracy. To overcome the above problem, we propose the Recommender Forest (a.k.a., RecForest), which jointly learns latent embedding and index for efficient and high-fidelity recommendation. RecForest consists of multiple k-ary trees, each of which is a partition of the item set via hierarchical balanced clustering such that each item is uniquely represented by a path from the root to a leaf. Given such a data structure, an encoder-decoder based routing network is developed: it first encodes the context, i.e., user information, into hidden states; then, leveraging a transformer-based decoder, it identifies the top-N items via beam search. Compared with the existing methods, RecForest brings in the following advantages: 1) the false partition of the boundary items can be effectively alleviated by the use of multiple trees; 2) the routing operation becomes much more accurate thanks to the powerful transformer decoder; 3) the tree parameters are shared across different tree levels, making the index to be extremely memory-efficient. The experimental studies are performed on five popular recommendation datasets: with a significantly simplified training cost, RecForest outperforms competitive baseline approaches in terms of both recommendation accuracy and efficiency.

Current state-of-the-art document retrieval solutions mainly follow an index-retrieve paradigm, where the index is hard to be directly optimized for the final retrieval target. In this paper, we aim to show that an end-to-end deep neural network unifying training and indexing stages can significantly improve the recall performance of traditional methods. To this end, we propose Neural Corpus Indexer (NCI), a sequence-to-sequence network that generates relevant document identifiers directly for a designated query. To optimize the recall performance of NCI, we invent a prefix-aware weight-adaptive decoder architecture, and leverage tailored techniques including query generation, semantic document identifiers, and consistency-based regularization. Empirical studies demonstrated the superiority of NCI on two commonly used academic benchmarks, achieving +21.4% and +16.8% relative enhancement for Recall@1 on NQ320k dataset and R-Precision on TriviaQA dataset, respectively, compared to the best baseline method.

A large-scale recommender system usually consists of recall and ranking modules. The goal of ranking modules (aka rankers) is to elaborately discriminate users' preference on item candidates proposed by recall modules. With the success of deep learning techniques in various domains, we have witnessed the mainstream rankers evolve from traditional models to deep neural models. However, the way that we design and use rankers remains unchanged: offline training the model, freezing the parameters, and deploying it for online serving. Actually, the candidate items are determined by specific user requests, in which underlying distributions (e.g., the proportion of items for different categories, the proportion of popular or new items) are highly different from one another in a production environment. The classical parameter-frozen inference manner cannot adapt to dynamic serving circumstances, making rankers' performance compromised.

With the development of deep learning techniques, deep recommendation models also achieve remarkable improvements in terms of recommendation accuracy. However, due to the large number of candidate items in practice and the high cost of preference computation, these methods also suffer from low efficiency of recommendation. The recently proposed tree-based deep recommendation models alleviate the problem by directly learning tree structure and representations under the guidance of recommendation objectives. However, such models have two shortcomings. First, the max-heap assumption in the hierarchical tree, in which the preference for a parent node should be the maximum between the preferences for its children, is difficult to satisfy in their binary classification objectives. Second, the learned index only includes a single tree, which is different from the widely-used multiple trees index, providing an opportunity to improve the accuracy of recommendation.

Vector quantization (VQ) based ANN indexes, such as Inverted File System (IVF) and Product Quantization (PQ), have been widely applied to embedding based document retrieval thanks to the competitive time and memory efficiency. Originally, VQ is learned to minimize the reconstruction loss, i.e., the distortions between the original dense embeddings and the reconstructed embeddings after quantization. Unfortunately, such an objective is inconsistent with the goal of selecting ground-truth documents for the input query, which may cause severe loss of retrieval quality. Recent works identify such a defect, and propose to minimize the retrieval loss through contrastive learning. However, these methods intensively rely on queries with ground-truth documents, whose performance is limited by the insufficiency of labeled data. In this paper, we propose Distill-VQ, which unifies the learning of IVF and PQ within a knowledge distillation framework. In Distill-VQ, the dense embeddings are leveraged as "teachers'', which predict the query's relevance to the sampled documents. The VQ modules are treated as the "students'', which are learned to reproduce the predicted relevance, such that the reconstructed embeddings may fully preserve the retrieval result of the dense embeddings. By doing so, Distill-VQ is able to derive substantial training signals from the massive unlabeled data, which significantly contributes to the retrieval quality. We perform comprehensive explorations for the optimal conduct of knowledge distillation, which may provide useful insights for the learning of VQ based ANN index. We also experimentally show that the labeled data is no longer a necessity for high-quality vector quantization, which indicates Distill-VQ's strong applicability in practice. The evaluations are performed on MS MARCO and Natural Questions benchmarks, where Distill-VQ notably outperforms the SOTA VQ methods in Recall and MRR. Our code is avaliable at https://github.com/staoxiao/LibVQ.

