Building agentic systems that can autonomously self-improve from experience is a longstanding goal of AI. Large language models (LLMs) today primarily self-improve via two mechanisms: self-reflection for context updates, and reinforcement learning (RL) for weight updates. In this work, we propose Evolutionary System Prompt Learning (E-SPL), a method for jointly improving a model's experiential context and its weights. In each RL iteration, E-SPL samples trajectories under multiple system prompts in parallel, then uses them to jointly update the weights with RL and evolve the system prompts with self-reflection. E-SPL encourages a natural division between procedural knowledge in weights and interpretable declarative knowledge in system prompts. Most notably, in an easy-to-hard generalization setting (AIME $\rightarrow$ BeyondAIME), E-SPL improves RL success rate from 38.8% $\rightarrow$ 45.1%. E-SPL also improves RL on AIME 2025 (56.3% $\rightarrow$ 60.6%), HMMT 2025 (50.0% $\rightarrow$ 52.7%), and agentic search (44.2% $\rightarrow$ 48.6% on gpt-oss-120b). Across all settings, E-SPL outperforms both RL-only and evolution-only baselines, demonstrating that weight updates and context updates are deeply synergistic and can together yield gains in generalization that neither achieves alone. Code is available at https://github.com/LunjunZhang/E-SPL.
Medical world models offer a framework for extending medical artificial intelligence beyond static prediction by representing evolving patient states and modelling how they change over time and in response to clinical interventions. This Review uses structured narrative synthesis and reproducible evidence mapping to define conceptual boundaries, technical foundations, application domains, and clinical evidence requirements. We screened 1,455 unique records and assembled a corpus of 300 sources, including 14 studies that met a strict empirical definition of a medical world model. The field is organized around four capabilities: patient state representation, temporal dynamics modelling, intervention-conditioned simulation, and clinician-supervised planning. Evidence spans medical imaging, longitudinal electronic health records, treatment response modelling, physiological and multimodal state modelling, ultrasound and surgical interaction, and population and health-system simulation; clinical digital twins are treated as a cross-cutting integration framework. Studies provide early evidence of technical feasibility for trajectory forecasting and comparison of candidate interventions, but most remain retrospective, task-specific, or preclinical. We distinguish functional capability from clinical evidence maturity and specify evidence needed to support progressively stronger clinical claims, from temporal forecasting to intervention comparison and planning-oriented decision support. These requirements include precise intervention representations, robust causal and mechanistic grounding, calibrated trajectory-level uncertainty, safety-constrained planning, external validation, and prospective multicentre evaluation against clinically meaningful endpoints. This framework aims to distinguish technically plausible simulation from clinically supportable prediction and decision support.
Yulin Yu, Yan Chen, Rui Hu, Siddharth Suri, Scott Counts
Generative AI is transforming the workplace by augmenting and automating cognitive tasks, reshaping how organizations work and innovate while raising questions about workplace inequality and the future of work. Despite rapid adoption, empirical evidence on how these tools alter work practices and generate productivity gains remains limited. We examine how AI use affects the quantity and nature of information work using digital trace data from the Microsoft M365 application suite across multiple large international companies. Specifically, we study how generative AI adoption shifts the balance between communication and productivity-oriented activities, such as content creation in Word. Difference-in-Differences analyses show that AI adoption is associated with significant increases in both productivity (21.2%) and communication (7.1%) application actions among users who used the AI system more than 100 times over a 20-week post-adoption period. Among users with 100-500 AI use instances, higher AI usage is also associated with continued increases in both types of activity. The smaller increase in communication represents an overall shift toward individual, documentation-focused work and reflects mixed changes in communication, including decreases in reading and organizing email, compared with more uniform increases in productivity actions. These findings suggest potential efficiency gains and reductions in information overload, while highlighting the need to ensure that AI adoption does not weaken interpersonal communication and the diffusion of diverse information that supports innovation.
Environmental, Social, and Governance (ESG) principles are reshaping the foundations of global financial governance, transforming capital allocation architectures, regulatory frameworks, and systemic risk coordination mechanisms. However, as the core medium for assessing corporate ESG performance, the ESG reports present significant challenges for large-scale understanding, due to chaotic reading order from slide-like irregular layouts and implicit hierarchies arising from lengthy, weakly structured content. To address these challenges, we propose Pharos-ESG, a unified framework that transforms ESG reports into structured representations through multimodal parsing, contextual narration, and hierarchical labeling. It integrates a reading-order modeling module based on layout flow, hierarchy-aware segmentation guided by table-of-contents anchors, and a multi-modal aggregation pipeline that contextually transforms visual elements into coherent natural language. The framework further enriches its outputs with ESG, GRI, and sentiment labels, yielding annotations aligned with the analytical demands of financial research. Extensive experiments on annotated benchmarks demonstrate that Pharos-ESG consistently outperforms both dedicated document parsing systems and general-purpose multimodal models. In addition, we release Aurora-ESG, the first large-scale public dataset of ESG reports, spanning Mainland China, Hong Kong, and U.S. markets, featuring unified structured representations of multi-modal content, enriched with fine-grained layout and semantic annotations to better support ESG integration in financial governance and decision-making.
