论文检索

Lotem Peled-Cohen, Maya Zadok, Nitay Calderon, Hila Gonen, Roi Reichart

Cognitive decline often surfaces in language years before diagnosis. It is frequently non-experts, such as those closest to the patient, who first sense a change and raise concern. As LLMs become integrated into daily communication and used over prolonged periods, it may even be an LLM that notices something is off. But what exactly do they notice–and should be noticing–when making that judgment? This paper investigates how dementia is perceived through language by non-experts. We presented transcribed picture descriptions to non-expert humans and LLMs, asking them to intuitively judge whether each text was produced by someone healthy or with dementia. We introduce an explainable method that uses LLMs to extract high-level, expert-guided features representing these picture descriptions, and use logistic regression to model human and LLM perceptions and compare with clinical diagnoses. Our analysis reveals that human perception of dementia is inconsistent and relies on a narrow, and sometimes misleading, set of cues. LLMs, by contrast, draw on a richer, more nuanced feature set that aligns more closely with clinical patterns. Still, both groups show a tendency toward false negatives, frequently overlooking dementia cases. Through our interpretable framework and the insights it provides, we hope to help non-experts better recognize the linguistic signs that matter.

Yinghao Hu, Yaoyao Yu, Leilei Gan, Bin Wei, Kun Kuang, Fei Wu

Recent advances in test-time scaling of large language models (LLMs), exemplified by DeepSeek-R1 and OpenAI’s o1, show that extending the chain of thought during inference can significantly improve general reasoning performance. However, the impact of this paradigm on legal reasoning remains insufficiently explored. To address this gap, we present the first systematic evaluation of 12 LLMs, including both reasoning-focused and general-purpose models, across 17 Chinese and English legal tasks spanning statutory and case-law traditions. In addition, we curate a bilingual chain-of-thought dataset for legal reasoning through distillation from DeepSeek-R1 and develop Legal-R1, an open-source model specialized for the legal domain. Experimental results show that Legal-R1 delivers competitive performance across diverse tasks. DeepSeek-R1 exhibits clear advantages in Chinese legal reasoning, while OpenAI’s o1 achieves comparable results on English tasks. We further conduct a detailed error analysis, which reveals recurring issues such as outdated legal knowledge, limited capacity for legal interpretation, and susceptibility to factual hallucinations. These findings delineate the main obstacles confronting legal-domain LLMs and suggest promising directions for future research. We release the dataset and model at https://github.com/YinghaoHu/Legal-R1-14B.

Nathan Inkiriwang, Necva Bölücü, Garth Tarr, Maciej Rybinski

High-dimensional text embeddings are foundational to modern NLP but costly to store and use. While embedding compression addresses these challenges, selecting the best compression method remains difficult. Existing evaluation methods for compressed embeddings are either expensive or too simplistic. We introduce a comprehensive intrinsic evaluation framework featuring a suite of task-agnostic metrics that together provide a reliable proxy for downstream performance. A key contribution is EOS_k, a novel spectral fidelity measure specifically designed to be robust to embedding anisotropy. Through extensive experiments on diverse embeddings across four downstream tasks, we demonstrate that our intrinsic metrics reliably predict extrinsic performance and reveal how different embedding architectures depend on distinct geometric properties. Our framework provides a practical, efficient, and interpretable alternative to standard evaluations for compressed embeddings.

Anuj Kumar, Mohammed Faisal Sayed, Satyadev Ahlawat, Yamuna Prasad

Marathi, the third most widely spoken language in India with over 83 million native speakers, remains significantly underrepresented in Natural Language Processing (NLP) research. While sentiment analysis has achieved substantial progress in high-resource languages such as English, Chinese, and Hindi, available Marathi datasets are limited to coarse sentiment labels and lack fine-grained emotional categorization or interpretability through explanations. To address this gap, we present a new annotated dataset of 10,762 Marathi sentences, each labeled with sentiment (positive, negative, or neutral), emotion (joy, anger, surprise, disgust, sadness, fear, or neutral), and a corresponding natural language justification. Justifications are written in English and generated using GPT-4 under a human-in-the-loop framework to ensure label fidelity and contextual alignment. Extensive experiments with both classical and transformer-based models demonstrate the effectiveness of the dataset for interpretable affective computing in a low-resource language setting, offering a benchmark for future research in multilingual and explainable NLP.