Recommender system is playing an increasingly important role in online news platforms nowadays. Recently, there is a growing demand for applying reinforcement learning (RL) algorithms to news recommendation aiming to maximize long-term and/or non-differentiable objectives. However, without an interactive simulated environment, it is extremely costly to develop powerful RL agents for news recommendation. In this paper, we build a user simulator, namely MINDSim, for news recommendation. Targeting at new user generation and corresponding behavior simulation, we first construct a hidden space for users using a generative adversarial network, so that new users can be generated by sampling from this hidden space. To capture complex and fast user interest drifts over time, we adopt an encoder-decoder architecture, which takes the clicked news during the simulation as input and outputs the new user interests for the next period of time. Finally, we build the MINDSim simulator using MIcrosoft News Dataset (MIND), and extensive experimental results on this large-scale real-world dataset demonstrate that MINDSim can simulate the behaviors of real users with high quality.

Ad-hoc search calls for the selection of appropriate answers from a massive-scale corpus. Nowadays, the embedding-based retrieval (EBR) becomes a promising solution, where deep learning based document representation and ANN search techniques are allied to handle this task. However, a major challenge is that the ANN index can be too large to fit into memory, given the considerable size of answer corpus. In this work, we tackle this problem with Bi-Granular Document Representation, where the lightweight sparse embeddings are indexed and standby in memory for coarse-grained candidate search, and the heavyweight dense embeddings are hosted in disk for fine-grained post verification. For the best of retrieval accuracy, a Progressive Optimization framework is designed. The sparse embeddings are learned ahead for high-quality search of candidates. Conditioned on the candidate distribution induced by the sparse embeddings, the dense embeddings are continuously learned to optimize the discrimination of ground-truth from the shortlisted candidates. Besides, two techniques: the contrastive quantization and the locality-centric sampling are introduced for the learning of sparse and dense embeddings, which substantially contribute to their performances. Thanks to the above features, our method effectively handles massive-scale EBR with strong advantages in accuracy: with up to recall gain on million-scale corpus, and up to recall gain on billion-scale corpus. Besides, Our method is applied to a major sponsored search platform with substantial gains on revenue (), Recall () and CTR (). Our code is available at https://github.com/microsoft/BiDR.

Product quantization (PQ) is a widely used technique for ad-hoc retrieval.Recent studies propose supervised PQ, where the embedding and quantization models can be jointly trained with supervised learning.However, there is a lack of appropriate formulation of the joint training objective; thus, the improvements over previous non-supervised baselines are limited in reality.In this work, we propose the Matching-oriented Product Quantization (MoPQ), where a novel objective Multinoulli Contrastive Loss (MCL) is formulated.With the minimization of MCL, we are able to maximize the matching probability of query and ground-truth key, which contributes to the optimal retrieval accuracy.Given that the exact computation of MCL is intractable due to the demand of vast contrastive samples, we further propose the Differentiable Cross-device Sampling (DCS), which significantly augments the contrastive samples for precise approximation of MCL.We conduct extensive experimental studies on four realworld datasets, whose results verify the effectiveness of MoPQ.The code is available at https://github.com/microsoft/MoPQ.1. We adopt the Asymmetric Distance Computation (ADC) (Jégou et al., 2011), where only keys need to be quantized.

News recommendation systems play a key role in online news reading service. Knowledge graphs (KG), which contain comprehensive structural knowledge, are well known for their potential to enhance both accuracy and explainability. While existing works intensively study using KG to improve news recommendation accuracy, using KG for news recommendation reasoning has not been fully explored. A few works such as KPRN [18], [22] and ADAC [25] have discussed knowledge reasoning in some other recommendation domains such as music or movie, but their methods are not practical for the news. How to make reasoning scalable to generic KGs, easy to deploy for real-time serving and meanwhile elastic for both recall and ranking stages remains an open question.

