Large language models have shown strong fluency in scientific writing, yet the evaluation of related work generation (RWG) remains limited. Existing RWG evaluations largely inherit summarization-oriented metrics, using lexical or semantic similarity to reference sections as proxies for quality. However, related work writing is fundamentally a citation-level scholarly positioning task: it requires selecting, organizing, and framing prior work to clarify how a target paper relates to, differs from, and contributes beyond existing research.As a result, models may generate coherent and semantically-relevant text while exhibiting academically critical failures, such as inappropriate citation selection or misplaced references, that conventional metrics do not capture.To this end, we introduce RWGBench, a benchmark that evaluates RWG from the perspective of citation decision-making rather than text similarity. RWGBench is constructed from a large-scale collection of 40,108 computer science papers and a retrieval corpus of 1.09 million documents, with a carefully curated test set comprising 100 papers and their corresponding published related work sections.We propose a multi-dimensional evaluation framework that assesses citation selection, contextual appropriateness, organization, and discourse structure.Experiments reveal systematic limitations in current systems that are obscured by standard evaluations, while Oracle studies further disentangle retrieval-level and generation-level bottlenecks. Human evaluation further shows that our citation-centric metrics align substantially better with expert judgment than surface-level text metrics. RWGBench offers a citation-centric testbed for developing and evaluating related work generation systems that are better aligned with scholarly writing practices.

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

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.

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

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.

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

Cross-domain recommendation (CDR) has emerged as a promising direction for expanding the applicability of recommendation systems. Recent advances in CDR have demonstrated the effectiveness of the unified distribution paradigm, which leverages shared distributions to transfer knowledge across domains and employs domain-specific adapters for targeted recommendations. While this well-designed paradigm shows promising performance, existing methods require extra supervision signals (e.g. contrastive learning on domain-masked embeddings) to maintain unified distributions across domains, leading to an inherent trade-off between unified objectives and domain-specific preference modeling. To address these limitations, we propose CD-CDR (Conditional Diffusion-CDR), a novel approach that leverages a shared conditional diffusion model to learn unified item distributions and facilitate knowledge transfer across domains. The key insight is to utilize the powerful generative capabilities of diffusion models to learn a shared distribution while naturally incorporating domain-specific characteristics through conditional generation. This design enables CD-CDR to replace traditional adapters with generation conditions as an integral part of the distribution model, thereby eliminating extra supervision signals and fundamentally resolving the trade-off between unified and domain-specific objectives. Extensive experiments on six domain pairs from two real-world datasets demonstrate that CD-CDR significantly outperforms existing methods for both normal and cold-start settings. To the best of our knowledge, this is the first work to explore the unified distribution paradigm in CDR using conditional diffusion models.

The sequential recommendation task involves predicting the items users will be interested in next based on their past interaction sequence. Recently, sequential recommender systems with generative retrieval have garnered significant attention. However, during training, these generative recommenders focus only on maximizing the prediction probability of the next target item in the temporal sequence, while neglecting awareness of diverse plausible potential items.Although introducing large language models (LLMs) with world knowledge and adding a set of auxiliary tasks that can link item identifiers to their real-world meanings can alleviate this issue, the high inference costs associated with these LLM-based recommenders make them challenging to deploy in practical scenarios. In this paper, we propose a novel learning framework, LOHRec, which leverages the order and hierarchy in generative recommendation using quantized identifiers to further explore the performance ceiling of lightweight generative recommenders. Under fair comparisons with approximate backbone parameter sizes, comprehensive experiments show that all variants of generative recommenders using our framework outperform strong prior baselines across multiple datasets. Furthermore, we empirically demonstrate that LOHRec can efficiently align lightweight generative recommenders with LLM recommendation preferences in low-resource scenarios, further demonstrating its practical utility. Our code repository is available at [https://github.com/xjw-nlp/LOHRec](https://github.com/xjw-nlp/LOHRec).

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.

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.

Click-Through Rate (CTR) prediction is a fundamental task in online advertising and recommender systems, requiring the effective modeling of feature interactions. While existing methods have improved overall prediction performance, the performance of long-tail users with limited historical data remains suboptimal. These users face two primary challenges: (i) insufficient training data leading to inaccurate predictions, and (ii) an imbalanced sample distribution that biases model learning toward head users. To address these challenges, we propose a novel framework that enhances long-tail user performance through hierarchical distribution alignment, hierarchical residual learning, and adaptive distrbution calibration. Our method first captures the shared patterns between head and long-tail users via hierarchical distribution alignment, then learns group-specific information through hierarchical residual learning. Additionally, we re-balance the sample distributions of head users and long-tail users by dynamic reweighting. To counteract potential biases introduced by reweighting, we further incorporate a distribution calibration module. Our method is model-agnostic and can be seamlessly integrated into various CTR prediction architectures that rely on feature interactions. Extensive experiments on public datasets and an online experiment demonstrate that our approach significantly improves accuracy and fairness for long-tail users while maintaining similar or even better overall performance.