Huaming Du, Lei Yuan, Cancan Feng, Guisong Liu, Gang Kou, Carl Yang

Large Language Models (LLMs) have achieved impressive success across a range of natural language processing tasks. However, they still underperform in text classification tasks compared to fine-tuned small models. This can be linked to complexities in addressing context-dependent expressions and complex linguistic phenomena. In contrast, fine-tuned small models typically achieve high prediction accuracy but often lack explanations for predictions. Existing explanation methods that generate keywords may be less effective due to missing critical contextual information. To mitigate these challenges, we propose a novel method termed Dialectical Explanation Training (**DET**). This method introduces a new prompting strategy, Dialectical Prompting, and integrates it with Explanation-Guided Training. Dialectical Prompting uses LLMs with our designed dialectical prompt to generate explanations for possible labels. These explanations handle context-dependent expressions and complex linguistic phenomena by considering multiple perspectives and providing rich, contextually relevant information. Explanation-Guided Training employs these explanations as features for training a small model, which combines the advantages of dialectical explanations and the predictive power of fine-tuned models to improve overall accuracy and interpretability. In addition, we incorporate the theory of Evidential Deep Learning, which further enhances the model’s classification performance and quantify the uncertainty of its predictions. Extensive experiments on multiple datasets from diverse domains have demonstrated that our proposed model significantly improves accuracy and explanation quality over state-of the-art methods in text classification.

Shin-nosuke Ishikawa, Masato Todo, Taiki Ogihara, Hirotsugu Ohba

We present a tendency of large language models (LLMs) to generate absurd patterns despite their clear inappropriateness in a simple task of identifying regularities in number series. Several approaches have been proposed to apply LLMs to complex real-world tasks, such as providing knowledge through retrieval-augmented generation and executing multi-step tasks using AI agent frameworks. However, these approaches rely on the logical consistency and self-coherence of LLMs, making it crucial to evaluate these aspects and consider potential countermeasures. To identify cases where LLMs fail to maintain logical consistency, we conducted an experiment in which LLMs were asked to explain the patterns in various integer sequences, ranging from arithmetic sequences to randomly generated integer series. While the models successfully identified correct patterns in arithmetic and geometric sequences, they frequently over-recognized patterns that were inconsistent with the given numbers when analyzing randomly generated series. This issue was observed even in multi-step reasoning models, including OpenAI o3, o4-mini, and Google Gemini 2.5 Flash Preview Thinking. This tendency to perceive non-existent patterns can be interpreted as the AI model equivalent of Idola Tribus and highlights potential limitations in their capability for applied tasks requiring logical reasoning, even when employing chain-of-thought reasoning mechanisms.

Mohammad R. Rezaei, Reza Saadati Fard, Jayson L. Parker, Rahul G. Krishnan, Milad Lankarany

Large Language Models (LLMs) have greatly advanced medical Question Answering (QA) by leveraging vast clinical data and medical literature. However, the rapid evolution of medical knowledge and the labor-intensive process of manually updating domain-specific resources can undermine the reliability of these systems. We address this challenge with Agentic Medical Graph-RAG (AMG-RAG), a comprehensive framework that automates the construction and continuous updating of Medical Knowledge Graph (MKG), integrates reasoning, and retrieves current external evidence from the MKG for medical QA.Evaluations on the MEDQA and MEDMCQA benchmarks demonstrate the effectiveness of AMG-RAG, achieving an F1 score of 74.1% on MEDQA and an accuracy of 66.34% on MEDMCQA—surpassing both comparable models and those 10 to 100 times larger. By dynamically linking new findings and complex medical concepts, AMG-RAG not only boosts accuracy but also enhances interpretability for medical queries, which has a critical impact on delivering up-to-date, trustworthy medical insights.

Xinyi Chen, Yifei Yuan, Jiaang Li, Serge Belongie, Maarten de Rijke, Anders Søgaard

Language models are often said to face a symbol grounding problem. While some have argued the problem can be solved without resort to other modalities, many have speculated that grounded learning is more efficient. We explore this question in Othello, a simplified, rule-based world that offers a controlled and interpretable testbed for studying world understanding. Building on prior work, we introduce VISOTHELLO, a multi-modal model trained jointly on move sequences and board images. Using the Othello rule understanding task, we examine whether multi-modal learning provides advantages over text-only approaches. We further evaluate robustness under semantically irrelevant perturbations and analyze the consistency of cross-modal alignment. Our results suggest that multi-modal training not only improves performance and robustness but also promotes convergence toward shared internal representations across different model architectures.

