论文检索

Yuxuan Ye, Raul Santos-Rodriguez, Edwin Simpson

Reinforcement learning with evaluation metrics as rewards is widely used to enhance specific capabilities of language models. However, for tasks such as factually consistent summarisation, existing metrics remain underdeveloped, limiting their effectiveness as signals for shaping model behaviour.While individual factuality metrics are unreliable, their combination can more effectively capture diverse factual errors. We leverage this insight to introduce an automated training pipeline that improves factual consistency in summaries by aggregating scores from different weak metrics. Our approach avoids the need for complex reward shaping by mapping scores to preferences and filtering out cases with high disagreement between metrics. For each source document, we generate lexically similar summary pairs by varying decoding strategies, enabling the model to learn from factual differences caused by subtle lexical differences. This approach constructs a high-quality preference dataset using only source documents.Experiments demonstrate consistent factuality gains across models, ranging from early encoder-decoder architectures to modern large language models, with smaller models reaching comparable factuality to larger ones.

Charlott Jakob, David Harbecke, Patrick Parschan, Pia Wenzel Neves, Vera Schmitt

Large Language Models are increasingly used in applications requiring objective assessment, which could be compromised by political bias. Many studies found preferences for left-leaning positions in LLMs, but downstream effects on tasks like fact-checking remain underexplored. In this study, we systematically investigate political bias through exchanging words with euphemisms or dysphemisms in German claims. We construct minimal pairs of factually equivalent claims that differ in political connotation, to assess the consistency of LLMs in classifying them as true or false. We evaluate six LLMs and find that, more than political leaning, the presence of judgmental words significantly influences truthfulness assessment. While a few models show tendencies of political bias, this is not mitigated by explicitly calling for objectivism in prompts. Warning: This paper contains content that may be offensive or upsetting.

Qinhong Lin, Zhongliang Yang, Yuang Cai, Dingfu Yu, Xuan Xu, Yu Li, Linna Zhou

Retrieval-Augmented Generation (RAG) plays a critical role in mitigating hallucinations and improving factual accuracy for Large Language Models (LLMs). While dynamic retrieval techniques aim to determine retrieval timing and content based on model intrinsic needs, existing approaches struggle to generalize effectively in black-box model scenarios. To address this limitation, we propose the Semantic Contribution-Aware Adaptive Retrieval (SCAAR) framework. SCAAR iteratively leverages the semantic importance of words in upcoming sentences to dynamically adjust retrieval thresholds and filter information, retaining the top-\alpha% most semantically significant words for constructing retrieval queries. We comprehensively evaluate SCAAR against baseline methods across four long-form, knowledge-intensive generation datasets using four models. Our method achieved the highest score on each dataset with GPT-4o. Extensive experiments also analyze the impact of various hyperparameters within the framework. Our results demonstrate SCAAR’s superior or competitive performance, showcasing its ability to effectively detect model retrieval needs and construct efficient retrieval queries for relevant knowledge about problem-solving in black-box scenarios. Our code is available on https://github.com/linqinhong/SAC.

Juli Bakagianni, Korbinian Randl, Guido Rocchietti, Cosimo Rulli, Franco Maria Nardini, Salvatore Trani, Aron Henriksson, Anna Romanova, John Pavlopoulos

Food safety demands timely detection, regulation, and public communication, yet the lack of structured datasets hinders Natural Language Processing (NLP) research. We present and release a new dataset of human-written and Large Language Model (LLM)-generated summaries of food safety documents, plus food safety related metadata. We evaluate its utility on three NLP tasks directly reflecting food safety practices: multilabel classification for organizing documents into domain-specific categories; document retrieval for accessing regulatory and scientific evidence; and question answering via retrieval-augmented generation that improves factual accuracy.We show that LLM summaries perform comparably or better than human ones across tasks. We also demonstrate clustering of summaries for event tracking and compliance monitoring. This dataset enables NLP applications that support core food safety practices, including the organization of regulatory and scientific evidence, monitoring of compliance issues, and communication of risks to the public.