My Chiffon Nguyen, Tianyi Hu, Vinod Anbalagan, Henry Gagnier, Pouya Sadeghi, Farah Abdou, Tidor-Vlad Pricope, Diego MMS Oliveira 等 164 位作者
As LLMs expand their language coverage, evaluating how well they understand and represent diverse cultures becomes increasingly important. However, many benchmarks remain English-centric, with limited cultural coverage and a focus on factual knowledge over its application in real-world cultural contexts. We propose CulturalRiddles, a benchmark of original, culturally grounded riddles created entirely by humans through a participatory community effort, spanning 62 cultures (45 countries, 51 languages). These riddles are deeply rooted in their respective cultures and require abductive and compositional reasoning, alongside detailed factual and social knowledge, language skills, and an understanding of culturally specific associations. We benchmark 28 LLM configurations using both automatic and large-scale human evaluation. Even frontier models achieve at most 22% accuracy in some communities. We publicly release all artifacts alongside an open contribution process to expand the benchmark's linguistic and cultural coverage.
Large language models (LLMs) perform well across many real-world tasks but remain prone to hallucination, limiting their reliability in domains such as biomedicine and healthcare, where answers must be grounded in established knowledge. Retrieval-augmented generation (RAG) mitigates this problem by incorporating external information, but often overlooks logical relationships and relies heavily on the LLM's own reasoning, which can compromise answer faithfulness. We propose OA-RAG, an agent that integrates formal reasoning with LLM reasoning, drawing on both the logical structure and informal annotations of domain ontologies. For each answer, OA-RAG reports the ontology observations used as evidence and a model-assessed grounding level (entailed, supported, related, or unrelated) that characterises the strength of that evidence. Across benchmarks derived from three large-scale ontologies and six real-world medical question-answering tasks, OA-RAG achieves the best overall performance among the evaluated ontology-fine-tuned and popular RAG methods, with particularly strong results on reasoning-intensive tasks. The reported grounding levels also offer a useful indication of answer reliability, with stronger grounding generally associated with higher accuracy. Finally, our results demonstrate the value of ontologies as knowledge sources: OA-RAG is competitive with, and sometimes outperforms, domain-specific RAG systems that draw on sources such as textbooks.
Zishan Su, Junjie Huang, Shiwen Shan, Xingyan Chen, Hui Zeng, Yuxin Su, Yanlin Wang, Michael R. Lyu
The adoption of Microservice Architecture (MSA) has revolutionized software engineering by enhancing scalability, agility, and maintainability over traditional monolithic applications. As more developers transition their legacy systems to microservice-based architectures, effective microservice decomposition-partitioning monolithic applications into highly cohesive services-becomes vital. However, this decomposition task presents significant challenges. Manual approaches are time-consuming and labor-intensive. Existing automated methods often fail to capture the necessary semantic insights from complex applications, while naive applications of Large Language Models tend to overlook crucial contextual information and design principles, leading to suboptimal results. To address these challenges, we propose MicroAgent, a Context-Augmented Multi-Agent Framework for Microservice Decomposition. Our framework divides the decomposition process into five distinct subtasks and assigns each to a specialized agent. To enhance the effectiveness of each agent, we provide tailored, multi-granularity context that keeps its analysis focused and mitigates information overload. Furthermore, to ensure the decomposition adheres to established design principles, we integrate analytical tools that guide the agents' decision-making. Experimental evaluations on 10 Java Web applications demonstrate that MicroAgent achieves an average decomposition accuracy of 89.2%, outperforming the state-of-the-art method by 24.6%. We also conduct a case study to highlight the practical benefits of our design.
Yuyan Chen, Zhendong Hou, Lei Xia, Jiahao Li, Zhuolin Ji, Zhixu Li
In the fields of advertising design, artistic creation, and cultural dissemination, there is an increasingly urgent demand for high-quality images that cater to fine-grained aesthetic preferences. Although existing large-scale models can generally meet basic requirements for clarity and alignment with textual elements, they still face significant bottlenecks in achieving precise control and aesthetic optimization. To address this limitation, we propose a set of comprehensive preference indicators across two major dimensions, text-image consistency and aesthetic quality, encompassing multiple criteria ranging from exposure and clarity to visual guidance and innovativeness. Building on these indicators, we have developed a generative framework named AesX to steer the model consistently toward a generation path that more closely aligns with human aesthetic sensibilities. Our experimental findings demonstrate that this approach yields significant improvements in both target recognition accuracy and overall visual aesthetic presentation.
