Fresh from the feed

arXiv:2508.13953v2 Announce Type: replace Abstract: In the hospitality industry, understanding the factors that drive customer review ratings is critical for improving guest satisfaction and business performance. This work proposes ReviewGraph for Review Rating Prediction (RRP), a novel framework that transforms textual customer reviews into knowledge graphs by extracting (subject, predicate, object) triples and associating sentiment scores. Using graph embeddings (Node2Vec) and sentiment features, the framework predicts review rating scores through machine learning classifiers. We compare ReviewGraph performance with traditional NLP baselines (such as Bag of Words, TF-IDF, and Word2Vec) and large language models (LLMs), evaluating them in the HotelRec dataset. In comparison to the state of the art literature, our proposed model performs similar to their best performing model but with lower computational cost (without ensemble). While ReviewGraph achieves comparable predictive performance to LLMs and outperforms baselines on agreement-based metrics such as Cohen's Kappa, it offers additional advantages in interpretability, visual exploration, and potential integration into Retrieval-Augmented Generation (RAG) systems. This work highlights the potential of graph-based representations for enhancing review analytics and lays the groundwork for future research integrating advanced graph neural networks and fine-tuned LLM-based extraction methods. We will share ReviewGraph output and platform open-sourced on our GitHub page https://github.com/aaronlifenghan/ReviewGraph

arXiv:2508.14031v2 Announce Type: replace Abstract: Beyond simple text generation, Large Language Models (LLMs) have evolved into agentic systems capable of planning and interacting with external tools to solve complex tasks. This evolution involves fine-tuning LLMs on agent-specific tasks to enhance their proficiency. However, safety concerns are frequently overlooked during this fine-tuning process. In this work, we show that aligned LLMs can become unintentionally misaligned, leading to a higher likelihood of executing harmful tasks and a reduced tendency to refuse them when fine-tuned to execute agentic tasks. To address these safety challenges, we propose Prefix INjection Guard (PING), a simple yet effective method that prepends automatically generated natural language prefixes to agent responses, guiding them to refuse harmful requests while preserving performance on benign tasks. Specifically, we introduce an iterative approach that alternates between (1) generating candidate prefixes and (2) selecting those that optimize both task performance and refusal behavior. Experimental results demonstrate that PING significantly enhances the safety of fine-tuned LLM agents without sacrificing their effectiveness. PING consistently outperforms existing prompting approaches across diverse benchmarks in both web navigation and code generation tasks. Our analysis of internal hidden states via linear probes reveals that prefix tokens are crucial for behavior modification, explaining the performance gains. WARNING: This paper contains contents that are unethical or offensive in nature.

arXiv:2509.00974v3 Announce Type: replace Abstract: Medical question answering requires advanced reasoning that integrates domain knowledge with logical inference. However, existing large language models (LLMs) often generate reasoning chains that lack factual accuracy and clinical reliability. We propose Ranked Preference Reinforcement Optimization (RPRO), a novel framework that combines reinforcement learning with preference-driven reasoning refinement to enhance clinical chain-of-thought (CoT) performance. RPRO distinguishes itself from prior approaches by employing task-adaptive reasoning templates and a probabilistic evaluation mechanism that aligns model outputs with established clinical workflows, while automatically identifying and correcting low-quality reasoning chains. Unlike traditional pairwise preference methods, RPRO introduces a groupwise ranking optimization based on the Bradley--Terry model and incorporates KL-divergence regularization for stable training. Experiments on PubMedQA, MedQA-USMLE, and a real-world clinical dataset from Far Eastern Memorial Hospital (FEMH) demonstrate consistent improvements over strong baselines. Remarkably, our 2B-parameter model outperforms much larger 7B--20B models, including medical-specialized variants. These findings demonstrate that combining preference optimization with quality-driven refinement provides a scalable and clinically grounded approach to building more reliable medical LLMs.

