Shaoping Ma

Jingtao Zhan, Qingyao Ai, Yiqun Liu et al. · tsinghua

Recent advance in Dense Retrieval (DR) techniques has significantly improved the effectiveness of first-stage retrieval. Trained with large-scale supervised data, DR models can encode queries and documents into a low-dimensional dense space and conduct effective semantic matching. However, previous studies have shown that the effectiveness of DR models would drop by a large margin when the trained DR models are adopted in a target domain that is different from the domain of the labeled data. One of the possible reasons is that the DR model has never seen the target corpus and thus might be incapable of mitigating the difference between the training and target domains. In practice, unfortunately, training a DR model for each target domain to avoid domain shift is often a difficult task as it requires additional time, storage, and domain-specific data labeling, which are not always available. To address this problem, in this paper, we propose a novel DR framework named Disentangled Dense Retrieval (DDR) to support effective and flexible domain adaptation for DR models. DDR consists of a Relevance Estimation Module (REM) for modeling domain-invariant matching patterns and several Domain Adaption Modules (DAMs) for modeling domain-specific features of multiple target corpora. By making the REM and DAMs disentangled, DDR enables a flexible training paradigm in which REM is trained with supervision once and DAMs are trained with unsupervised data. Comprehensive experiments in different domains and languages show that DDR significantly improves ranking performance compared to strong DR baselines and substantially outperforms traditional retrieval methods in most scenarios.

Yunqiu Shao, Haitao Li, Yueyue Wu et al. · tsinghua

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

Weihang Su, Xiangsheng Li, Yiqun Liu et al.

Our team(THUIR2) participated in both FOSS and POSS subtasks of the NTCIR-161 Session Search (SS) Task. This paper describes our approaches and results. In the FOSS subtask, we submit five runs using learning-to-rank and fine-tuned pre-trained language models. We fine-tuned the pre-trained language model with ad-hoc data and session information and assembled them by a learning-to-rank method. The assembled model achieves the best performance among all participants in the preliminary evaluation. In the POSS subtask, we used an assembled model which also achieves the best performance in the preliminary evaluation.

Yangkun Li, Weizhi Ma, Chong Chen et al.

Overfitting is a common problem in machine learning, which means the model too closely fits the training data while performing poorly in the test data. Among various methods of coping with overfitting, dropout is one of the representative ways. From randomly dropping neurons to dropping neural structures, dropout has achieved great success in improving model performances. Although various dropout methods have been designed and widely applied in past years, their effectiveness, application scenarios, and contributions have not been comprehensively summarized and empirically compared by far. It is the right time to make a comprehensive survey. In this paper, we systematically review previous dropout methods and classify them into three major categories according to the stage where dropout operation is performed. Specifically, more than seventy dropout methods published in top AI conferences or journals (e.g., TKDE, KDD, TheWebConf, SIGIR) are involved. The designed taxonomy is easy to understand and capable of including new dropout methods. Then, we further discuss their application scenarios, connections, and contributions. To verify the effectiveness of distinct dropout methods, extensive experiments are conducted on recommendation scenarios with abundant heterogeneous information. Finally, we propose some open problems and potential research directions about dropout that worth to be further explored.

Changyue Wang, Weihang Su, Hu Yiran et al.

