论文检索

Yuto Nishida, Naoki Shikoda, Yosuke Kishinami, Ryo Fujii, Makoto Morishita, Hidetaka Kamigaito, Taro Watanabe

Understanding what kinds of factual knowledge large language models (LLMs) memorize is essential for evaluating their reliability and limitations.Entity-based QA is a common framework for analyzing non-verbatim memorization, but typical evaluations query each entity using a single canonical surface form, making it difficult to disentangle fact memorization from access through a particular name.We introduce RedirectQA, an entity-based QA dataset that uses Wikipedia redirect information to associate Wikidata factual triples with categorized surface forms for each entity, including alternative names, abbreviations, spelling variants, and common erroneous forms.Across 13 LLMs, we examine surface-conditioned factual memorization and find that prediction outcomes often change when only the entity surface form changes.This inconsistency is category-dependent: models are more robust to minor orthographic variations than to larger lexical variations such as aliases and abbreviations.Frequency analyses further suggest that both entity- and surface-level frequencies are associated with accuracy, and that entity frequency often contributes beyond surface frequency.Overall, factual memorization appears neither purely surface-specific nor fully surface-invariant, highlighting the importance of surface-form diversity in evaluating non-verbatim memorization.

Tejas Anvekar, Junha Park, Aparna Garimella, Vivek Gupta

Evaluating the quality of tables generated by large language models (LLMs) remains an open challenge: existing metrics either flatten tables into text, ignoring structure, or rely on fixed references that limit generalization. We present TabReX, a reference-less, property-driven framework for evaluating tabular generation via graph-based reasoning. TabReX converts both source text and generated tables into canonical knowledge graphs, aligns them through an LLM-guided matching process, and computes interpretable, rubric-aware scores that quantify structural and factual fidelity. The resulting metric provides controllable trade-offs between sensitivity and specificity, yielding human-aligned judgments and cell-level error traces. To systematically asses metric robustness, we introduce TabReX-Bench, a large-scale benchmark spanning six domains and twelve planner-driven perturbation types across three difficulty tiers. Empirical results show that TabReX achieves the highest correlation with expert rankings, remains stable under harder perturbations, and enables fine-grained model-vs-prompt analysis establishing a new paradigm for trustworthy, explainable evaluation of structured generation systems.

Arka Dutta, Sujan Dutta, Rijul Magu, Soumyajit Datta, Munmun De Choudhury, Ashiqur R. KhudaBukhsh

Hallucinations pose a critical challenge to the real-world deployment of large language models (LLMs) in high-stakes domains. In this paper, we present a framework for stress testing factual fidelity in LLMs in the presence of adversarial nudge. Our framework consists of three steps. First, we instruct the LLM to produce sets of truths and lies consistent with the closed domain in question. Next, we instruct the LLM to verify the same set of assertions as truths and lies consistent with the same closed domain. Finally, we test the robustness of the LLM against the lies generated (and verified) by itself. Our extensive evaluation, conducted using five widely known proprietary and six open LLMs across two closed domains of popular movies and novels, reveals a wide range of susceptibility to adversarial nudges: even among the strongest proprietary LLMs, Claude exhibits strong resilience, GPT and Grok demonstrate moderate resilience, while Gemini and DeepSeek show weak resilience and open models fall short significantly.

Hail Hochman, Natalie Shapira, Yoav Goldberg

Large language models (LLMs) store and recall factual knowledge, yet the precise mechanism of how entity representations are transformed to enable specific attribute retrieval remains underexplored. In this work, we investigate this mechanism through the lens of an “attribute-computation path”—a sequence of computational steps over the entity representation required to elicit a target attribute. We then propose an iterative patching protocol to identify a minimal subset of layers necessary for this computation. Applying our method to LLaMA 3.1 8B and Qwen 3 8B, we find that these paths are non-contiguous, often skipping layers, and that models possess multiple, functionally-equivalent paths for the same entity and fact, highlighting a high degree of redundancy in attribute computation. This implies that knowledge computation is highly distributed, potentially explaining the localization-editing mismatch and suggesting that knowledge storage and retrieval in LLMs is far from being well understood.