arXiv:2509.12440v2 Announce Type: replace Abstract: Deploying Large Language Models (LLMs) in medical applications requires fact-checking capabilities to ensure patient safety and regulatory compliance. We introduce MedFact, a challenging Chinese medical fact-checking benchmark with 2,116 expert-annotated instances from diverse real-world texts, spanning 13 specialties, 8 error types, 4 writing styles, and 5 difficulty levels. Construction uses a hybrid AI-human framework where iterative expert feedback refines AI-driven, multi-criteria filtering to ensure high quality and difficulty. We evaluate 20 leading LLMs on veracity classification and error localization, and results show models often determine if text contains errors but struggle to localize them precisely, with top performers falling short of human performance. Our analysis reveals the "over-criticism" phenomenon, a tendency for models to misidentify correct information as erroneous, which can be exacerbated by advanced reasoning techniques such as multi-agent collaboration and inference-time scaling. MedFact highlights the challenges of deploying medical LLMs and provides resources to develop factually reliable medical AI systems.

arXiv:2509.23188v2 Announce Type: replace Abstract: Large Language Model (LLM)-powered multi-agent systems (MAS) have rapidly advanced collaborative reasoning, tool use, and role-specialized coordination in complex tasks. However, reliability-critical deployment remains hindered by a systemic failure mode: hierarchical compliance under instruction conflicts (system-user, peer-peer), where agents misprioritize system-level rules in the presence of competing demands. Moreover, widely used macro-level metrics (e.g., pass@k) obscure these micro-level violations and offer little actionable guidance for remedy. In this work, we present a full-stack, three-stage framework: (1) Diagnose - Contextualized Role Adherence Score (CRAS), a query-wise, context-aware scoring metric that decomposes role adherence into four measurable dimensions; (2) Localize - attention drift analysis revealing that instruction conflicts are resolved by attention heads that are largely concentrated in middle layers; (3) Align - Surgical Alignment of Instruction Layers (SAIL), which installs LoRA only on the localized focal layers and optimizes a token-weighted DPO-style preference objective that credits tokens by their focal attentional contribution. Across standard benchmarks and MAS frameworks, our surgical approach improves instruction hierarchy compliance (e.g., +5.60% with AutoGen on MedQA) without full-model finetuning.

arXiv:2509.24130v2 Announce Type: replace Abstract: The performance of Large Language Models (LLMs) hinges on carefully engineered prompts. However, prevailing prompt optimization methods, ranging from heuristic edits and reinforcement learning to evolutionary search, primarily target point-wise accuracy. They seldom enforce paraphrase invariance or searching stability, and therefore cannot remedy this brittleness in practice. Automated prompt search remains brittle: small, semantically preserving paraphrases often cause large performance swings. We identify this brittleness as the textual sharpness of the prompt landscape. In this work, we provide the first formal treatment of textual sharpness in the discrete, semantic space of prompts, together with an operational robustness criterion over a semantic neighborhood; the design is black-box or API-only, requiring no gradients to update the model's parameters. Then we introduce TARE (Textual Sharpness-Aware Evolving), a derivative-free framework that alternates between an inner, sampling-based adversarial search that stresses a prompt with hard paraphrases and an outer, robust selection that prefers candidates whose neighborhoods remain strong. We further propose ATARE, which learns anisotropic weights to shape the semantic neighborhood and adapts its radius over time to balance exploration and fidelity. Diverse tasks evaluate our methods, whose design for minimizing textual sharpness gap leads to prompts that preserve accuracy under paraphrasing, outperforming accuracy-only prompt search while remaining computationally practical.