Recent advances in Large Language Models (LLMs) have significantly shaped the applications of AI in multiple fields, including the studies of legal intelligence. Trained on extensive legal texts, including statutes and legal documents, the legal LLMs can capture important legal knowledge/concepts effectively and provide important support for downstream legal applications such as legal consultancy. Yet, the dynamic nature of legal statutes and interpretations also poses new challenges to the use of LLMs in legal applications. Particularly, how to update the legal knowledge of LLMs effectively and efficiently has become an important research problem in practice. Existing benchmarks for evaluating knowledge update methods are mostly designed for the open domain and cannot address the specific challenges of the legal domain, such as the nuanced application of new legal knowledge, the complexity and lengthiness of legal regulations, and the intricate nature of legal reasoning. To address this gap, we introduce the Legal Knowledge Update BEnchmark, i.e. LeKUBE, which evaluates knowledge update methods for legal LLMs across five dimensions. Specifically, we categorize the needs of knowledge updates in the legal domain with the help of legal professionals, and then hire annotators from law schools to create synthetic updates to the Chinese Criminal and Civil Code as well as sets of questions of which the answers would change after the updates. Through a comprehensive evaluation of state-of-the-art knowledge update methods, we reveal a notable gap between existing knowledge update methods and the unique needs of the legal domain, emphasizing the need for further research and development of knowledge update mechanisms tailored for legal LLMs.

Weizhi Ma, Min Zhang, Yue Cao et al.

Explainability and effectiveness are two key aspects for building recommender systems. Prior efforts mostly focus on incorporating side information to achieve better recommendation performance. However, these methods have some weaknesses: (1) prediction of neural network-based embedding methods are hard to explain and debug; (2) symbolic, graph-based approaches (e.g., meta path-based models) require manual efforts and domain knowledge to define patterns and rules, and ignore the item association types (e.g. substitutable and complementary). In this paper, we propose a novel joint learning framework to integrate \textit{induction of explainable rules from knowledge graph} with \textit{construction of a rule-guided neural recommendation model}. The framework encourages two modules to complement each other in generating effective and explainable recommendation: 1) inductive rules, mined from item-centric knowledge graphs, summarize common multi-hop relational patterns for inferring different item associations and provide human-readable explanation for model prediction; 2) recommendation module can be augmented by induced rules and thus have better generalization ability dealing with the cold-start issue. Extensive experiments\footnote{Code and data can be found at: \url{https://github.com/THUIR/RuleRec}} show that our proposed method has achieved significant improvements in item recommendation over baselines on real-world datasets. Our model demonstrates robust performance over "noisy" item knowledge graphs, generated by linking item names to related entities.

Yucheng Wu, Yuekui Yang, Hongzheng Li et al.

Deploying new architectures in large-scale user response prediction systems incurs high model switching costs due to expensive retraining on massive historical data and performance degradation under data retention constraints. Existing knowledge distillation methods struggle with architectural heterogeneity and the prohibitive cost of transferring large embedding tables. We propose CrossAdapt, a two-stage framework for efficient cross-architecture knowledge transfer. The offline stage enables rapid embedding transfer via dimension-adaptive projections without iterative training, combined with progressive network distillation and strategic sampling to reduce computational cost. The online stage introduces asymmetric co-distillation, where students update frequently while teachers update infrequently, together with a distribution-aware adaptation mechanism that dynamically balances historical knowledge preservation and fast adaptation to evolving data. Experiments on three public datasets show that CrossAdapt achieves 0.27-0.43% AUC improvements while reducing training time by 43-71%. Large-scale deployment on Tencent WeChat Channels (~10M daily samples) further demonstrates its effectiveness, significantly mitigating AUC degradation, LogLoss increase, and prediction bias compared to standard distillation baselines.

Yifan Wang, Peijie Sun, Weizhi Ma et al.

Fairness of recommender systems (RS) has attracted increasing attention recently. Based on the involved stakeholders, the fairness of RS can be divided into user fairness, item fairness, and two-sided fairness which considers both user and item fairness simultaneously. However, we argue that the intersectional two-sided unfairness may still exist even if the RS is two-sided fair, which is observed and shown by empirical studies on real-world data in this paper, and has not been well-studied previously. To mitigate this problem, we propose a novel approach called Intersectional Two-sided Fairness Recommendation (ITFR). Our method utilizes a sharpness-aware loss to perceive disadvantaged groups, and then uses collaborative loss balance to develop consistent distinguishing abilities for different intersectional groups. Additionally, predicted score normalization is leveraged to align positive predicted scores to fairly treat positives in different intersectional groups. Extensive experiments and analyses on three public datasets show that our proposed approach effectively alleviates the intersectional two-sided unfairness and consistently outperforms previous state-of-the-art methods.