Cross-domain recommender (CDR) systems aim to enhance the performance of the target domain by utilizing data from other related domains. However, irrelevant information from the source domain may instead degrade target domain performance, which is known as the negative transfer problem. There have been some attempts to address this problem, mostly by designing adaptive representations for overlapped users. Whereas, representation adaptions solely rely on the expressive capacity of the CDR model, lacking explicit constraint to filter the irrelevant source-domain collaborative information for the target domain. In this paper, we propose a novel Collaborative information regularized User Transformation (CUT) framework to tackle the negative transfer problem by directly filtering users' collaborative information. In CUT, user similarity in the target domain is adopted as a constraint for user transformation learning to filter the user collaborative information from the source domain. CUT first learns user similarity relationships from the target domain. Then, source-target information transfer is guided by the user similarity, where we design a user transformation layer to learn target-domain user representations and a contrastive loss to supervise the user collaborative information transferred. The results show significant performance improvement of CUT compared with SOTA single and cross-domain methods. Further analysis of the target-domain results illustrates that CUT can effectively alleviate the negative transfer problem.

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.

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.

Cross-domain recommendation (CDR) is an important method to improve recommender system performance, especially when observations in target domains are sparse. However, most existing cross-domain recommendations fail to fully utilize the target domain's special features and are hard to be generalized to new domains. The designed network is complex and is not suitable for rapid industrial deployment. Our method introduces a two-step domain-aware cross-attention, extracting transferable features of the source domain from different granularity, which allows the efficient expression of both domain and user interests. In addition, we simplify the training process, and our model can be easily deployed on new domains. We conduct experiments on both public datasets and industrial datasets, and the experimental results demonstrate the effectiveness of our method. We have also deployed the model in an online advertising system and observed significant improvements in both Click-Through-Rate (CTR) and effective cost per mille (ECPM).

Large-scale recommender systems usually contain hundreds of multi-valued feature fields, which have different number of values in each field. For easier computation in traditional fixed-shape neural networks, pooling operators are widely used to compress the multi-valued feature into a fixed-dimension vector. Most existing works set a single pooling method for all fields, but this leads to sub-optimal results, because different feature fields have different information distributions and thus require different pooling methods. In this work, we propose an AutoML-based framework, called AutoPooling, which can automatically and efficiently search for the optimal pooling operator for each multivalued feature. Specifically, learnable weights are assigned to all candidate pooling operators in each feature field. Then an AutoML-based algorithm is used to learn both model parameters and the field-aware weights. Finally, the optimal pooling operator can be acquired based on the associated weights. We evaluate the proposed framework on both public and industrial datasets. The results show that AutoPooling significantly outperforms the benchmarks. Further experiment results show that our method is robust in various deep recommend models and different search spaces.

With the applications of recommendation systems rapidly expanding, an increasing number of studies have focused on every aspect of recommender systems with different data inputs, models, and task settings. Therefore, a flexible library is needed to help researchers implement the experimental strategies they require. Existing open libraries for recommendation scenarios have enabled reproducing various recommendation methods and provided standard implementations. However, these libraries often impose certain restrictions on data and seldom support the same model to perform different tasks and input formats, limiting users from customized explorations. To fill the gap, we propose ReChorus2.0, a modular and task-flexible library for recommendation researchers. Based on ReChorus, we upgrade the supported input formats, models, and training&evaluation strategies to help realize more recommendation tasks with more data types. The main contributions of ReChorus2.0 include: (1) Realization of complex and practical tasks, including re-ranking and CTR prediction tasks; (2) Inclusion of various context-aware and re-ranking recommenders; (3) Extension of existing and new models to support different tasks with the same models; (4) Support of highly-customized input with impression logs, negative items, or click labels, as well as user, item, and situation contexts. To summarize, ReChorus2.0 serves as a comprehensive and flexible library that better addresses the practical problems in the recommendation scenario and caters to more diverse research needs. The implementation and detailed tutorials of ReChorus2.0 can be found at https://github.com/THUwangcy/ReChorus.

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.

Nowadays, personalized recommender systems play an increasingly important role in music scenarios in our daily life with the preference prediction ability. However, existing methods mainly rely on users' implicit feedback (e.g., click, dwell time) which ignores the detailed user experience. This paper introduces Electroencephalography (EEG) signals to personal music preferences as a basis for the personalized recommender system. To realize collection in daily life, we use a dry-electrodes portable device to collect data. We perform a user study where participants listen to music and record preferences and moods. Meanwhile, EEG signals are collected with a portable device. Analysis of the collected data indicates a significant relationship between music preference, mood, and EEG signals. Furthermore, we conduct experiments to predict personalized music preference with the features of EEG signals. Experiments show significant improvement in rating prediction and preference classification with the help of EEG. Our work demonstrates the possibility of introducing EEG signals in personal music preference with portable devices. Moreover, our approach is not restricted to the music scenario, and the EEG signals as explicit feedback can be used in personalized recommendation tasks.

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