arXiv:2509.24189v2 Announce Type: replace Abstract: Understanding how user preference evolves over time is a fundamental challenge central to modern digital ecosystems, for which Large Language Models (LLMs) are an increasingly prominent and popular approach due to their ability to comprehend the rich semantic context within behavioral data. A common practice is to use LLMs to predict a user's next action by directly generating a ranked list of preferred items. Although effective for short-term prediction, the end-to-end generation paradigm inherently limits personalization. Its opaque decision-making process obscures holistic user profiling and exacerbates popularity bias. To address these limitations, we propose Preference Evolution Tracking (PET), a framework that reframes the task as inferring a dynamic probability distribution over a stable and interpretable lattice of preference clusters. By applying logit-probing and generative classification techniques, PET infers a user's preference as a probability distribution, enabling transparent preference learning. On public benchmarks (Yelp, MovieLens), PET improves ranking quality by up to 40% in NDCG over direct generation baselines. On a large-scale, real-world dataset from a short-video platform, it excels at ranking long-tail contents, significantly outperforming a SOTA production model by 7 times in the NDCG score. Ultimately, PET transforms the user profile model from direct preference list generation to a transparent distributional preference mapping, paving the way for more explainable, fair, and diverse personalization systems.

arXiv:2510.00829v2 Announce Type: replace Abstract: \textbf{RE}trieval-\textbf{A}ugmented \textbf{L}LM-based \textbf{M}achine \textbf{T}ranslation (REAL-MT) shows promise for knowledge-intensive tasks like idiomatic translation, but its reliability under noisy retrieval contexts remains poorly understood despite this being a common challenge in real-world deployment. To address this gap, we propose a noise synthesis framework and new metrics to evaluate the robustness of REAL-MT systematically. Using this framework, we instantiate REAL-MT with Qwen-series models, including standard LLMs and large reasoning models (LRMs) with enhanced reasoning, and evaluate their performance on idiomatic translation across high-, medium-, and low-resource language pairs under synthesized noise. Our results show that low-resource language pairs, which rely more heavily on retrieved context, degrade more severely under noise than high-resource ones and often produce nonsensical translations. Although LRMs possess enhanced reasoning capabilities, they show no improvement in error correction and are even more susceptible to noise, tending to rationalize incorrect contexts. We find that this stems from an attention shift away from the source idiom to noisy content, while confidence increases despite declining accuracy, indicating poor calibration. To mitigate these issues, we investigate training-free and fine-tuning strategies, which improve robustness at the cost of performance in clean contexts, revealing a fundamental trade-off. Our findings highlight the limitations of current approaches, underscoring the need for self-verifying integration mechanisms.

arXiv:2510.08956v2 Announce Type: replace Abstract: We study how open source communities describe participation and control through version controlled governance documents. Using a corpus of 710 projects with paired snapshots, we parse text into actors, rules, actions, and objects, then group them and measure change with entropy for evenness, richness for diversity, and Jensen Shannon divergence for drift. Projects define more roles and more actions over time, and these are distributed more evenly, while the composition of rules remains stable. These findings indicate that governance grows by expanding and balancing categories of participation without major shifts in prescriptive force. The analysis provides a reproducible baseline for evaluating whether future AI mediated workflows concentrate or redistribute authority.

arXiv:2510.14128v2 Announce Type: replace Abstract: Computational gastronomy increasingly relies on diverse, high-quality recipe datasets to capture regional culinary traditions. Although there are large-scale collections for major languages, Macedonian recipes remain under-represented in digital research. In this work, we present the first systematic effort to construct a Macedonian recipe dataset through web scraping and structured parsing. We address challenges in processing heterogeneous ingredient descriptions, including unit, quantity, and descriptor normalization. An exploratory analysis of ingredient frequency and co-occurrence patterns, using measures such as Pointwise Mutual Information and Lift score, highlights distinctive ingredient combinations that characterize Macedonian cuisine. The resulting dataset contributes a new resource for studying food culture in underrepresented languages and offers insights into the unique patterns of Macedonian culinary tradition.