The representation learning on textual graph is to generate low-dimensional embeddings for the nodes based on the individual textual features and the neighbourhood information. Recent breakthroughs on pretrained language models and graph neural networks push forward the development of corresponding techniques. The existing works mainly rely on the cascaded model architecture: the textual features of nodes are independently encoded by language models at first; the textual embeddings are aggregated by graph neural networks afterwards. However, the above architecture is limited due to the independent modeling of textual features. In this work, we propose GraphFormers, where layerwise GNN components are nested alongside the transformer blocks of language models. With the proposed architecture, the text encoding and the graph aggregation are fused into an iterative workflow, making each node's semantic accurately comprehended from the global perspective. In addition, a progressive learning strategy is introduced, where the model is successively trained on manipulated data and original data to reinforce its capability of integrating information on graph. Extensive evaluations are conducted on three large-scale benchmark datasets, where GraphFormers outperform the SOTA baselines with comparable running efficiency. The source code is released at https://github.com/microsoft/GraphFormers .

Self-attention networks (SANs) have been intensively applied for sequential recommenders, but they are limited due to: (1) the quadratic complexity and vulnerability to over-parameterization in self-attention; (2) inaccurate modeling of sequential relations between items due to the implicit position encoding. In this work, we propose the low-rank decomposed self-attention networks (LightSANs) to overcome these problems. Particularly, we introduce the low-rank decomposed self-attention, which projects user's historical items into a small constant number of latent interests and leverages item-to-interest interaction to generate the context-aware representation. It scales linearly w.r.t. the user's historical sequence length in terms of time and space, and is more resilient to over-parameterization. Besides, we design the decoupled position encoding, which models the sequential relations between items more precisely. Extensive experimental studies are carried out on three real-world datasets, where LightSANs outperform the existing SANs-based recommenders in terms of both effectiveness and efficiency.

Sponsored search ads appear next to search results when people look for products and services on search engines. In recent years, they have become one of the most lucrative channels for marketing. As the fundamental basis of search ads, relevance modeling has attracted increasing attention due to the significant research challenges and tremendous practical value. Most existing approaches solely rely on the semantic information in the input query-ad pair, while the pure semantic information in the short ads data is not sufficient to fully identify user's search intents. Our motivation lies in incorporating the tremendous amount of unsupervised user behavior data from the historical search logs as the complementary graph to facilitate relevance modeling. In this paper, we extensively investigate how to naturally fuse the semantic textual information with the user behavior graph, and further propose three novel AdsGNN models to aggregate topological neighborhood from the perspectives of nodes, edges and tokens. Furthermore, two critical but rarely investigated problems, domain-specific pre-training and long-tail ads matching, are studied thoroughly. Empirically, we evaluate the AdsGNN models over the large industry dataset, and the experimental results of online/offline tests consistently demonstrate the superiority of our proposal.

Personalized news recommendation is a critical technology to improve users' online news reading experience.The core of news recommendation is accurate matching between user's interests and candidate news.The same user usually has diverse interests that are reflected in different news she has browsed.Meanwhile, important semantic features of news are implied in text segments of different granularities.Existing studies generally represent each user as a single vector and then match the candidate news vector, which may lose fine-grained information for recommendation.In this paper, we propose FIM, a Finegrained Interest Matching method for neural news recommendation.Instead of aggregating user's all historical browsed news into a unified vector, we hierarchically construct multilevel representations for each news via stacked dilated convolutions.Then we perform finegrained matching between segment pairs of each browsed news and the candidate news at each semantic level.High-order salient signals are then identified by resembling the hierarchy of image recognition for final click prediction.Extensive experiments on a real-world dataset from MSN news validate the effectiveness of our model on news recommendation.