Ruizhe Zhang, Haitao Li, Yueyue Wu et al.

In recent years, the utilization of large language models for natural language dialogue has gained momentum, leading to their widespread adoption across various domains. However, their universal competence in addressing challenges specific to specialized fields such as law remains a subject of scrutiny. The incorporation of legal ethics into the model has been overlooked by researchers. We asserts that rigorous ethic evaluation is essential to ensure the effective integration of large language models in legal domains, emphasizing the need to assess domain-specific proficiency and domain-specific ethic. To address this, we propose a novelty evaluation methodology, utilizing authentic legal cases to evaluate the fundamental language abilities, specialized legal knowledge and legal robustness of large language models (LLMs). The findings from our comprehensive evaluation contribute significantly to the academic discourse surrounding the suitability and performance of large language models in legal domains.

Jingtao Zhan, Qingyao Ai, Yiqun Liu et al. · tsinghua

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

Deyi Ji, Yuekui Yang, Haiyang Wu et al.

Advertisement (Ad) video violation detection is critical for ensuring platform compliance, but existing methods struggle with precise temporal grounding, noisy annotations, and limited generalization. We propose RAVEN, a novel framework that integrates curriculum reinforcement learning with multimodal large language models (MLLMs) to enhance reasoning and cognitive capabilities for violation detection. RAVEN employs a progressive training strategy, combining precisely and coarsely annotated data, and leverages Group Relative Policy Optimization (GRPO) to develop emergent reasoning abilities without explicit reasoning annotations. Multiple hierarchical sophisticated reward mechanism ensures precise temporal grounding and consistent category prediction. Experiments on industrial datasets and public benchmarks show that RAVEN achieves superior performances in violation category accuracy and temporal interval localization. We also design a pipeline to deploy the RAVEN on the online Ad services, and online A/B testing further validates its practical applicability, with significant improvements in precision and recall. RAVEN also demonstrates strong generalization, mitigating the catastrophic forgetting issue associated with supervised fine-tuning.

Yichen Tang, Weihang Su, Yujia Zhou et al.

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

Jingtao Zhan, Jiahao Zhao, Jiayu Li et al. · tsinghua

Intelligence is a crucial trait for species to find solutions within a limited number of trial-and-error attempts. Building on this idea, we introduce Survival Game as a framework to evaluate intelligence based on the number of failed attempts in a trial-and-error process. Fewer failures indicate higher intelligence. When the expectation and variance of failure counts are both finite, it signals the ability to consistently find solutions to new challenges, which we define as the Autonomous Level of intelligence. Using Survival Game, we comprehensively evaluate existing AI systems. Our results show that while AI systems achieve the Autonomous Level in simple tasks, they are still far from it in more complex tasks, such as vision, search, recommendation, and language. While scaling current AI technologies might help, this would come at an astronomical cost. Projections suggest that achieving the Autonomous Level for general tasks would require $10^{26}$ parameters. To put this into perspective, loading such a massive model requires so many H100 GPUs that their total value is $10^{7}$ times that of Apple Inc.'s market value. Even with Moore's Law, supporting such a parameter scale would take $70$ years. This staggering cost highlights the complexity of human tasks and the inadequacies of current AI technologies. To further investigate this phenomenon, we conduct a theoretical analysis of Survival Game and its experimental results. Our findings suggest that human tasks possess a criticality property. As a result, Autonomous Level requires a deep understanding of the task's underlying mechanisms. Current AI systems, however, do not fully grasp these mechanisms and instead rely on superficial mimicry, making it difficult for them to reach an autonomous level. We believe Survival Game can not only guide the future development of AI but also offer profound insights into human intelligence.

Deyi Ji, Yuekui Yang, Liqun Liu et al.