arXiv:2510.15501v2 Announce Type: replace Abstract: Despite the remarkable advances of Large Language Models (LLMs) across diverse cognitive tasks, the rapid enhancement of these capabilities also introduces emergent deceptive behaviors that may induce severe risks in high-stakes deployments. More critically, the characterization of deception across realistic real-world scenarios remains underexplored. To bridge this gap, we establish DeceptionBench, the first benchmark that systematically evaluates how deceptive tendencies manifest across different societal domains, what their intrinsic behavioral patterns are, and how extrinsic factors affect them. Specifically, on the static count, the benchmark encompasses 150 meticulously designed scenarios in five domains, i.e., Economy, Healthcare, Education, Social Interaction, and Entertainment, with over 1,000 samples, providing sufficient empirical foundations for deception analysis. On the intrinsic dimension, we explore whether models exhibit self-interested egoistic tendencies or sycophantic behaviors that prioritize user appeasement. On the extrinsic dimension, we investigate how contextual factors modulate deceptive outputs under neutral conditions, reward-based incentivization, and coercive pressures. Moreover, we incorporate sustained multi-turn interaction loops to construct a more realistic simulation of real-world feedback dynamics. Extensive experiments across LLMs and Large Reasoning Models (LRMs) reveal critical vulnerabilities, particularly amplified deception under reinforcement dynamics, demonstrating that current models lack robust resistance to manipulative contextual cues and the urgent need for advanced safeguards against various deception behaviors. Code and resources are publicly available at https://github.com/Aries-iai/DeceptionBench.

arXiv:2510.15859v2 Announce Type: replace Abstract: Reinforcement learning has powered many of the recent breakthroughs in large language models, especially for tasks where rewards can be computed automatically, such as code generation. However, these methods deteriorate in open-ended domains like medical consultation, where feedback is inherently ambiguous, highly context-dependent, and cannot be reduced to a reliable scalar signal. In such settings, RL must either rely on supervision-intensive reward models that often fail to generalize, or it falls into pathological behaviors such as reward hacking - an especially troubling risk for high-stakes medical dialogue. To address these limitations, we introduce ORBIT, an open-ended rubric-based incremental training framework for high-stakes medical dialogue. ORBIT integrates synthetic dialogue generation with dynamically constructed rubrics that serve as adaptive guides for incremental RL. Instead of relying on external medical knowledge bases or handcrafted rule sets, ORBIT uses rubric-driven feedback to steer the learning process. Its judge component can be instantiated with general-purpose instruction-following LLMs, removing the need for any task-specific fine-tuning. Applied to the Qwen3-4B-Instruct model, ORBIT raises the HealthBench-Hard score from 7.0 to 27.5 using only 2k training samples, achieving SOTA performance for models at this scale. With larger rubric datasets, ORBIT-trained models further compete with the strongest open-source baselines on HealthBench-Hard. Our analysis shows that rubric-guided RL consistently improves consultation quality across diverse medical scenarios. We also apply such rubric generation and training pipeline to InfoBench, where ORBIT enhances instruction-following performance, highlighting the generality of rubric-based feedback.

arXiv:2510.25117v2 Announce Type: replace Abstract: Large Language Models (LLMs) demonstrate remarkable capabilities, but their training on massive corpora poses significant risks from memorized sensitive information. To mitigate these issues and align with legal standards, unlearning has emerged as a critical technique to selectively erase specific knowledge from LLMs without compromising their overall performance. This survey provides a systematic review of over 180 papers on LLM unlearning published since 2021. First, it introduces a novel taxonomy that categorizes unlearning methods based on the phase in the LLM pipeline of the intervention. This framework further distinguishes between parameter modification and parameter selection strategies, thus enabling deeper insights and more informed comparative analysis. Second, it offers a multidimensional analysis of evaluation paradigms. For datasets, we compare 18 existing benchmarks from the perspectives of task format, content, and experimental paradigms to offer actionable guidance. For metrics, we move beyond mere enumeration by dividing knowledge memorization metrics into 10 categories to analyze their advantages and applicability, while also reviewing metrics for model utility, robustness, and efficiency. By discussing current challenges and future directions, this survey aims to advance the field of LLM unlearning and the development of secure AI systems.