Recommendation techniques are important approaches for alleviating information overload. Being often trained on implicit user feedback, many recommenders suffer from the sparsity challenge due to the lack of explicitly negative samples. The GAN-style recommenders (i.e., IRGAN) addresses the challenge by learning a generator and a discriminator adversarially, such that the generator produces increasingly difficult samples for the discriminator to accelerate optimizing the discrimination objective. However, producing samples from the generator is very time-consuming, and our empirical study shows that the discriminator performs poor in top-k item recommendation. To this end, a theoretical analysis is made for the GAN-style algorithms, showing that the generator of limit capacity is diverged from the optimal generator. This may interpret the limitation of discriminator's performance. Based on these findings, we propose a Sampling-Decomposable Generative Adversarial Recommender (SD-GAR). In the framework, the divergence between some generator and the optimum is compensated by self-normalized importance sampling; the efficiency of sample generation is improved with a sampling-decomposable generator, such that each sample can be generated in O(1) with the Vose-Alias method. Interestingly, due to decomposability of sampling, the generator can be optimized with the closed-form solutions in an alternating manner, being different from policy gradient in the GAN-style algorithms. We extensively evaluate the proposed algorithm with five real-world recommendation datasets. The results show that SD-GAR outperforms IRGAN by 12.4% and the SOTA recommender by 10% on average. Moreover, discriminator training can be 20x faster on the dataset with more than 120K items.

Candidate generation is a critical task for recommendation system, which is technically challenging from two perspectives. On the one hand, recommendation system requires the comprehensive inclusion of user's interested candidates, yet typical deep user modeling approaches would represent each user as an onefold vector, which is hard to capture user's diverse interests. On the other hand, for the sake of practicability, the candidate generation process needs to be both accurate and efficient. Although existing "multi-channel structures'', like memory networks, are more capable of representing user's diverse interests, they may bring in substantial irrelevant candidates and lead to rapid growth of temporal cost. As a result, it remains a tough issue to comprehensively acquire user's interested items in a practical way.

Knowledge graphs have been widely adopted to improve recommendation accuracy. The multi-hop user-item connections on knowledge graphs also endow reasoning about why an item is recommended. However, reasoning on paths is a complex combinatorial optimization problem. Traditional recommendation methods usually adopt brute-force methods to find feasible paths, which results in issues related to convergence and explainability. In this paper, we address these issues by better supervising the path finding process. The key idea is to extract imperfect path demonstrations with minimum labeling efforts and effectively leverage these demonstrations to guide path finding. In particular, we design a demonstration-based knowledge graph reasoning framework for explainable recommendation. We also propose an ADversarial Actor-Critic (ADAC) model for the demonstration-guided path finding. Experiments on three real-world benchmarks show that our method converges more quickly than the state-of-the-art baseline and achieves better recommendation accuracy and explainability.

Deep recommender systems have achieved remarkable improvements in recent years. Despite its superior ranking precision, the running efficiency and memory consumption turn out to be severe bottlenecks in reality. To overcome both limitations, we propose LightRec, a lightweight recommender system which enjoys fast online inference and economic memory consumption. The backbone of LightRec is a total of B codebooks, each of which is composed of W latent vectors, known as codewords. On top of such a structure, LightRec will have an item represented as additive composition of B codewords, which are optimally selected from each of the codebooks. To effectively learn the codebooks from data, we devise an end-to-end learning workflow, where challenges on the inherent differentiability and diversity are conquered by the proposed techniques. In addition, to further improve the representation quality, several distillation strategies are employed, which better preserves user-item relevance scores and relative ranking orders. LightRec is extensively evaluated with four real-world datasets, which gives rise to two empirical findings: 1) compared with those the state-of-the-art lightweight baselines, LightRec achieves over 11% relative improvements in terms of recall performance; 2) compared to conventional recommendation algorithms, LightRec merely incurs negligible accuracy degradation while leads to more than 27x speedup in top-k recommendation.

Personalized news recommendation is important to help users find their interested news and improve reading experience.A key problem in news recommendation is learning accurate user representations to capture their interests.Users usually have both long-term preferences and short-term interests.However, existing news recommendation methods usually learn single representations of users, which may be insufficient.In this paper, we propose a neural news recommendation approach which can learn both long-and short-term user representations.The core of our approach is a news encoder and a user encoder.In the news encoder, we learn representations of news from their titles and topic categories, and use attention network to select important words.In the user encoder, we propose to learn long-term user representations from the embeddings of their IDs.In addition, we propose to learn short-term user representations from their recently browsed news via GRU network.Besides, we propose two methods to combine long-term and short-term user representations.The first one is using the long-term user representation to initialize the hidden state of the GRU network in short-term user representation.The second one is concatenating both long-and short-term user representations as a unified user vector.Extensive experiments on a real-world dataset show our approach can effectively improve the performance of neural news recommendation.

