The rapid progress of large language models (LLMs) has increased the demand for efficient and reliable evaluation of question answering (QA) systems. Existing evaluation methods either rely on rule-based matching with shallow semantic understanding or adopt LLM-as-a-Judge approaches that incur high cost and latency while offering limited error interpretability. Accordingly, we propose HiEval, a curriculum learning based hierarchical framework for QA task evaluation that supports both quick scoring and fine-grained error analysis. HiEval contains a quick scoring model (HiEval-QS) that predicts three-level correctness labels, and an error analysis model (HiEval-EA) that identifies incorrect responses into five error types. HiEval incorporates a class-balanced focal loss to handle label imbalance, experience replay to prevent forgetting, and contrastive unlikelihood optimization to improve error discrimination. We also construct two large-scale human-annotated evaluation datasets collected from 50 QA-related datasets, covering 8 task types and release two challenging benchmarks. Extensive experiments show that HiEval achieves state-of-the-art performance on both quick scoring and error analysis tasks, outperforming all baseline methods, including GPT-5, while being approximately 25× faster.

Text-guided image editors can now manipulate authentic medical scans with high fidelity, enabling lesion implantation/removal that threatens clinical trust and safety. Existing defenses are inadequate for healthcare. Medical detectors are largely black-box, while MLLM-based explainers are typically post-hoc, lack medical expertise, and may hallucinate evidence on ambiguous cases. We present MedForge, a data-and-method solution for pre-hoc, evidence-grounded medical forgery detection. We introduce MedForge-90K, a large-scale benchmark of realistic lesion edits across 19 pathologies with expert-guided reasoning supervision via doctor inspection guidelines and gold edit locations. Building on it, MedForge-Reasoner performs localize-then-analyze reasoning, predicting suspicious regions before producing a verdict, and is further aligned with Forgery-aware GSPO to strengthen grounding and reduce hallucinations. Experiments demonstrate state-of-the-art detection accuracy and trustworthy, expert-aligned explanations.

This study introduces Crab, a novel Configurable Role-Playing (RP) LLM with Assessing Benchmark, which consists of Role-Centric Dataset Curation, Persona-Embodying LLM Construction, and Comprehensive Benchmark Creation for RP dialogue generation. Distinct from traditional RP models that employ only several preset roles, Crab enables dynamic configuration of desired roles, thereby enhancing related flexibility and adaptability. To effectively train RP-LLMs, we curated the largest RP training dataset. The dataset provides a detailed role overview for each dialogue, including character profile, conversation scenario, and tagged topic, capturing a broad range of role-based behaviors, emotions, and interactions. We also noticed that current benchmarks lack both proper evaluation standards and methods. Thus, to validate RP-LLMs’ effectiveness, we introduced a new benchmark containing an evaluation standard, a test dataset with manual annotations, and a reward model RoleRM designed to automatically assess specific aspects of RP while aligning with human perception. Sufficient experiments reveal that RoleRM significantly outperforms ChatGPT and other evaluation methods in conducting fine-grained evaluations of RP. Also, RP-LLMs powered by Crab demonstrate superior performance across various fine-grained aspects.

Detecting LLM-generated text in specialized and high-stakes domains like medicine and law is crucial for combating misinformation and ensuring authenticity. However, current zero-shot detectors, while effective on general text, often fail when applied to specialized content due to domain shift. We provide a theoretical analysis showing this failure is fundamentally linked to the KL divergence between human, detector, and source text distributions. To address this, we propose DivScore, a zero-shot detection framework using normalized entropy-based scoring and domain knowledge distillation to robustly identify LLM-generated text in specialized domains. Experiments on medical and legal datasets show that DivScore consistently outperforms state-of-the-art detectors, with 14.4% higher AUROC and 64.0% higher recall at 0.1% false positive rate threshold. In adversarial settings, DivScore demonstrates superior robustness to other baselines, achieving on average 22.8% advantage in AUROC and 29.5% in recall.

Pre-hospital Emergency Care (PEC) systems are critical for managing life-threatening emergencies where rapid intervention can significantly impact patient outcomes. The rising global demand for PEC services, coupled with increased emergency calls and strained emergency departments, necessitates efficient resource utilization through Telephone Triage (TT) systems. However, existing TT processes face challenges such as incomplete data collection, communication barriers, and manual errors, leading to high over-triage and under-triage rates. This study proposes InTriage, an AI-driven multilingual TT system to provide decision support for triage. InTriage enhances accuracy by transcribing emergency calls, extracting critical patient information, prompting supplementary, and providing real-time triage decisions support. We conducted an evaluation on a real-world corpus of approximately 40 hours of telephone data, achieving a word error rate of 14.57% for speech recognition and an F1 score of 73.34% for key information extraction.By improving communication efficiency and reducing triage errors, InTriage offers a scalable solution to potentially help address the growing demands on PEC systems globally.