Khanh-Tung Tran, Nguyet-Hang Vu, Barry O’Sullivan, Hoang D. Nguyen

Cross-lingual chain-of-thought prompting techniques have proven effective for investigating diverse reasoning paths in Large Language Models (LLMs), especially for low-resource languages. Despite these empirical gains, the mechanisms underlying cross-lingual improvements remain perplexing. This study, therefore, addresses whether the benefits of cross-lingual prompting arise from language-specific reasoning structures intrinsic to each language, or are simply a consequence of improved comprehension through cross-linguistic exposure. We employ neuron intervention and perturbation techniques to analyze and deactivate language-specific reasoning neurons during cross-lingual prompting, leading to performance disparities across languages, up to 27.4%. Our findings disentangle that these neurons are essential for reasoning in their respective languages, but have minimal effect on reasoning in other languages, providing evidence for the existence of language-specific local reasoning structures and guiding the development of more interpretable and effective multilingual AI systems.

Matthieu Tehenan

Sentence embeddings are central to modern natural language processing, powering tasks such as clustering, semantic search, and retrieval-augmented generation. Yet, they remain largely opaque: their internal features are not directly interpretable, and users lack fine-grained control for downstream tasks. To address this issue, we introduce a formal framework to characterize the organization of features in sentence embeddings through information-theoretic means. Building on this foundation, we develop a method to identify interpretable feature directions and show how they can be composed to capture richer semantic structures. Experiments on both synthetic and real-world datasets confirm the presence of this semantic geometry and highlight the utility of our approach for enhancing interpretability and fine-grained control in sentence embeddings.

Wen-Han Hsieh, Elvis Hsieh, Dantong Niu, Trevor Darrell, Roei Herzig, David M. Chan

Recently, Vision-Language-Action (VLA) models have demonstrated strong performance on a range of robotic tasks. These models rely on multimodal inputs, with language instructions playing a crucial role-not only in predicting actions, but also in robustly interpreting user intent, even when the requests are impossible to fulfill. In this work, we investigate how VLAs can recognize, interpret, and respond to false-premise instructions-natural language commands that reference objects or conditions absent from the environment. We propose — Instruct-Verify-and-Act (IVA) — a unified framework that (i) detects when an instruction cannot be executed due to a false premise, (ii) engages in language-based clarification or correction, and (iii) grounds plausible alternatives in perception and action. Towards this end, we construct a large-scale instruction tuning setup with structured language prompts and train a VLA model capable of handling both accurate and erroneous requests. Our approach leverages a contextually augmented, semi-synthetic dataset containing paired positive and false-premise instructions, enabling robust detection and natural language correction. Our experiments show that IVA can improves false premise detection accuracy by 58.89% over baselines, while increasing successful responses in false-premise scenarios by 27.89%.

Qingsong Lv, Yangning Li, Zihua Lan, Zishan Xu, Jiwei Tang, Tingwei Lu, Yinghui Li, Wenhao Jiang, Hong-Gee Kim, Hai-Tao Zheng 等

Instruction tuning of large language models (LLMs) benefits more from a handful of high-quality examples than from hordes of low-quality ones. Existing selection methods typically rely on static, heuristic quality scores and are executed only once before training. Consequently, they neither adapt to the changing state of the model nor target downstream objectives, leaving substantial room for optimization. We propose RAISE (**R**einforced **A**daptive **I**nstruction **SE**lection), a *dynamic*, *task-driven* framework that integrates selection into every training step. At each step, RAISE estimates the expected contribution of each candidate instruction to task performance and admits only the most helpful. By modeling this process as sequential decision making, we optimize the selector with reinforcement learning, yielding an interpretable policy specialized for the target task. Extensive experiments show that RAISE reaches comparable or better results than full-data training while updating only 1% of the steps, demonstrating both high efficacy and significant computational savings.