arXiv:2511.01891v2 Announce Type: replace Abstract: Multi-personality generation for LLMs, enabling simultaneous embodiment of multiple personalization attributes, is a fundamental challenge. Existing retraining-based approaches are costly and poorly scalable, while decoding-time methods often rely on external models or heuristics, limiting flexibility and robustness. In this paper, we propose a novel Multi-Personality Generation (MPG) framework under the decoding-time combination paradigm. It flexibly controls multi-personality without relying on scarce multi-dimensional models or extra training, leveraging implicit density ratios in single-dimensional models as a "free lunch" to reformulate the task as sampling from a target strategy aggregating these ratios. To implement MPG efficiently, we design Speculative Chunk-level based Rejection sampling (SCR), which generates responses in chunks and parallelly validates them via estimated thresholds within a sliding window. This significantly reduces computational overhead while maintaining high-quality generation. Experiments on MBTI personality and Role-Playing demonstrate the effectiveness of MPG, showing improvements up to 16%-18%. Code and data are available at https://github.com/Libra117/MPG .

arXiv:2511.03369v2 Announce Type: replace Abstract: Safety-aligned large language models (LLMs) are becoming increasingly widespread, especially in sensitive applications where fairness is essential and biased outputs can cause significant harm. However, evaluating the fairness of models is a complex challenge, and approaches that do so typically utilize standard question-answer (QA) styled schemes. Such methods often overlook deeper issues by interpreting the model's refusal responses as positive fairness measurements, which creates a false sense of fairness. In this work, we introduce the concept of silenced biases, which are unfair preferences encoded within models' latent space and are effectively concealed by safety-alignment. Previous approaches that considered similar indirect biases often relied on prompt manipulation or handcrafted implicit queries, which present limited scalability and risk contaminating the evaluation process with additional biases. We propose the Silenced Bias Benchmark (SBB), which aims to uncover these biases by employing activation steering to reduce model refusals during QA. SBB supports easy expansion to new demographic groups and subjects, presenting a fairness evaluation framework that encourages the future development of fair models and tools beyond the masking effects of alignment training. We demonstrate our approach over multiple LLMs, where our findings expose an alarming distinction between models' direct responses and their underlying fairness issues.

arXiv:2511.07998v2 Announce Type: replace Abstract: Structured data question answering (QA), including table QA, Knowledge Graph (KG) QA, and temporal KG QA, is a pivotal research area. Advances in large language models (LLMs) have driven significant progress in unified structural QA frameworks like TrustUQA. However, these frameworks face challenges when applied to small-scale LLMs since small-scale LLMs are prone to errors in generating structured queries. To improve the structured data QA ability of small-scale LLMs, we propose a self-correction distillation (SCD) method. In SCD, an error prompt mechanism (EPM) is designed to detect errors and provide customized error messages during inference, and a two-stage distillation strategy is designed to transfer large-scale LLMs' query-generation and error-correction capabilities to small-scale LLM. Experiments across 5 benchmarks with 3 structured data types demonstrate that our SCD achieves the best performance and superior generalization on small-scale LLM (8B) compared to other distillation methods, and closely approaches the performance of GPT4 on some datasets. Furthermore, large-scale LLMs equipped with EPM surpass the state-of-the-art results on most datasets.

arXiv:2511.08230v2 Announce Type: replace Abstract: The development of multi-modal large language models (LLMs) leads to intelligent approaches capable of speech interactions. As one of the most widely spoken languages globally, Mandarin is supported by most models to enhance their applicability and reach. However, the scarcity of comprehensive speech-to-speech (S2S) benchmarks in Mandarin contexts impedes systematic evaluation for developers and hinders fair model comparison for users. In this work, we propose VocalBench-zh, an ability-level divided evaluation suite adapted to Mandarin context consisting of 10 well-crafted subsets and over 10K high-quality instances, covering 12 user-oriented characters. The evaluation experiment on 14 mainstream models reveals the common challenges for current routes, and highlights the need for new insights into next-generation speech interactive systems. The evaluation codes and datasets will be available at https://github.com/SJTU-OmniAgent/VocalBench-zh.