Advertising (Ad) is a cornerstone of the digital economy, yet the moderation of video advertisements remains a significant challenge due to their complexity and the need for precise violation localization. While recent advancements, such as the RAVEN model, have improved coarse-grained violation detection, critical gaps persist in fine-grained understanding, explainability, and generalization. To address these limitations, we propose RAVEN++, a novel framework that introduces three key innovations: 1) Active Reinforcement Learning (RL), which dynamically adapts training to samples of varying difficulty; 2) Fine-Grained Violation Understanding, achieved through hierarchical reward functions and reasoning distillation; and 3) Progressive Multi-Stage Training, which systematically combines knowledge injection, curriculum-based passive RL, and active RL. Extensive experiments on both public and proprietary datasets, on both offline scenarios and online deployed A/B Testing, demonstrate that RAVEN++ outperforms general-purpose LLMs and specialized models like RAVEN in terms of fine-grained violation understanding, reasoning capabilities, and generalization ability.

He Wei, Yuekui Yang, Yang Zhang et al.

Deep Learning Recommendation Model(DLRM)s utilize the embedding layer to represent various categorical features. Traditional DLRMs adopt unified embedding size for all features, leading to suboptimal performance and redundant parameters. Thus, lots of Automatic Embedding size Search (AES) works focus on obtaining mixed embedding sizes with strong model performance. However, previous AES works can hardly address several challenges together: (1) The search results of embedding sizes are unstable; (2) Recommendation effect with AES results is unsatisfactory; (3) Memory cost of embeddings is uncontrollable. To address these challenges, we propose a novel one-shot AES framework called AdaS&S, in which a supernet encompassing various candidate embeddings is built and AES is performed as searching network architectures within it. Our framework contains two main stages: In the first stage, we decouple training parameters from searching embedding sizes, and propose the Adaptive Sampling method to yield a well-trained supernet, which further helps to produce stable AES results. In the second stage, to obtain embedding sizes that benefits the model effect, we design a reinforcement learning search process which utilizes the supernet trained previously. Meanwhile, to adapt searching to specific resource constraint, we introduce the resource competition penalty to balance the model effectiveness and memory cost of embeddings. We conduct extensive experiments on public datasets to show the superiority of AdaS&S. Our method could improve AUC by about 0.3% while saving about 20% of model parameters. Empirical analysis also shows that the stability of searching results in AdaS&S significantly exceeds other methods.

Junjie Chen, Weihang Su, Zhumin Chu et al.

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

Jingtao Zhan, Qingyao Ai, Yiqun Liu et al.

For text-to-image generation, automatically refining user-provided natural language prompts into the keyword-enriched prompts favored by systems is essential for the user experience. Such a prompt refinement process is analogous to translating the prompt from "user languages" into "system languages". However, the scarcity of such parallel corpora makes it difficult to train a prompt refinement model. Inspired by zero-shot machine translation techniques, we introduce Prompt Refinement with Image Pivot (PRIP). PRIP innovatively uses the latent representation of a user-preferred image as an intermediary "pivot" between the user and system languages. It decomposes the refinement process into two data-rich tasks: inferring representations of user-preferred images from user languages and subsequently translating image representations into system languages. Thus, it can leverage abundant data for training. Extensive experiments show that PRIP substantially outperforms a wide range of baselines and effectively transfers to unseen systems in a zero-shot manner.

Jingtao Zhan, Jiaxin Mao, Yiqun Liu et al.

Dense Retrieval (DR) reaches state-of-the-art results in first-stage retrieval, but little is known about the mechanisms that contribute to its success. Therefore, in this work, we conduct an interpretation study of recently proposed DR models. Specifically, we first discretize the embeddings output by the document and query encoders. Based on the discrete representations, we analyze the attribution of input tokens. Both qualitative and quantitative experiments are carried out on public test collections. Results suggest that DR models pay attention to different aspects of input and extract various high-level topic representations. Therefore, we can regard the representations learned by DR models as a mixture of high-level topics.