Yitao Long, Tiansheng Hu, Yilun Zhao, Arman Cohan, Chen Zhao

Large Language Models (LLMs) frequently hallucinate to long-form questions, producing plausible yet factually incorrect answers. A common mitigation strategy is to provide attribution to LLM outputs. However, existing benchmarks primarily focus on simple attribution that retrieves supporting textual evidence as references. We argue that in real-world scenarios such as financial applications, attribution goes beyond reference retrieval.We introduce FinLFQA, a benchmark designed to evaluate the ability of LLMs to generate long-form answers to complex financial questions with reliable and nuanced attributions. FinLFQA evaluates three critical aspects of attribution through human annotations: (1) supporting evidence extracted from financial reports, (2) intermediate numerical reasoning steps, and (3) domain-specific financial knowledge that informs the reasoning process.We further provide an automatic evaluation framework covering both answer quality and attribution quality. Through extensive experiments on eight LLMs across multiple attribution-generation paradigms, we find that fine-grained metrics are important to distinguish model capabilities, that end-to-end generation achieves comparable performance to post-hoc approaches, and that iterative refinement only helps when guided by external feedback.

Jiatong Han, Neil Band, Muhammed Razzak, Jannik Kossen, Tim G. J. Rudner, Yarin Gal

Large language models (LLMs) often mislead users with confident hallucinations. Current approaches to detect hallucination require many samples from the LLM generator, which is computationally infeasible as frontier model sizes and generation lengths continue to grow. We present a remarkably simple baseline for detecting hallucinations in long-form LLM generations, with performance comparable to expensive multi-sample approaches while drawing only a single sample from the LLM generator. Our key finding is that LLM hidden states are highly predictive of factuality in long-form natural language generation and that this information can be efficiently extracted at inference time using a lightweight probe. We benchmark a variety of long-form hallucination detection methods across open-weight models up to 405B parameters and demonstrate that our approach achieves competitive performance with up to 100x fewer FLOPs. Furthermore, our probes generalize to out-of-distribution model outputs, evaluated using hidden states of smaller open-source models. Our results demonstrate the promise of hidden state probes in detecting long-form LLM hallucinations.

Chimaobi Okite, Naihao Deng, Kiran Bodipati, Huaidian Hou, Joyce Chai, Rada Mihalcea

Recent years have witnessed a growing interest in personalizing the responses of large language models (LLMs). While existing evaluations primarily focus on whether a response aligns with a user’s preferences, we argue that factuality is an equally important yet often overlooked dimension. In the context of personalization, we define a model as robust if its responses are both factually accurate and align with the user preferences. To assess this, we introduce PERG, a scalable framework for evaluating robustness of LLMs in personalization, along with a new dataset, PERGData. We evaluate fourteen models from five different model families using different prompting methods. Our findings show that current LLMs struggle with robust personalization: even the strongest models (GPT-4.1, LLaMA3-70B) fails to maintain correctness in 5% of previously successful cases without personalization, while smaller models (e.g., 7B scale) can fail more than 20% of the time. Further analysis reveals that robustness is significantly affected by the nature of the query and the type of user preference. To mitigate these failures, we propose Pref-Aligner, a two-stage approach that improves robustness by an average of 25% across models. Our work highlights critical gaps in current evaluation practices and introduces tools and metrics to support more reliable, user-aligned LLM deployments.

Haonan Ge, Yiwei Wang, Ming-Hsuan Yang, Yujun Cai

Large Vision-Language Models (LVLMs) have shown strong performance across multimodal tasks. However, they often produce hallucinations—text that is inconsistent with visual input, due to the limited ability to verify information in different regions of the image. To address this, we propose **Multi-Region Fusion Decoding (MRFD)**, a training-free decoding method that improves factual grounding by modeling inter-region consistency. MRFD identifies salient regions using cross-attention, generates initial responses for each, and computes reliability weights based on Jensen-Shannon Divergence (JSD) among the responses. These weights guide a consistency-aware fusion of per-region predictions, using region-aware prompts inspired by Chain-of-Thought reasoning. Experiments across multiple LVLMs and benchmarks show that MRFD significantly reduces hallucinations and improves response factuality without requiring model updates.