Deep learning techniques have been widely applied to modern recommendation systems, bringing in flexible and effective ways of user representation. Conventionally, user representations are generated purely in the offline stage. Without referencing to the specific candidate item for recommendation, it is difficult to fully capture user preference from the perspective of interest. More recent algorithms tend to generate user representation at runtime, where user's historical behaviors are attentively summarized w.r.t. the presented candidate item. In spite of the improved efficacy, it is too expensive for many real-world scenarios because of the repetitive access to user's entire history. In this work, a novel user representation framework, Hi-Fi Ark, is proposed. With Hi-Fi Ark, user history is summarized into highly compact and complementary vectors in the offline stage, known as archives. Meanwhile, user preference towards a specific candidate item can be precisely captured via the attentive aggregation of such archives. As a result, both deployment feasibility and superior recommendation efficacy are achieved by Hi-Fi Ark. The effectiveness of Hi-Fi Ark is empirically validated on three real-world datasets, where remarkable and consistent improvements are made over a variety of well-recognized baseline methods.

Realtime traffic speed estimation is an important issue in urban computation. Existing approaches usually focus on exploiting the periodicity properties of the traffic speed and utilize crowdsourcing techniques to facilitate real-time estimation. The quality of such estimation is limited in real world: 1) the accuracy of existing estimation over-relies on the probed data; 2) the accidental traffic variance is ignored; 3) existing strategies incur exhaustive usage of human workers to get fine-grained estimation results. Thus, a more intelligent RTSE approach is desired. In this paper, we propose the framework of CrowdRTSE (Crowdsourcing-based Real-time Traffic Speed Estimation), which adopts a hybrid offline-online process to collaboratively exploit the historical and real-time data to produce high-quality RTSE. To accomplish such a framework, we devise effective algorithms to judiciously select the best group of human workers with a constant approximation ratio, and effectively propagate the crowdsourced data with high efficiency. Comprehensive evaluations have been conducted on both synthetic and real world datasets. The experimental results verify the effectiveness and efficiency of our proposed methods.

Collaborator Recommendation is a useful application in exploiting big academic data. However, existing works leave out the contextual restriction (i.e., research topics) of people's academic collaboration, thus cannot recommend suitable collaborators for the required research topics. In this work, we propose Context-aware Collaborator Recommendation (CACR), which aims to recommend high-potential new collaborators for people's context-restricted requests. To this end, we design a novel recommendation framework, which consists of two fundamental components: the Collaborative Entity Embedding network (CEE) and the Hierarchical Factorization Model (HFM). In particular, CEE jointly represents researchers and research topics as compact vectors based on their co-occurrence relationships, whereby capturing researchers' context-aware collaboration tendencies and topics' underlying semantics. Meanwhile, HFM extracts researchers' activenesses and conservativenesses, which reflect their intensities of making academic collaborations and tendencies of working with non-collaborated fellows. The extracted activenesses and conservativenesses work collaboratively with the context-aware collaboration tendencies, such that high-quality recommendation can be produced. Extensive experimental studies are conducted with large-scale academic data, whose results verify the effectiveness of our proposed approaches.

As crowdsourcing has been dramatically investigated and utilized to address problems in the real world, it is essential and important to think about performance optimization. Analogous to computer systems with CPUs, treating each worker as a HPU (Human Processing Unit [1]) and studying the performance optimization on top of HPUs are interesting perspectives to resolve crowdsourcing issues. However, as we characterize HPUs in detail for this purpose, we find that there are significant differences between CPUs and HPUs, leading to the need of completely new optimization algorithms. In this paper, we study the specific optimization problem of obtaining results the fastest for a crowdsourced job with a fixed total budget. In crowdsourcing, jobs are usually broken down into sets of small tasks, which are assigned to workers one at a time. We consider three scenarios of increasing complexity: Identical Round Homogeneous Tasks, Multiplex Round Homogeneous Tasks, and Multiple Round Heterogeneous Tasks. For each scenario, we analyze the stochastic behavior of the HPU clock rate as a function of the remuneration offered. After that, we develop an optimum Budget Allocation Strategy to minimize the latency of the job completion. We validate our results through extensive simulations and experiments on Amazon Mechanical Turk.