Xuesong Chen, Ziyi Ye, Xiaohui Xie et al.

While search technologies have evolved to be robust and ubiquitous, the fundamental interaction paradigm has remained relatively stable for decades. With the maturity of the Brain-Machine Interface, we build an efficient and effective communication system between human beings and search engines based on electroencephalogram(EEG) signals, called Brain-Machine Search Interface(BMSI) system. The BMSI system provides functions including query reformulation and search result interaction. In our system, users can perform search tasks without having to use the mouse and keyboard. Therefore, it is useful for application scenarios in which hand-based interactions are infeasible, e.g, for users with severe neuromuscular disorders. Besides, based on brain signals decoding, our system can provide abundant and valuable user-side context information(e.g., real-time satisfaction feedback, extensive context information, and a clearer description of information needs) to the search engine, which is hard to capture in the previous paradigm. In our implementation, the system can decode user satisfaction from brain signals in real-time during the interaction process and re-rank the search results list based on user satisfaction feedback. The demo video is available at http://www.thuir.cn/group/YQLiu/datasets/BMSISystem.mp4.

Jingtao Zhan, Jiaxin Mao, Yiqun Liu et al.

Dense Retrieval (DR) has achieved state-of-the-art first-stage ranking effectiveness. However, the efficiency of most existing DR models is limited by the large memory cost of storing dense vectors and the time-consuming nearest neighbor search (NNS) in vector space. Therefore, we present RepCONC, a novel retrieval model that learns discrete Representations via CONstrained Clustering. RepCONC jointly trains dual-encoders and the Product Quantization (PQ) method to learn discrete document representations and enables fast approximate NNS with compact indexes. It models quantization as a constrained clustering process, which requires the document embeddings to be uniformly clustered around the quantization centroids and supports end-to-end optimization of the quantization method and dual-encoders. We theoretically demonstrate the importance of the uniform clustering constraint in RepCONC and derive an efficient approximate solution for constrained clustering by reducing it to an instance of the optimal transport problem. Besides constrained clustering, RepCONC further adopts a vector-based inverted file system (IVF) to support highly efficient vector search on CPUs. Extensive experiments on two popular ad-hoc retrieval benchmarks show that RepCONC achieves better ranking effectiveness than competitive vector quantization baselines under different compression ratio settings. It also substantially outperforms a wide range of existing retrieval models in terms of retrieval effectiveness, memory efficiency, and time efficiency.

Ziyi Ye, Xiaohui Xie, Yiqun Liu et al.

Web search heavily relies on click-through behavior as an essential feedback signal for performance improvement and evaluation. Traditionally, click is usually treated as a positive implicit feedback signal of relevance or usefulness, while non-click (especially non-click after examination) is regarded as a signal of irrelevance or uselessness. However, there are many cases where users do not click on any search results but still satisfy their information need with the contents of the results shown on the Search Engine Result Page (SERP). This raises the problem of measuring result usefulness and modeling user satisfaction in "Zero-click" search scenarios. Previous works have solved this issue by (1) detecting user satisfaction for abandoned SERP with context information and (2) considering result-level click necessity with external assessors' annotations. However, few works have investigated the reason behind non-click behavior and estimated the usefulness of non-click results. A challenge for this research question is how to collect valuable feedback for non-click results. With neuroimaging technologies, we design a lab-based user study and reveal differences in brain signals while examining non-click search results with different usefulness levels. The findings in significant brain regions and electroencephalogram~(EEG) spectrum also suggest that the process of usefulness judgment might involve similar cognitive functions of relevance perception and satisfaction decoding. Inspired by these findings, we conduct supervised learning tasks to estimate the usefulness of non-click results with brain signals and conventional information (i.e., content and context factors). Results show that it is feasible to utilize brain signals to improve usefulness estimation performance and enhancing human-computer interactions in "Zero-click" search scenarios.