Dhananjay Ashok, Ashutosh Chaubey, Hirona Jacqueline Arai, Jonathan May, Jesse Thomason

Through a controlled study, we identify a systematic deficiency in the multimodal grounding of Vision Language Models (VLMs). While VLMs can recall factual associations when provided a textual reference to an entity, their ability to do so is significantly diminished when the reference is visual instead. Forcing VLMs to rely on image representations of an entity halves their ability to recall factual knowledge, suggesting that VLMs struggle to link their internal knowledge of an entity with its image representation. We show that such linking failures are correlated with the expression of distinct patterns in model internal states, and that probes on these internal states achieve over 92% accuracy at flagging cases where the VLM response is unreliable. These probes can be applied, without retraining, to identify when a VLM will fail to correctly answer a question that requires an understanding of multimodal input. When used to facilitate selective prediction on a visual question answering task, the probes increase coverage by 7.87% (absolute) while also reducing the risk of error by 0.9% (absolute). Addressing the systematic, detectable deficiency is an important avenue in language grounding, and we provide informed recommendations for future directions.

Luca Rolshoven, Vishvaksenan Rasiah, Srinanda Brügger Bose, Sarah Hostettler, Lara Burkhalter, Matthias Stürmer, Joel Niklaus

Legal research depends on headnotes: concise summaries that help lawyers quickly identify relevant cases. Yet, many court decisions lack them due to the high cost of manual annotation. To address this gap, we introduce the Swiss Landmark Decisions Summarization (SLDS) dataset containing 20K rulings from the Swiss Federal Supreme Court, each with headnotes in German, French, and Italian. SLDS has the potential to significantly improve access to legal information and transform legal research in Switzerland. We fine-tune open models (Qwen2.5, Llama 3.2, Phi-3.5) and compare them to larger general-purpose and reasoning-tuned LLMs, including GPT-4o, Claude 3.5 Sonnet, and the open-source DeepSeek R1. Using an LLM-as-a-Judge framework, we find that fine-tuned models perform well in terms of lexical similarity, while larger models generate more legally accurate and coherent summaries. Interestingly, reasoning-focused models show no consistent benefit, suggesting that factual precision is more important than deep reasoning in this task. We release SLDS under a CC BY 4.0 license to support future research in cross-lingual legal summarization.

Rochana Prih Hastuti, Rian Adam Rajagede, Mansour Al Ghanim, Mengxin Zheng, Qian Lou

As large language models (LLMs) are adapted to sensitive domains such as medicine, their fluency raises safety risks, particularly regarding provenance and accountability. Watermarking embeds detectable patterns to mitigate these risks, yet its reliability in medical contexts remains untested. Existing benchmarks focus on detection-quality tradeoffs and overlook factual risks. In medical text, watermarking often reweights low-entropy tokens, which are highly predictable and often carry critical medical terminology. Shifting these tokens can cause inaccuracy and hallucinations, risks that prior general-domain benchmarks fail to capture.We propose a medical-focused evaluation workflow that jointly assesses factual accuracy and coherence. Using GPT-Judger and further human validation, we introduce the Factuality-Weighted Score (FWS), a composite metric prioritizing factual accuracy beyond coherence to guide watermarking deployment in medical domains. Our evaluation shows current watermarking methods substantially compromise medical factuality, with entropy shifts degrading medical entity representation. These findings underscore the need for domain-aware watermarking approaches that preserve the integrity of medical content.

Jiashi Lin, Changhong Jiang, Yixiao Wang, Xinyi Zhu, Zhongtian Hu, Wei Zhang

Knowledge Graph Embedding (KGE) seeks to learn latent representations of entities and relations to support knowledge-driven AI systems. However, existing KGE approaches often exhibit a growing discrepancy between the learned embedding space and the intrinsic structural semantics of the underlying knowledge graph. This divergence primarily stems from the over-reliance on geometric criteria for assessing triple plausibility, whose effectiveness is inherently limited by the sparsity of factual triples and the disregard of higher-order structural dependencies in the knowledge graph. To overcome this limitation, we introduce Structure-aware Calibration (SaCa), a versatile framework designed to calibrate KGEs through the integration of global structural patterns. SaCa designs two new components: (i) Structural Proximity Measurement, which captures multi-order structural signals from both entity and entity-relation perspectives; and (ii) KG-Induced Soft-weighted Contrastive Learning (KISCL), which assigns soft weights to hard-to-distinguish positive and negative pairs, enabling the model to better reflect nuanced structural dependencies. Extensive experiments on seven benchmarks demonstrate that SaCa consistently boosts performance across ten KGE models on link prediction and entity classification tasks with minimal overhead.