Xingyu Sui, Yanyan Zhao, Yulin Hu, Jiahe Guo, Weixiang Zhao, Bing Qin

Emotional Support Conversation requires not only affective expression but also grounded instrumental support to provide trustworthy guidance. However, existing ESC systems and benchmarks largely focus on affective support in text-only settings, overlooking how external tools can enable factual grounding and reduce hallucination in multi-turn emotional support. We introduce **TEA-Bench**, the first interactive benchmark for evaluating tool-augmented agents in ESC, featuring realistic emotional scenarios, an MCP-style tool environment, and process-level metrics that jointly assess the quality and factual grounding of emotional support. Experiments on nine LLMs show that tool augmentation generally improves emotional support quality and reduces hallucination, but the gains are strongly capacity-dependent: stronger models use tools more selectively and effectively, while weaker models benefit only marginally. We further release **TEA-Dialog**, a dataset of tool-enhanced ESC dialogues, and find that supervised fine-tuning improves in-distribution support but generalizes poorly. Our results underscore the importance of tool use in building reliable emotional support agents.

Javier Carnerero-Cano, Massimiliano Pronesti, Radu Marinescu, Tigran T. Tchrakian, James Barry, Jasmina Gajcin, Yufang Hou, Alessandra Pascale, Elizabeth M. Daly

Large language models (LLMs) are widely used in knowledge-intensive applications but often generate factually incorrect responses. A promising approach to rectify these flaws is correcting LLMs using feedback. Therefore, in this paper, we introduce FactCorrector, a new post-hoc correction method that adapts across domains without retraining and leverages structured feedback about the factuality of the original response to generate a correction. To support rigorous evaluations of factuality correction methods, we also develop the VELI5 benchmark, a novel dataset containing systematically injected factual errors and ground-truth corrections. Experiments on VELI5 and several popular long-form factuality datasets show that the FactCorrector approach significantly improves factual precision while preserving relevance, outperforming strong baselines. We release our code at https://ibm.biz/factcorrector.

Ashish Mishra, Tarun Kumar, Arpit Shah, Suparna Bhattacharya, Martin Foltin

Hallucinated object mentions remain a persistent failure mode of vision–language models (VLMs) across generation tasks such as image captioning and visual question answering: outputs may be fluent yet include entities not supported by visual evidence. Existing mitigation approaches often reduce hallucinations at the cost of degraded generation quality or require expensive retraining and task-specific supervision. We introduce CEBC, a lightweight, training-free framework for low-hallucination vision–language generation based on conformal evidence-bounded minimal editing. CEBC first produces a strong base output (via greedy decoding or best-of-K sampling), then applies an evidence-bounded editing step that minimally revises or suppresses unsupported object mentions using constraints derived from an external visual detector. Crucially, the evidence threshold is conformally calibrated on a small held-out set via quantiles of detector confidence scores, enabling explicit and controllable hallucination risk at test time.To balance factuality and informativeness, we further introduce a risk-first, quality-aware selection rule that prioritizes evidence-consistent generations while regularizing unnecessary length or lexical drift. Extensive experiments on MS-COCO and GQA for image captioning, and POPE for VQA evaluation across multiple VLMs demonstrate that CEBC consistently reduces hallucination rates(CHAIR_S, CHAIR_I, POPE) while maintaining or improving standard generation quality metrics (CIDEr, BLEU, CLIPScore). CEBC establishes a stronger factuality–quality Pareto frontier without any additional model training or access to paired supervision beyond an off-the-shelf detector.

Yash Kumar Atri, Steven L. Johnson, Thomas Hartvigsen

Language models are increasingly deployed in interactive settings where users reason about facts over time rather than in isolation.In such scenarios, correct behavior requires models to maintain and update implicit temporal assumptions established earlier in a conversation. We study this challenge through the lens of temporal scope stability: the ability to preserve, override, or transfer time-scoped factual context across dialogue turns. We introduce ChronoScope, a large-scale diagnostic benchmark designed to isolate temporal scope behavior in controlled multi-turn interactions, comprising over one million deterministically generated question chains grounded in Wikidata. ChronoScope evaluates whether models can correctly retain inferred temporal scope when follow-up questions omit explicit time references, spanning implicit carryover, explicit scope switching, cross-entity transfer, and longer temporal trajectories.Through extensive evaluation of state-of-the-art language models, we find that temporal scope stability is frequently violated in controlled multi-turn settings, with models often drifting toward present-day assumptions despite correct underlying knowledge.These failures intensify with interaction length and persist even under oracle context conditions, revealing a gap between single-turn factual accuracy and coherent temporal reasoning under sequential interaction.We make our dataset and evaluation suite publicly available at https://github.com/yashkumaratri/ChronoScope.