Ziyi Ye, Xiaohui Xie, Yiqun Liu et al.

Reading comprehension is a complex cognitive process involving many human brain activities. Plenty of works have studied the patterns and attention allocations of reading comprehension in information retrieval related scenarios. However, little is known about what happens in human brain during reading comprehension and how these cognitive activities can affect information retrieval process. Additionally, with the advances in brain imaging techniques such as electroencephalogram (EEG), it is possible to collect brain signals in almost real time and explore whether it can be utilized as feedback to facilitate information acquisition performance. In this paper, we carefully design a lab-based user study to investigate brain activities during reading comprehension. Our findings show that neural responses vary with different types of reading contents, i.e., contents that can satisfy users' information needs and contents that cannot. We suggest that various cognitive activities, e.g., cognitive loading, semantic-thematic understanding, and inferential processing, underpin these neural responses at the micro-time scale during reading comprehension. From these findings, we illustrate several insights for information retrieval tasks, such as ranking models construction and interface design. Besides, we suggest the possibility of detecting reading comprehension status for a proactive real-world system. To this end, we propose a Unified framework for EEG-based Reading Comprehension Modeling (UERCM). To verify its effectiveness, we conduct extensive experiments based on EEG features for two reading comprehension tasks: answer sentence classification and answer extraction. Results show that it is feasible to improve the performance of two tasks with brain signals.

Jingtao Zhan, Jiaxin Mao, Yiqun Liu et al.

Recently, Information Retrieval community has witnessed fast-paced advances in Dense Retrieval (DR), which performs first-stage retrieval with embedding-based search. Despite the impressive ranking performance, previous studies usually adopt brute-force search to acquire candidates, which is prohibitive in practical Web search scenarios due to its tremendous memory usage and time cost. To overcome these problems, vector compression methods have been adopted in many practical embedding-based retrieval applications. One of the most popular methods is Product Quantization (PQ). However, although existing vector compression methods including PQ can help improve the efficiency of DR, they incur severely decayed retrieval performance due to the separation between encoding and compression. To tackle this problem, we present JPQ, which stands for Joint optimization of query encoding and Product Quantization. It trains the query encoder and PQ index jointly in an end-to-end manner based on three optimization strategies, namely ranking-oriented loss, PQ centroid optimization, and end-to-end negative sampling. We evaluate JPQ on two publicly available retrieval benchmarks. Experimental results show that JPQ significantly outperforms popular vector compression methods. Compared with previous DR models that use brute-force search, JPQ almost matches the best retrieval performance with 30x compression on index size. The compressed index further brings 10x speedup on CPU and 2x speedup on GPU in query latency.

Bin Hao, Min Zhang, Weizhi Ma et al.

Data plays a vital role in machine learning studies. In the research of recommendation, both user behaviors and side information are helpful to model users. So, large-scale real scenario datasets with abundant user behaviors will contribute a lot. However, it is not easy to get such datasets as most of them are only hold and protected by companies. In this paper, a new large-scale dataset collected from a knowledge-sharing platform is presented, which is composed of around 100M interactions collected within 10 days, 798K users, 165K questions, 554K answers, 240K authors, 70K topics, and more than 501K user query keywords. There are also descriptions of users, answers, questions, authors, and topics, which are anonymous. Note that each user's latest query keywords have not been included in previous open datasets, which reveal users' explicit information needs. We characterize the dataset and demonstrate its potential applications for recommendation study. Multiple experiments show the dataset can be used to evaluate algorithms in general top-N recommendation, sequential recommendation, and context-aware recommendation. This dataset can also be used to integrate search and recommendation and recommendation with negative feedback. Besides, tasks beyond recommendation, such as user gender prediction, most valuable answerer identification, and high-quality answer recognition, can also use this dataset. To the best of our knowledge, this is the largest real-world interaction dataset for personalized recommendation.

Jingtao Zhan, Jiaxin Mao, Yiqun Liu et al.