Feiyu Wang, Ziran Zhao, Dong Yu, Pengyuan Liu

Multimodal Large Language Models (MLLMs) have demonstrated exceptional performance across various tasks. However, the internal mechanisms by which they interpret and integrate cross-modal information remain insufficiently understood. In this paper, to address the limitations of prior studies that could only identify neurons corresponding to single-token and rely on the vocabulary of LLMs, we propose a novel method to identify multimodal neurons in Transformer-based MLLMs. Then we introduce fuzzy set theory to model the complex relationship between neurons and semantic concepts and to characterize how multiple neurons collaboratively contribute to semantic concepts. Through both theoretical analysis and empirical validation, we demonstrate the effectiveness of our method and present some meaningful findings. Furthermore, by modulating neuron activation values based on the constructed fuzzy sets, we enhance performance on the Visual Question Answering (VQA) task, showing the practical value of our approach in downstream applications in MLLMs.

Qiyao Peng, Hongtao Liu, Hua Huang, Jian Yang, Qing Yang, Minglai Shao

Recently, Large Language Models (LLMs) have demonstrated remarkable capabilities in natural language understanding, reasoning, and generation, prompting the recommendation community to leverage these powerful models to address fundamental challenges in traditional recommender systems, including limited comprehension of complex user intents, insufficient interaction capabilities, and inadequate recommendation interpretability. This survey presents a comprehensive synthesis of this rapidly evolving field. We consolidate existing studies into three paradigms: (i) recommender-oriented methods, which directly enhance core recommendation mechanisms; (ii) interaction-oriented methods, which conduct multi-turn conversations to elicit preferences and deliver interpretable explanations; and (iii) simulation-oriented methods, that model user-item interactions through multi-agent frameworks. Then, we dissect a four-module agent architecture: profile, memory, planning, and action. Then we review representative designs, public datasets, and evaluation protocols. Finally, we give the open challenges that impede real-world deployment, including cost-efficient inference, robust evaluation, and security.

Hyomin Kim, Yunhui Jang, Sungsoo Ahn

Large language models (LLMs) have large potential for molecular optimization, as they can gather external chemistry tools and enable collaborative interactions to iteratively refine molecular candidates. However, this potential remains underexplored, particularly in the context of structured reasoning, interpretability, and comprehensive tool-grounded molecular optimization. To address this gap, we introduce MT-Mol, a multi-agent framework for molecular optimization that leverages tool-guided reasoning and role-specialized LLM agents. Our system incorporates comprehensive RDKit tools, categorized into five distinct domains: structural descriptors, electronic and topological features, fragment-based functional groups, molecular representations, and miscellaneous chemical properties. Each category is managed by an expert analyst agent, responsible for extracting task-relevant tools and enabling interpretable, chemically grounded feedback. MT-Mol produces molecules with tool-aligned and stepwise reasoning through the interaction between the analyst agents, a molecule-generating scientist, a reasoning-output verifier, and a reviewer agent. As a result, we show that our framework shows the state-of-the-art performance of the PMO-1K benchmark on 15 out of 23 tasks and outperforms LLM baselines on ChemCoTBench benchmark.

Shijie Zhang, Renhao Li, Songsheng Wang, Philipp Koehn, Min Yang, Derek F. Wong

The advancement of Large Language Models (LLMs) enables flexible and interpretable automatic evaluations. In the field of machine translation evaluation, utilizing LLMs with translation error annotations based on Multidimensional Quality Metrics (MQM) yields more human-aligned judgments. However, current LLM-based evaluation methods still face challenges in accurately identifying error spans and assessing their severity. In this paper, we propose HiMATE, a Hierarchical Multi-Agent Framework for Machine Translation Evaluation. We argue that existing approaches inadequately exploit the fine-grained structural and semantic information within the MQM hierarchy. To address this, we develop a hierarchical multi-agent system grounded in the MQM error typology, enabling granular evaluation of subtype errors. Two key strategies are incorporated to further mitigate systemic hallucinations within the framework: the utilization of the model’s self-reflective capability and the facilitation of agent discussion involving asymmetric information. Empirically, HiMATE outperforms competitive baselines across different datasets in conducting human-aligned evaluations. Further analyses underscore its significant advantage in error span detection and severity assessment, achieving an average F1-score improvement of 89% over the best-performing baseline. We make our code and data publicly available at https://github.com/nlp2ct-shijie/HiMATE.