Minzhu Tu, Shiyu Ni, Keping Bi

Large language models (LLMs) are increasingly adopted as scalable judges for open-ended generation, yet how they form judgments remains insufficiently understood. Meanwhile, modern LLMs frequently produce answers accompanied by explicit reasoning, making reasoning chains a natural but understudied source of information for model-based evaluation. This work takes a first step toward understanding how exposing reasoning influences LLM-based judgment. Empirical results across factual question-answering (QA) and mathematical datasets show that the presence of reasoning substantially alters judgment behavior, with clear differences across judge capabilities. Weaker judges become more likely to accept incorrect answers when reasoning is present, suggesting over-reliance on persuasive explanations. In contrast, stronger judges exhibit more selective behavior and, in some cases, achieve higher judgment accuracy by leveraging reasoning content. Further analysis reveals that both reasoning fluency and factuality critically shape judgment outcomes. Together, these findings suggest that examining how models interpret reasoning is essential for understanding and improving LLM-based evaluation, with broader implications for the design of reliable automatic judges and evaluation protocols.

Hao Huang, JiaTang Luo, Ruihua Zhou, Yunpeng Li, Yuling Liu

While Retrieval-Augmented Generation (RAG) systems are designed to enhance factual fidelity by grounding LLMs in provided sources, the application of current watermarking techniques creates a paradoxical failure mode. These methods, being inherently fact-agnostic, force the model to deviate from the very source documents it is supposed to follow. This leads to “faithfulness hallucinations"—a critical flaw where the generated output contradicts its own grounding context. Consequently, these watermarks undermine the core value of RAG, rendering even the most secure schemes untrustworthy for high-stakes applications. To resolve this RAG-specific conflict, we introduce the Dual Factual Shield (DFS) framework, a novel architecture designed to enforce knowledge loyalty. The DFS framework employs a defense-in-depth strategy through two synergistic layers: a source-anchored algorithmic safeguard that shields critical terms from the retrieved context, and prompt-based semantic guidance that protects against factual corruption. To demonstrate its effectiveness, we enhance a state-of-the-art, spoofing-aware contrastive watermarking baseline with our framework. Experiments show that our framework drastically reduces the Knowledge Corruption Rate (KCR)—a new metric we introduce—while preserving its original high security and robustness. This work establishes a new paradigm for watermarking, evolving it from merely secure to truly trustworthy. We demonstrate that traceability and truth can, and must, coexist, paving the way for the responsible deployment of traceable AI in knowledge-critical domains.

Yangjunqi, Yuecong Min, Jie Zhang, Shiguang Shan, Xilin Chen

Despite rapid progress, Video Large Language Models (Video-LLMs) remain unreliable due to hallucinations, which are outputs that contradict either video evidence (faithfulness) or verifiable world knowledge (factuality).Existing benchmarks provide limited coverage of factuality hallucinations and predominantly evaluate models only in clean settings.We introduce INFACT, a diagnostic benchmark comprising 9,800 QA instances with fine-grained taxonomies for faithfulness and factuality, spanning real and synthetic videos.INFACT evaluates models in four modes: Base (clean), Visual Degradation, Evidence Corruption, and Temporal Intervention for order-sensitive items.Reliability under induced modes is quantified using Resist Rate (RR) and Temporal Sensitivity Score (TSS).Experiments on 14 representative Video-LLMs reveal that higher Base-mode accuracy does not reliably translate to higher reliability in the induced modes, with evidence corruption reducing stability and temporal intervention yielding the largest degradation.Notably, many open-source baselines exhibit near-zero TSS on factuality, indicating pronounced temporal inertia on order-sensitive questions.

Ruifeng Yuan, Wanxing Chang, Weiwei Cao, Bowen Shi, Zhongyu Wei, Ling Zhang, Jianpeng Zhang

The evaluation of generated reports remains a critical challenge in Computed Tomography (CT) report generation, due to the large volume of text, the diversity and complexity of findings, and the presence of fine-grained, disease-oriented attributes. Conventional evaluation metrics offer only coarse measures of lexical overlap or entity matching and fail to reflect the granular diagnostic accuracy required for clinical use. To address this gap, we propose CT-FineBench, a benchmark built from CT-RATE and Merlin to evaluate the fine-grained factual consistency of CT reports, constructed from CT-RATE and Merlin. Our benchmark is constructed through a meticulous, Question-Answering (QA) based process: first, we identify and structure key, finding-specific clinical attributes (e.g., location, size, margin). Second, we systematically transform these attributes into a QA dataset, where questions probe for specific clinical details grounded in gold-standard reports. The evaluation protocol for CT-FineBench involves using this QA dataset to query a machine-generated report and scoring the correctness of the answers. This allows for a comprehensive, interpretable, and clinically-relevant assessment, moving beyond superficial lexical overlap to pinpoint specific clinical errors. Experiments show that CT-FineBench correlates better with expert clinical assessment and is substantially more sensitive to fine-grained factual errors than prior metrics.