Ranking has always been one of the top concerns in information retrieval researches. For decades, the lexical matching signal has dominated the ad-hoc retrieval process, but solely using this signal in retrieval may cause the vocabulary mismatch problem. In recent years, with the development of representation learning techniques, many researchers turn to Dense Retrieval (DR) models for better ranking performance. Although several existing DR models have already obtained promising results, their performance improvement heavily relies on the sampling of training examples. Many effective sampling strategies are not efficient enough for practical usage, and for most of them, there still lacks theoretical analysis in how and why performance improvement happens. To shed light on these research questions, we theoretically investigate different training strategies for DR models and try to explain why hard negative sampling performs better than random sampling. Through the analysis, we also find that there are many potential risks in static hard negative sampling, which is employed by many existing training methods. Therefore, we propose two training strategies named a Stable Training Algorithm for dense Retrieval (STAR) and a query-side training Algorithm for Directly Optimizing Ranking pErformance (ADORE), respectively. STAR improves the stability of DR training process by introducing random negatives. ADORE replaces the widely-adopted static hard negative sampling method with a dynamic one to directly optimize the ranking performance. Experimental results on two publicly available retrieval benchmark datasets show that either strategy gains significant improvements over existing competitive baselines and a combination of them leads to the best performance.

Yunqiu Shao, Bulou Liu, Jiaxin Mao et al.

In this paper, we present our methodologies for tackling the challenges of legal case retrieval and entailment in the Competition on Legal Information Extraction / Entailment 2020 (COLIEE-2020). We participated in the two case law tasks, i.e., the legal case retrieval task and the legal case entailment task. Task 1 (the retrieval task) aims to automatically identify supporting cases from the case law corpus given a new case, and Task 2 (the entailment task) to identify specific paragraphs that entail the decision of a new case in a relevant case. In both tasks, we employed the neural models for semantic understanding and the traditional retrieval models for exact matching. As a result, our team (TLIR) ranked 2nd among all of the teams in Task 1 and 3rd among teams in Task 2. Experimental results suggest that combing models of semantic understanding and exact matching benefits the legal case retrieval task while the legal case entailment task relies more on semantic understanding.

Jingtao Zhan, Jiaxin Mao, Yiqun Liu et al.

Ranking has always been one of the top concerns in information retrieval research. For decades, lexical matching signal has dominated the ad-hoc retrieval process, but it also has inherent defects, such as the vocabulary mismatch problem. Recently, Dense Retrieval (DR) technique has been proposed to alleviate these limitations by capturing the deep semantic relationship between queries and documents. The training of most existing Dense Retrieval models relies on sampling negative instances from the corpus to optimize a pairwise loss function. Through investigation, we find that this kind of training strategy is biased and fails to optimize full retrieval performance effectively and efficiently. To solve this problem, we propose a Learning To Retrieve (LTRe) training technique. LTRe constructs the document index beforehand. At each training iteration, it performs full retrieval without negative sampling and then updates the query representation model parameters. Through this process, it teaches the DR model how to retrieve relevant documents from the entire corpus instead of how to rerank a potentially biased sample of documents. Experiments in both passage retrieval and document retrieval tasks show that: 1) in terms of effectiveness, LTRe significantly outperforms all competitive sparse and dense baselines. It even gains better performance than the BM25-BERT cascade system under reasonable latency constraints. 2) in terms of training efficiency, compared with the previous state-of-the-art DR method, LTRe provides more than 170x speed-up in the training process. Training with a compressed index further saves computing resources with minor performance loss.

Jingtao Zhan, Jiaxin Mao, Yiqun Liu et al.

Although exact term match between queries and documents is the dominant method to perform first-stage retrieval, we propose a different approach, called RepBERT, to represent documents and queries with fixed-length contextualized embeddings. The inner products of query and document embeddings are regarded as relevance scores. On MS MARCO Passage Ranking task, RepBERT achieves state-of-the-art results among all initial retrieval techniques. And its efficiency is comparable to bag-of-words methods.