Nakyung Lee, Yeongoon Kim, Minhae Oh, Suhwan Kim, Jin Woo Koo, Hyewon Jo, Jungwoo Lee

Transformer-based self-attention mechanism serves as the core of modern language models, yet it often suffers from *localization*, where attentions collapse onto a limited subset of tokens and fail to capture long-range dependencies. To address this issue, we propose **Self-Attention One-step Belief Propagation (SAOBP)**, a refinement framework that injects *multi-hop* relationships through a belief propagation process. To interpret and quantify these interactions, we introduce **Global Token Dependency (GTD)** that captures the relative contribution of multi-hop connections within the attention graph. Empirical results indicate that SAOBP helps prevent entropy collapse in deeper layers and adaptively maintains GTD at task-appropriate levels, thereby supporting improvements in model performance. Importantly, we observe competitive gains in small-scale models, highlighting its potential for improving inference quality in resource-constrained scenarios.

Jian Jornbowrl Wang, Xiaofei Xie, Qiang Hu, Shangqing Liu, Yi Li

Code Large Language Models (Code LLMs) have opened a new era in programming with their impressive capabilities. However, recent research has revealed critical limitations in their ability to reason about runtime behavior and understand the actual functionality of programs, which poses significant challenges for their post-training and practical deployment. Specifically, Code LLMs encounter two principal issues: (1) a lack of proficiency in reasoning about program execution behavior, as they struggle to interpret what programs actually do during runtime, and (2) inconsistent and fragmented representation of semantic information, such as execution traces, across existing methods, which hinders their ability to generalize and reason effectively. These challenges underscore the necessity for more systematic approaches to enhance the reasoning capabilities of Code LLMs. To address these issues, we introduce a generic framework to support integrating semantic information (e.g., execution trace) to code task-relevant prompts, and conduct a comprehensive study to explore the role of semantic information in enhancing the reasoning ability of Code LLMs accordingly. Specifically, we focus on investigating the usefulness of trace-based semantic information in boosting supervised fine-tuning(SFT) and post-phase inference of Code LLMs. The experimental results surprisingly disagree with previous works and demonstrate that semantic information has limited usefulness for SFT and test time scaling of Code LLM.

Ping Chen, Xiang Liu, Zhaoxiang Liu, Zezhou Chen, Xingpeng Zhang, Huan Hu, Zipeng Wang, Kai Wang, Shuming Shi, Shiguo Lian

With the rapid advancement of large language models (LLMs), natural language processing (NLP) has achieved remarkable progress. Nonetheless, significant challenges remain in handling texts with ambiguity, polysemy, or uncertainty. We introduce the Fuzzy Reasoning Chain (FRC) framework, which integrates LLM semantic priors with continuous fuzzy membership degrees, creating an explicit interaction between probability-based reasoning and fuzzy membership reasoning. This transition allows ambiguous inputs to be gradually transformed into clear and interpretable decisions while capturing conflicting or uncertain signals that traditional probability-based methods cannot. We validate FRC on sentiment analysis tasks, where both theoretical analysis and empirical results show that it ensures stable reasoning and facilitates knowledge transfer across different model scales. These findings indicate that FRC provides a general mechanism for managing subtle and ambiguous expressions with improved interpretability and robustness.

Yi Su, Jiayi Zhang, Shu Yang, Xinhai Wang, Lijie Hu, Di Wang

Rapid integration of large language models (LLMs) into societal applications has intensified concerns about their alignment with universal ethical principles, as their internal value representations remain opaque despite behavioral alignment advancements. Current approaches struggle to systematically interpret how values are encoded in neural architectures, limited by datasets that prioritize superficial judgments over mechanistic analysis. We introduce ValueLocate, a mechanistic interpretability framework grounded in the Schwartz Values Survey, to address this gap. Our method first constructs ValueInsight, a dataset that operationalizes four dimensions of universal value through behavioral contexts in the real world. Leveraging this dataset, we develop a neuron identification method that calculates activation differences between opposing value aspects, enabling precise localization of value-critical neurons without relying on computationally intensive attribution methods. Our proposed validation method demonstrates that targeted manipulation of these neurons effectively alters model value orientations, establishing causal relationships between neurons and value representations. This work advances the foundation for value alignment by bridging psychological value frameworks with neuron analysis in LLMs.