Zehua Cheng, Wei Dai, Jiahao Sun, Thomas Lukasiewicz

The large language models offer a scaleable solution for the generation of synthetic data faced with a trade-off between maintaining the diversity of generation and achieving factually accurate results. This paper introduces Graphsynth, a framework which leverages a probabilistic factor graph modeling the universe of attributes. The framework leverages a high-level schema mapping compiled into efficient hard masks during the decoding phase for maintaining the syntactic truth and a span-synchronized verifier for dismissing logical contradictions at the decode time. The experiments conducted on biomedical, legal, and generic domains show that the method outperforms the state-of-the-art baselines with a structural integrity approaching perfection, a coverage of around 94% attributes on the factor graph solution, and a boost in performance on downstream tasks such as +17.9% on TruthfulQA.

Fangda Ye, Kuicai Dong, Xie Zhifei, Yuxin Hu, Yihang Yin, Shurui Huang, Shikai Dong, Chen Zhang, Jianzhu Bao, Shuicheng Yan

Recent agentic search frameworks enable deep research via iterative planning and retrieval, reducing hallucinations and enhancing factual grounding. However, they remain text-centric, overlooking the multimodal evidence that characterizes real-world expert reports. We introduce a pressing task: multimodal long-form generation. Accordingly, we propose Deep-Reporter, a unified agentic framework for grounded multimodal long-form generation. It orchestrates: (i) Agentic Multimodal Search and Filtering to retrieve and filter textual passages and information-dense visuals; (ii) Checklist-Guided Incremental Synthesis to ensure coherent image-text integration and optimal citation placement; and (iii) Recurrent Context Management to balance long-range coherence with local fluency. We develop a rigorous curation pipeline producing 8K high-quality agentic traces for model optimization. We further introduce M^2LongBench, a comprehensive testbed comprising 247 research tasks across 9 domains and a stable multimodal sandbox. It enables unified multimodal assessment, fair comparison, and accessible evaluation without commercial APIs. Extensive experiments demonstrate that long-form multimodal generation is a challenging task, especially in multimodal selection and integration, and effective post-training can bridge the gap. Our code is available at https://github.com/fangda-ye/Deep-Report.

Ashley Lewis, Andrew Perrault, Eric Fosler-Lussier, Michael White

Hallucination—defined here as generated statements unsupported or contradicted by available evidence or conversational context—remains a major obstacle to using conversational AI systems in settings that demand factual reliability. Existing metrics evaluate isolated responses or treat unverifiable content as errors, limiting their use for multi-turn dialogue. We introduce VISTA (Verification In Sequential Turn-based Assessment), a framework for evaluating conversational factuality via claim-level verification and sequential consistency tracking. VISTA decomposes each turn into atomic claims, verifies them against trusted sources and dialogue history, and categorizes unverifiable statements (subjective, contradicted, lacking evidence, or abstaining). Across eight large language models and four dialogue factuality benchmarks (Ais, Begin, FaithDial, and Fade), VISTA substantially improves hallucination detection over FActScore and LLM-as-Judge baselines. Human evaluation confirms that VISTA’s decomposition improves annotator agreement and reveals inconsistencies in existing benchmarks. By modeling factuality as a dynamic property of conversation, VISTA offers a more transparent, human-aligned measure of truthfulness in dialogue systems.

Miao Xie, Xiao Zhang, Yi Li, Chunli Lv

Retrieval-Augmented Generation (RAG) has been proposed to mitigate hallucinations in large language models (LLMs), where generated outputs may be factually incorrect. However, existing RAG approaches predominantly rely on vector similarity for retrieval, which is prone to semantic noise and fails to ensure that generated responses fully satisfy the complex conditions specified by factual queries, often leading to incorrect answers. To address this challenge, we introduce a novel research problem, named Exact Retrieval Problem (ERP). To the best of our knowledge, this is the first problem formulation that explicitly incorporates structural information into RAG for factual questions to satisfy all query conditions. For this novel problem, we propose Structure Guided Retrieval-Augmented Generation (SG-RAG), which models the retrieval process as an embedding-based subgraph matching task, and uses the retrieved topological structures to guide the LLM to generate answers that meet all specified query conditions. To facilitate evaluation of ERP, we construct and publicly release Exact Retrieval Question Answering (ERQA), a large-scale dataset comprising 120{,}000 fact-oriented QA pairs, each involving complex conditions, spanning 20 diverse domains. The experimental results demonstrate that SG-RAG significantly outperforms strong baselines on ERQA, delivering absolute improvements from 20.68 to 50.88 points across all evaluation metrics, while maintaining reasonable computational overhead.