尚无 AI 解读，展示原文摘要
cs.CL
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 6: Industry Track) · 2026 · 正式发表
Bryan Chen Zhengyu Tan, Zhengyuan Liu Agency for Science, Technology and Research (A*STAR), Singapore, Xiaoyuan Yi Microsoft Research, Jing Yao Microsoft, Xing Xie Microsoft Research Asia, Nancy F. Chen A*STAR, Roy Ka-Wei Lee Singapore University of Technology and Design
Despite their global prevalence, many Large Language Models (LLMs) are aligned to a monolithic, often Western-centric set of values. This paper investigates the more challenging task of fine-grained value alignment: examining whether LLMs can emulate the distinct cultural values of demographic subgroups. Using Singapore as a case study and the World Values Survey (WVS), we examine the value landscape and show that even state-of-the-art models like GPT-4.1 achieve only 57.4% accuracy in predicting subgroup modal preferences. We construct a dataset of over 20,000 samples to train and evaluate a range of models. We demonstrate that simple fine-tuning on structured numerical preferences yields substantial gains, improving accuracy on unseen, out-of-distribution subgroups by an average of 17.4%. These gains partially transfer to open-ended generation. However, we find significant pre-existing performance biases, where models better emulate young, male, Chinese, and Christian personas. Furthermore, while fine-tuning improves average performance, it widens the disparity between subgroups when measured by distance-aware metrics. Our work offers insights into the limits and fairness implications of subgroup-level cultural alignment.
尚无 AI 解读，展示原文摘要
cs.CL
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表
While large language models (LLMs) have substantially improved Text-to-SQL generation, a pronounced gap remains between AI systems and human experts on challenging benchmarks such as BIRD-SQL. We argue this gap stems largely from the prevailing single-pass paradigm, which lacks the iterative reasoning, schema exploration, and error-correction behaviors that humans naturally employ. To address this limitation, we introduce SQL-Trail, a multi-turn reinforcement learning (RL) agentic framework for Text-to-SQL. Rather than producing a query in one shot, SQL-Trail interacts with the database environment and uses execution feedback to iteratively refine its predictions. Our approach centers on two key ideas: (i) an adaptive turn-budget allocation mechanism that scales the agent’s interaction depth to match question difficulty, and (ii) a composite reward panel that jointly incentivizes SQL correctness and efficient exploration. Across benchmarks, SQL-Trail sets a new state of the art and delivers strong data efficiency—up to 18× higher than prior single-pass RL state-of-the-art methods. Notably, our 7B and 14B models outperform substantially larger proprietary systems by 5% on average, underscoring the effectiveness of interactive, agentic workflows for robust Text-to-SQL generation.
尚无 AI 解读，展示原文摘要
cs.CL · cs.AI
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表
Jie He, Nan Hu Southeast University, Wanqiu Long, Jiaoyan Chen The University of Manchester, Jeff Z. Pan University of Edinburgh, University of Edinburgh
Retrieval-Augmented Generation (RAG) enhances large language models (LLMs) by integrating external knowledge, enabling them to tackle knowledge-intensive tasks. However, limited research has explored how LLMs effectively leverage RAG techniques for multi-hop question answering (QA), particularly when handling knowledge with with varying degrees of familiarity. In this paper, we introduce MINTQA (Multi-hop Question Answering on New and Tail Knowledge), a benchmark designed to evaluate multi-hop QA performance on questions involving 10,479 question-answer pairs for evaluating old/new knowledge and 17,887 pairs for assessing popular/unpopular knowledge, with each question equipped with corresponding sub-questions and answers. This benchmark primarily evaluates the multi-hop reasoning ability of LLMs and their capacity to handle knowledge with varying levels of familiarity during the reasoning process. We evaluate 22 state-of-the-art LLMs using three distinct QA strategies: LLM-based parameterized knowledge QA, direct RAG-enhanced QA, and multi-hop RAG-enhanced QA. Our experiments reveal key challenges in how LLMs handle knowledge with different familiarity and offer insights into improving their multi-hop reasoning capabilities when combined with RAG techniques.
尚无 AI 解读，展示原文摘要
cs.CL
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表
Xu Liu East China University of Science and Technology, Yan Chen East China University of Science and Technology, Kan Ling East China University of Science and Technology, Yichi Zhu, Hengrun Zhang East China University of Science and Technology, Guisheng Fan East China University of Science and Technology, Huiqun Yu
Despite extensive safety alignment, Large Language Models (LLMs) remain vulnerable to jailbreak attacks. However, existing methods generally lack the capability for continuous learning and self-evolution from interactions, limiting the diversity and adaptability of attack strategies. To address this, we propose ASTRA, an automated framework capable of autonomously discovering, retrieving, and evolving attack strategies. ASTRA operates on a closed-loop "attack-evaluate-distill-reuse” mechanism, which not only generates attack prompts but also automatically distills reusable strategies from every interaction. To systematically manage these strategies, we introduce a dynamic three-tier strategy library (Effective, Promising, and Ineffective) that categorizes strategies based on performance. This hierarchical memory mechanism enables the framework to enhance efficiency by leveraging successful patterns while optimizing the exploration space by avoiding known failures. Extensive experiments in a black-box setting demonstrate that ASTRA significantly outperforms existing baselines.