Radu Marinescu, Debarun Bhattacharjya, Junkyu Lee, Tigran T. Tchrakian, Javier Carnerero-Cano, Yufang Hou, Elizabeth M. Daly, Alessandra Pascale

Large language models (LLMs) have achieved remarkable success in generative tasks, yet they often fall short in ensuring the factual accuracy of their outputs thus limiting their reliability in real-world applications where correctness is critical. In this paper, we present FactReasoner, a novel neuro-symbolic based factuality assessment framework that employs probabilistic reasoning to evaluate the truthfulness of long-form generated responses. FactReasoner decomposes a response into atomic units, retrieves relevant contextual information from external knowledge sources, and models the logical relationships (e.g., entailment, contradiction) between these units and their contexts using probabilistic encodings. It then estimates the posterior probability that each atomic unit is supported by the retrieved evidence. Our experiments on both labeled and unlabeled benchmark datasets demonstrate that FactReasoner often outperforms state-of-the-art prompt-based methods in terms of factual precision and recall.

Sahal Shaji Mullappilly, Mohammed Irfan Kurpath, Sara Pieri, Saeed Yahya Alseiari, Shanavas Cholakkal, Khaled M Aldahmani, Fahad Shahbaz Khan, Rao Muhammad Anwer, Salman Khan, Timothy Baldwin 等

We introduce BiMediX2, a bilingual (Arabic-English) Bio-Medical EXpert Large Multimodal Model that supports text-based and image-based medical interactions. It enables multi-turn conversation in Arabic and English and supports diverse medical imaging modalities, including radiology, CT, and histology. To train BiMediX2, we curate BiMed-V, an extensive Arabic-English bilingual healthcare dataset consisting of 1.6M samples of diverse medical interactions. This dataset supports a range of medical Large Language Model (LLM) and Large Multimodal Model (LMM) tasks, including multi-turn medical conversations, report generation, and visual question answering (VQA). We also introduce BiMed-MBench, the first Arabic-English medical LMM evaluation benchmark, verified by medical experts. BiMediX2 demonstrates excellent performance across multiple medical LLM and LMM benchmarks, achieving state-of-the-art results compared to other open-sourced models. On BiMed-MBench, BiMediX2 outperforms existing methods by over 9% in English and more than 20% in Arabic evaluations. Additionally, it surpasses GPT-4 by approximately 9% in UPHILL factual accuracy evaluations and excels in various medical VQA, report generation, and report summarization tasks. Our trained models, instruction set, and source code are available at - https://github.com/mbzuai-oryx/BiMediX2

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.

Geng Zhang, Yizhou Ying, Sihang Jiang, Jiaqing Liang, Guanglei Yue, Yifei Fu, Hailin Hu, Yanghua Xiao

Despite the rapid development of large language models (LLMs), existing benchmark datasets often focus on low-level cognitive tasks, such as factual recall and basic comprehension, while providing limited coverage of higher-level reasoning skills, including analysis, evaluation, and creation. In this work, we systematically assess the cognitive depth of popular LLM benchmarks using Bloom’s Taxonomy to evaluate both the cognitive and knowledge dimensions.Our analysis reveals a pronounced imbalance: most datasets concentrate on “Remembering” and “Understanding”, with metacognitive and creative reasoning largely underrepresented. We also find that incorporating higher-level cognitive instructions into the current instruction fine-tuning process improves model performance. These findings highlight the importance of future benchmarks incorporating metacognitive evaluations to more accurately assess and enhance model performance.