arXiv:2511.08614v2 Announce Type: replace Abstract: Medical decision-support and advising systems are critical for emergency physicians to quickly and accurately assess patients' conditions and make diagnosis. Artificial Intelligence (AI) has emerged as a transformative force in healthcare in recent years and Large Language Models (LLMs) have been employed in various fields of medical decision-support systems. We studied responses of a group of different LLMs to real cases in emergency medicine. The results of our study on five most renown LLMs showed significant differences in capabilities of Large Language Models for diagnostics acute diseases in medical emergencies with accuracy ranging between 58% and 65%. This accuracy significantly exceeds the reported accuracy of human doctors. We built a super-learner MEDAS (Medical Emergency Diagnostic Advising System) of five major LLMs - Gemini, Llama, Grok, GPT, and Claude). The super-learner produces higher diagnostic accuracy, 70%, even with a quite basic meta-learner. However, at least one of the integrated LLMs in the same super-learner produces 85% correct diagnoses. The super-learner integrates a cluster of LLMs using a meta-learner capable of learning different capabilities of each LLM to leverage diagnostic accuracy of the model by collective capabilities of all LLMs in the cluster. The results of our study showed that aggregated diagnostic accuracy provided by a meta-learning approach exceeds that of any individual LLM, suggesting that the super-learner can take advantage of the combined knowledge of the medical datasets used to train the group of LLMs.

arXiv:2511.09292v2 Announce Type: replace Abstract: Recent advancements in large language models (LLMs) have demonstrated remarkable text generation capabilities. However, controlling specific attributes of generated text remains challenging without architectural modifications or extensive fine-tuning. Current methods typically toggle a single, basic attribute but struggle with precise multi-attribute control. In scenarios where attribute requirements conflict, existing methods lack coordination mechanisms, causing interference between desired attributes. Furthermore, these methods fail to incorporate iterative optimization processes in the controlled generation pipeline. To address these limitations, we propose Conflict-aware, Composite, and Collaborative Controlled Text Generation (C$^3$TG), a two-phase framework for fine-grained, multi-dimensional text attribute control. During generation, C$^3$TG selectively pairs the LLM with the required attribute classifiers from the 17 available dimensions and employs weighted KL-divergence to adjust token probabilities. The optimization phase then leverages an energy function combining classifier scores and penalty terms to resolve attribute conflicts through iterative feedback, enabling precise control over multiple dimensions simultaneously while preserving natural text flow. Experiments show that C$^3$TG significantly outperforms baselines across multiple metrics including attribute accuracy, linguistic fluency, and output diversity, while simultaneously reducing toxicity. These results establish C$^3$TG as an effective and flexible solution for multi-dimensional text attribute control that requires no costly model modifications.

arXiv:2511.10045v2 Announce Type: replace Abstract: Sound symbolism is a linguistic concept that refers to non-arbitrary associations between phonetic forms and their meanings. We suggest that this can be a compelling probe into how Multimodal Large Language Models (MLLMs) interpret auditory information in human languages. We investigate MLLMs' performance on phonetic iconicity across textual (orthographic and IPA) and auditory forms of inputs with up to 25 semantic dimensions (e.g., sharp vs. round), observing models' layer-wise information processing by measuring phoneme-level attention fraction scores. To this end, we present LEX-ICON, an extensive mimetic word dataset consisting of 8,052 words from four natural languages (English, French, Japanese, and Korean) and 2,930 systematically constructed pseudo-words, annotated with semantic features applied across both text and audio modalities. Our key findings demonstrate (1) MLLMs' phonetic intuitions that align with existing linguistic research across multiple semantic dimensions and (2) phonosemantic attention patterns that highlight models' focus on iconic phonemes. These results bridge domains of artificial intelligence and cognitive linguistics, providing the first large-scale, quantitative analyses of phonetic iconicity in terms of MLLMs' interpretability.