尚无 AI 解读，展示原文摘要
cs.CL
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表
Fengbo Ma University of Georgia, Zixin Rao, Xiaoting Li University of Georgia, Zhetao Chen, Hongyue Sun University of Georgia, Yiping Zhao University of Georgia, Xianyan Chen University of Georgia, Zhen Xiang University of Georgia
Scientific research relies on accurate information retrieval from literature to support analytical decisions.In this work, we introduce a new task, INformation reTRieval through literAture reVIEW (IntraView), which aims to automate fine-grained information retrieval faithfully grounded in the provided content in response to research-driven queries, and propose IntrAgent, an LLM-based agent that addresses this challenging task.In particular, IntrAgent is designed to mimic human behaviors when reading literature for information retrieval - identifying relevant sections and then iteratively extracting key details to refine the retrieved information.It follows a two-stage pipeline: a Section Ranking stage that prioritizes relevant literature sections through structural-knowledge-enabled reasoning, and an Iterative Reading stage that continuously extracts details and synthesizes them into concise, contextually grounded answers.To support rigorous evaluation, we introduce IntraBench, a new benchmark consisting of 315 test instances built from expert-authored questions paired with literature spanning five STEM domains.Across seven backbone LLMs, IntrAgent achieves on average 13.2% higher cross-domain accuracy than state-of-the-art RAG and research-agent baselines.
尚无 AI 解读，展示原文摘要
cs.CL
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表
Qisheng Su, Shiting Huang, Zhen Fang University of Science and Technology of China, Ziyan Chen, Zehui Chen, Feng Zhao University of Science and Technology of China
In real-world Tool-Integrated Reasoning (TIR) scenarios, a major source of inefficiency is that the toolcalls create pauses between LLM requests and cause KV-cache eviction. Also, the long, unfiltered response returned by external tools inflates the KV-cache, so each decode step spends more time loading the growing cache and thus becomes steadily slower as context length increases. However, existing efficiency metrics like token counts and toolcall counts fail to capture this real computational cost. To address this, we introduce PTE (Prefill Token Equivalents), a hardware-aware TIR-efficiency metric that unifies internal reasoning and external tool-use costs while explicitly accounting for non-reusable KV-Cache and long-tool-response scenarios, thus better reflects real-world scenarios. We conduct extensive experiments across five TIR benchmarks, quantify their PTE costs, and identify four inefficiency patterns that appear in TIR. In a simulated high-concurrency industrial setting, PTE explains wall-clock latency significantly better than token-count metric. We also discover that trajectories with higher PTE costs tend to have lower reasoning correctness, indicating that simply using more tools does not improve the quality of the answer. PTE offers a new perspective on the efficiency of Tool-Integrated Reasoning. The code is available.
尚无 AI 解读，展示原文摘要
cs.CL
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表
Weihua Zheng, Zhengyuan Liu Agency for Science, Technology and Research (A*STAR), Singapore, Tanmoy Chakraborty Indian Institute of Technology, Delhi, Weiwen Xu Alibaba Group, Xiaoxue Gao, Bryan Chen Zhengyu Tan, Bowei Zou A*STAR, Chang Liu Shanghai University of Finance and Economics 等 35 位作者
The global deployment of Large Language Models (LLMs) underscores the urgent need to evaluate their cultural alignment. However, assessing genuine “cultural awareness” across modalities (text, vision, speech) and languages remains a significant challenge. To comprehensively investigate this domain, we propose MMAC, a systematic framework that encompasses a tri-modally aligned cultural benchmark creation pipeline and a five-dimensional evaluation protocol to assess cross-country awareness disparities, evaluate cross-lingual and cross-modal consistency, and verify cultural knowledge generalization and grounding validity. Given the prevailing Western cultural bias in current models, we focus on 8 Asian countries as our dataset foundation to more acutely reveal potential cultural deficiencies in LLMs. Our dataset, MMAC-bench, features 27,000 human-curated questions across 10 languages. Crucially, it is the first dataset aligned at the input level across text, image, and speech, enabling direct cross-modal transfer tests. Each question consists of multiple-choice options accompanied by open-ended generated explanations, where 79% require multi-step reasoning grounded in cultural context, moving beyond simple memorization. We probe the causes of modal divergence, offering insights into fostering culturally robust MLLMs.
尚无 AI 解读，展示原文摘要
cs.CL · cs.CV
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表