Due to the presence of the natural gap between Knowledge Graph (KG) structures and the natural language, the effective integration of holistic structural information of KGs with Large Language Models (LLMs) has emerged as a significant question. To this end, we propose a two-stage framework to learn and apply quantized codes for each entity, aiming for the seamless integration of KGs with LLMs. Firstly, a self-supervised quantized representation (SSQR) method is proposed to compress both KG structural and semantic knowledge into discrete codes (i.e., tokens) that align the format of language sentences. We further design KG instruction-following data by viewing these learned codes as features to directly input to LLMs, thereby achieving seamless integration. The experiment results demonstrate that SSQR outperforms existing unsupervised quantized methods, producing more distinguishable codes. Moreover, the fine-tuned LLaMA2 and LLaMA3.1 also have superior performance on KG link prediction and triple classification tasks, utilizing only 16 tokens per entity instead of thousands in conventional prompting methods.

Metaphor interpretation is a difficult task in natural language understanding. The development of relevant techniques in this domain is slow, mostly because of the lack of large annotated datasets and effective pre-trained language models (PLMs) for metaphor learning. Thus, we propose a large annotated dataset and a PLM for the metaphor interpretation task. Our foundation model is based on a novel anomalous language modeling (ALM) method, which we benchmark with comparable PLM baselines on the new dataset, finding that it largely improves model performance on metaphor identification and interpretation.

Metaphoric expressions are a special linguistic phenomenon, frequently appearing in everyday language. Metaphors do not take their literal meanings in contexts, which may cause obstacles for language learners to understand them. Metaphoric expressions also reflect the cognition of humans via concept mappings, attracting great attention from cognitive science and psychology communities. Thus, we aim to develop a computational metaphor processing online system, termed MetaPro Online, that allows users without a coding background, e.g., language learners and linguists, to easily query metaphoricity labels, metaphor paraphrases, and concept mappings for non-domain-specific text. The outputs of MetaPro can be directly used by language learners and natural language processing downstream tasks because MetaPro is an end-to-end system.

Sentiment analysis is a task that highly depends on the understanding of word senses. Traditional neural network models are black boxes that represent word senses as vectors that are uninterpretable for humans. On the other hand, the application of Word Sense Disambiguation (WSD) systems in downstream tasks poses challenges regarding i) which words need to be disambiguated, and ii) how to model explicit word senses into easily understandable terms for a downstream model. This work proposes a neurosymbolic framework that incorporates WSD by identifying and paraphrasing ambiguous words to improve the accuracy of sentiment predictions. The framework allows us to understand which words are paraphrased into which semantically unequivocal words, thus enabling a downstream task model to gain both accuracy and interpretability. To better fine-tune a lexical substitution model for WSD on a downstream task without ground-truth word sense labels, we leverage dynamic rewarding to jointly train sentiment analysis and lexical substitution models. Our framework proves to effectively improve the performance of sentiment analysis on corpora from different domains.

Distance metric learning has become a popular solution for few-shot Named Entity Recognition (NER). The typical setup aims to learn a similarity metric for measuring the semantic similarity between test samples and referents, where each referent represents an entity class. The effect of this setup may, however, be compromised for two reasons. First, there is typically a limited optimization exerted on the representations of entity tokens after initing by pre-trained language models. Second, the referents may be far from representing corresponding entity classes due to the label scarcity in the few-shot setting. To address these challenges, we propose a novel approach named COntrastive learning with Prompt guiding for few-shot NER (COPNER). We introduce a novel prompt composed of class-specific words to COPNER to serve as 1) supervision signals for conducting contrastive learning to optimize token representations; 2) metric referents for distance-metric inference on test samples. Experimental results demonstrate that COPNER outperforms state-of-the-art models with a significant margin in most cases. Moreover, COPNER shows great potential in the zero-shot setting.

Claims database and electronic health records database do not usually capture kinship or family relationship information, which is imperative for genetic research. We identify online obituaries as a new data source and propose a special named entity recognition and relation extraction solution to extract names and kinships from online obituaries. Built on 1,809 annotated obituaries and a novel tagging scheme, our joint neural model achieved macro-averaged precision, recall and F measure of 72.69%, 78.54% and 74.93%, and micro-averaged precision, recall and F measure of 95.74%, 98.25% and 96.98% using 57 kinships with 10 or more examples in a 10-fold cross-validation experiment. The model performance improved dramatically when trained with 34 kinships with 50 or more examples. Leveraging additional information such as age, death date, birth date and residence mentioned by obituaries, we foresee a promising future of supplementing EHR databases with comprehensive and accurate kinship information for genetic research.