Nishanth Sridhar Nakshatri, Eylon Caplan, Rajkumar Pujari, Dan Goldwasser

Health influencers play a growing role in shaping public beliefs, yet their content is often conveyed through conversational narratives and rhetorical strategies rather than explicit factual claims. As a result, claim-centric verification methods struggle to capture the pragmatic meaning of influencer discourse. In this paper, we propose TAIGR (Takeaway Argumentation Inference with Grounded References), a structured framework designed to analyze influencer discourse, which operates in 3 stages: (1) identifying the core influencer recommendation–takeaway; (2) constructing an argumentation graph that captures influencer justification for the takeaway; (3) performing factor graph-based probabilistic inference to validate the takeaway. We evaluate TAIGR on a content validation task over influencer video transcripts on health, showing that accurate validation requires modeling the discourse’s pragmatic and argumentative structure rather than treating transcripts as flat collections of claims.

Mutian Bao, Qiuyi Qi, Tian Liang, Jinjian Zhang, Wei Zhou, Ming Kong, Linjian Mo, Qiang Zhu

Current evaluations of geospatial reasoning in LLMs are frequently impeded by the entanglement of factual recall and spatial logic, which often obscures the models’ true capabilities in complex city-scale environments. To address this, we introduce UrbanGeoEval, a comprehensive benchmark featuring a dual-module framework designed to disentangle these competencies. The Knowledge Module assesses urban memory via scalable map-based queries, while the Reasoning Module isolates pure logical inference across 3,148 realistic tasks by providing necessary geospatial context. Unlike prior benchmarks that hand the model pre-computed spatial text, UrbanGeoEval provides raw geometry and forces the model to act as a spatial computing engine. Our evaluation methodology introduces a reliable hybrid pipeline that merges deterministic programmatic checks with an LLM-as-a-Judge, achieving expert-level evaluation accuracy. Extensive experiments on 18 widely used LLMs uncover critical insights: (1) models exhibit severe geographic biases and resolution gaps; (2) failures in complex multi-hop tasks often stem from brittle foundational spatial skills rather than high-level logic deficits. UrbanGeoEval provides a precise diagnostic tool for advancing urban geospatial intelligence in LLMs.

Ponhvoan Srey, Xiaobao Wu, Cong-Duy T Nguyen, Anh Tuan Luu

Uncertainty estimation is a promising approach to detect hallucinations in large language models (LLMs). Recent approaches commonly depend on model internal states to estimate uncertainty. However, they suffer from strict assumptions on how hidden states should evolve across layers, and from information loss by solely focusing on last or mean tokens. To address these issues, we present Sequential Internal Variance Representation (SIVR), a supervised hallucination detection framework that leverages token-wise, layer-wise features derived from hidden states. SIVR adopts a more basic assumption that uncertainty manifests in the degree of dispersion or variance of internal representations across layers, rather than relying on specific assumptions, which makes the method model and task agnostic. It additionally aggregates the full sequence of per-token variance features, learning temporal patterns indicative of factual errors and thereby preventing information loss. Experimental results demonstrate SIVR consistently outperforms strong baselines. Most importantly, SIVR enjoys stronger generalisation and avoids relying on large training sets, highlighting the potential for practical deployment.

Jingwen Pu, Mingjun Shi, Xinrui Ren, Yizhe Wang, Xinyu Zhang, Zhaokun Wang, Kun She

Large Language Models (LLMs) exhibit powerful capabilities but inevitably memorize sensitive information, raising privacy, copyright, and safety concerns. Existing LLM unlearning methods typically rely on updating model parameters. While effective, they are often limited in real-world scenarios: fine-tuning large-scale models is costly, may introduce potential irreversible risks, and depends on both forget and retain datasets, which are often difficult to obtain in full. To address these challenges, an ideal solution is to achieve unlearning at inference time. To this end, we propose SEGUE, a training-free, plug-and-play inference-time unlearning strategy. SEGUE employs a probe to detect queries involving forgettable concepts and applies entropy-guided decoding to suppress target knowledge, enabling controllable non-factual generation while preserving overall model capabilities. Experiments on the MUSE, RWKU, and WMDP datasets, covering copyright, entity, and potential-risk knowledge, show that SEGUE effectively balances sensitive knowledge suppression and generation quality, outperforming existing most inference-time unlearning methods.