Yunhai Hu, Yilun Zhao, Chen Zhao, Arman Cohan

We introduce MCTS-RAG, a novel approach that enhances the reasoning capabilities of small language models on knowledge-intensive tasks by leveraging retrieval-augmented generation (RAG) to provide relevant context and Monte Carlo Tree Search (MCTS) to refine reasoning paths. MCTS-RAG dynamically integrates retrieval and reasoning through an iterative decision-making process. Unlike standard RAG methods, which typically retrieve information independently from reasoning and thus integrate knowledge suboptimally, or conventional MCTS reasoning, which depends solely on internal model knowledge without external facts, MCTS-RAG combines structured reasoning with adaptive retrieval. This integrated approach enhances decision-making, reduces hallucinations, and ensures improved factual accuracy and response consistency. The experimental results on multiple reasoning and knowledge-intensive datasets datasets (ComplexWebQA, GPQA, and FoolMeTwice) show that our method enables small-scale LMs to achieve performance comparable to frontier LLMs like GPT-4o by effectively scaling inference-time compute, setting a new standard for reasoning in small-scale models.

Deema Alnuhait, Neeraja Kirtane, Muhammad Khalifa, Hao Peng

Language models (LMs) hallucinate. We inquire: Can we detect and mitigate hallucinations before they happen? This work answers this research question in the positive, by showing that the internal representations of LMs provide rich signals that can be used for this purpose. We introduce FactCheckmate, which preemptively detects hallucinations by learning a classifier that predicts whether the LM will hallucinate, based on the model’s hidden states produced over the inputs, before decoding begins. If a hallucination is detected, FactCheckmate then intervenes by adjusting the LM’s hidden states such that the model will produce more factual outputs. FactCheckmate provides fresh insights that the inner workings of LMs can be revealed by their hidden states. Practically, both its detection and mitigation models are lightweight, adding little inference overhead; FactCheckmate proves a more efficient approach for mitigating hallucinations compared to many post-hoc alternatives. We evaluate FactCheckmate over LMs of different scales and model families (including Llama, Mistral, Qwen and Gemma), across a variety of QA datasets from different domains. Our results demonstrate the effectiveness of FactCheckmate, achieving over 70% preemptive detection accuracy. On average, outputs generated by LMs with intervention are 34.4% more factual compared to those without.

Siying Zhou, Yiquan Wu, Hui Chen, Xueyu Hu, Kun Kuang, Adam Jatowt, Chunyan Zheng, Fei Wu

Legal claims refer to the plaintiff’s demands in a case and are essential to guiding judicial reasoning and case resolution. While many works have focused on improving the efficiency of legal professionals, the research on helping non-professionals (e.g., plaintiffs) remains unexplored. This paper explores the problem of legal claim generation based on the given case’s facts. First, we construct ClaimGen-CN, the first dataset for Chinese legal claim generation task, from various real-world legal disputes. Additionally, we design an evaluation metric tailored for assessing the generated claims, which encompasses two essential dimensions: factuality and clarity. Building on this, we conduct a comprehensive zero-shot evaluation of state-of-the-art general and legal-domain large language models. Our findings highlight the limitations of the current models in factual precision and expressive clarity, pointing to the need for more targeted development in this domain. To encourage further exploration of this important task, we will make the dataset publicly available.

Yangning Li, Weizhi Zhang, Yuyao Yang, Wei-Chieh Huang, Yaozu Wu, Junyu Luo, Yuanchen Bei, Henry Peng Zou, Xiao Luo, Yusheng Zhao 等

Retrieval-Augmented Generation (RAG) lifts the factuality of Large Language Models (LLMs) by injecting external knowledge, yet it falls short on problems that demand multi-step inference; conversely, purely reasoning-oriented approaches often hallucinate or mis-ground facts. This survey synthesizes both strands under a unified reasoning-search perspective. We first map how advanced reasoning optimizes each stage of RAG (Reasoning-Enhanced RAG). Then, we show how retrieved knowledge of different type supply missing premises and expand context for complex inference (RAG-Enhanced Reasoning). Finally, we spotlight emerging Synergized RAG-Reasoning frameworks, where (agentic) LLMs iteratively interleave search and thought to achieve state-of-the-art performance across knowledge-intensive benchmarks. We categorize methods, datasets, and open challenges, and outline research avenues toward deeper RAG-Reasoning systems that are more effective, multimodally-adaptive, trustworthy, and human-centric.