Xiaohui Xie, Jiaxin Mao, Maarten de Rijke et al.

User interaction behavior is a valuable source of implicit relevance feedback. In Web image search a different type of search result presentation is used than in general Web search, which leads to different interaction mechanisms and user behavior. For example, image search results are self-contained, so that users do not need to click the results to view the landing page as in general Web search, which generates sparse click data. Also, two-dimensional result placement instead of a linear result list makes browsing behaviors more complex. Thus, it is hard to apply standard user behavior models (e.g., click models) developed for general Web search to Web image search. In this paper, we conduct a comprehensive image search user behavior analysis using data from a lab-based user study as well as data from a commercial search log. We then propose a novel interaction behavior model, called grid-based user browsing model (GUBM), whose design is motivated by observations from our data analysis. GUBM can both capture users' interaction behavior, including cursor hovering, and alleviate position bias. The advantages of GUBM are two-fold: (1) It is based on an unsupervised learning method and does not need manually annotated data for training. (2) It is based on user interaction features on search engine result pages (SERPs) and is easily transferable to other scenarios that have a grid-based interface such as video search engines. We conduct extensive experiments to test the performance of our model using a large-scale commercial image search log. Experimental results show that in terms of behavior prediction (perplexity), and topical relevance and image quality (normalized discounted cumulative gain (NDCG)), GUBM outperforms state-of-the-art baseline models as well as the original ranking. We make the implementation of GUBM and related datasets publicly available for future studies.

Xiaohui Xie, Yiqun Liu, Maarten de Rijke et al.

What are the intents or goals behind human interactions with image search engines? Knowing why people search for images is of major concern to Web image search engines because user satisfaction may vary as intent varies. Previous analyses of image search behavior have mostly been query-based, focusing on what images people search for, rather than intent-based, that is, why people search for images. To date, there is no thorough investigation of how different image search intents affect users' search behavior. In this paper, we address the following questions: (1)Why do people search for images in text-based Web image search systems? (2)How does image search behavior change with user intent? (3)Can we predict user intent effectively from interactions during the early stages of a search session? To this end, we conduct both a lab-based user study and a commercial search log analysis. We show that user intents in image search can be grouped into three classes: Explore/Learn, Entertain, and Locate/Acquire. Our lab-based user study reveals different user behavior patterns under these three intents, such as first click time, query reformulation, dwell time and mouse movement on the result page. Based on user interaction features during the early stages of an image search session, that is, before mouse scroll, we develop an intent classifier that is able to achieve promising results for classifying intents into our three intent classes. Given that all features can be obtained online and unobtrusively, the predicted intents can provide guidance for choosing ranking methods immediately after scrolling.

Yongfeng Zhang, Min Zhang, Yiqun Liu et al.

Sentiment analysis on user reviews helps to keep track of user reactions towards products, and make advices to users about what to buy. State-of-the-art review-level sentiment classification techniques could give pretty good precisions of above 90%. However, current phrase-level sentiment analysis approaches might only give sentiment polarity labelling precisions of around 70%~80%, which is far from satisfaction and restricts its application in many practical tasks. In this paper, we focus on the problem of phrase-level sentiment polarity labelling and attempt to bridge the gap between phrase-level and review-level sentiment analysis. We investigate the inconsistency between the numerical star ratings and the sentiment orientation of textual user reviews. Although they have long been treated as identical, which serves as a basic assumption in previous work, we find that this assumption is not necessarily true. We further propose to leverage the results of review-level sentiment classification to boost the performance of phrase-level polarity labelling using a novel constrained convex optimization framework. Besides, the framework is capable of integrating various kinds of information sources and heuristics, while giving the global optimal solution due to its convexity. Experimental results on both English and Chinese reviews show that our framework achieves high labelling precisions of up to 89%, which is a significant improvement from current approaches.