Worker Recommendation (WR) is one of the most important functions for crowdsourced Q&A services. Specifically, given a set of tasks to be solved, WR recommends each task with a certain group of workers, whom are expected to give timely answers with high qualities. To address the WR problem, recent studies have introduced a number of recommendation approaches, which take advantage of workers' expertises or preferences towards different types of tasks. However, without a thorough consideration of workers' characters, such approaches will lead to either inadequate task fulfillment or inferior answer quality. In this work, we propose the Triple-factor Aware Worker Recommendation framework, which collectively considers workers' expertises, preferences and activenesses to maximize the overall production of high quality answers. We construct the Latent Hierarchical Factorization Model, which is able to infer the tasks' underlying categories and workers' latent characters from the historical data; and we propose a novel parameter inference method, which only requires the processing of positive instances, giving rise to significantly higher time efficiency and better inference quality. What's more, the sampling-based recommendation algorithm is developed, such that the near optimal worker recommendation can be generated for a presented batch of tasks with considerably reduced time consumption. Comprehensive experiments have been carried out using both real and synthetic datasets, whose results verify the effectiveness and efficiency of our proposed methods.

As one of the three major steps (question design, task assignment, answer aggregation) in crowdsourcing, task assignment directly affects the quality of the crowdsourcing result. A good assignment will not only improve the answers' quality, but also boost the workers' willingness to participate. Although a lot of works have been made to produce better assignment, most of them neglected one of its most important properties: the bipartition, which exists widely in real world scenarios. Such ignorance greatly limits their application under general settings.

Uncertain data has been emerged as an important problem in database systems due to the imprecise nature of many applications. To handle the uncertainty, probabilistic databases can be used to store uncertain data, and querying facilities are provided to yield answers with confidence. However, the uncertainty may propagate, hence the returned results from a query or mining process may not be useful. In this paper, we leverage the power of crowdsourcing for cleaning uncertain data. Specifically, we will design a set of Human Intelligence Tasks (HIT)s to ask a crowd to improve the quality of uncertain data. Each HIT is associated with a cost, thus, we need to design solutions to maximize the data quality with minimal number of HITs. There are two obstacles for this non-trivial optimization - first, the crowd has a probability to return incorrect answers; second, the HITs decomposed from uncertain data are often correlated. These two obstacles lead to very high computational cost for selecting the optimal set of HITs. Thus, in this paper, we have addressed these challenges by designing an effective approximation algorithm and an efficient heuristic solution. To further improve the efficiency, we derive tight lower and upper bounds, which are used for effective filtering and estimation. We have verified the solutions with extensive experiments on both a simulated crowd and a real crowdsourcing platform.

As one of the successful forms of using Wisdom of Crowd, crowdsourcing, has been widely used for many human intrinsic tasks, such as image labeling, natural language understanding, market predication and opinion mining. Meanwhile, with advances in pervasive technology, mobile devices, such as mobile phones and tablets, have become extremely popular. These mobile devices can work as sensors to collect multimedia data(audios, images and videos) and location information. This power makes it possible to implement the new crowdsourcing mode: spatial crowdsourcing. In spatial crowdsourcing, a requester can ask for resources related a specific location, the mobile users who would like to take the task will travel to that place and get the data. Due to the rapid growth of mobile device uses, spatial crowdsourcing is likely to become more popular than general crowdsourcing, such as Amazon Turk and Crowdflower. However, to implement such a platform, effective and efficient solutions for worker incentives, task assignment, result aggregation and data quality control must be developed. In this demo, we will introduce gMission, a general spatial crowdsourcing platform, which features with a collection of novel techniques, including geographic sensing, worker detection, and task recommendation. We introduce the sketch of system architecture and illustrate scenarios via several case analysis.