Rania Elbadry, Sarfraz Ahmad, Ahmed Heakl, Dani Bouch, Momina Ahsan, Muhra AlMahri, Marwa Elsaid Khalil, Yuxia Wang, Salem Lahlou, Sophia Ananiadou, Veselin Stoyanov, Jimin Huang, Xueqing Peng, Preslav Nakov, Zhuohan Xie. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Xueqing Peng, Lingfei Qian, Yan Wang, Ruoyu Xiang, Yueru He, Yang Ren, Mingyang Jiang, Vincent Jim Zhang, Yuqing Guo, Jeff Zhao, Huan He, Yi Han, Yun Feng, Yuechen Jiang, Yupeng Cao, Haohang Li, Yangyang Yu, Xiaoyu Wang, Penglei Gao, Shengyuan Lin, Keyi Wang, Shanshan Yang, Yilun Zhao, Zhiwei Liu, Peng Lu, Jerry Huang, Suyuchen Wang, Triantafillos Papadopoulos, Polydoros Giannouris, Efstathia Soufleri, Nuo Chen, Zhiyang Deng, Heming Fu, Yijia Zhao, Mingquan Lin, Meikang Qiu, Kaleb E Smith, Arman Cohan, Xiao-Yang Liu, Jimin Huang, Guojun Xiong, Alejandro Lopez-Lira, Xi Chen, Junichi Tsujii, Jian-Yun Nie, Sophia Ananiadou, Qianqian Xie. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Zhuohan Xie, Daniil Orel, Rushil Thareja, Dhruv Sahnan, Hachem Madmoun, Fan Zhang, Debopriyo Banerjee, Georgi Nenkov Georgiev, Xueqing Peng, Lingfei Qian, Jimin Huang, Jinyan Su, Aaryamonvikram Singh, Rui Xing, Rania Elbadry, Chen Xu, Haonan Li, Fajri Koto, Ivan Koychev, Tanmoy Chakraborty, Yuxia Wang, Salem Lahlou, Veselin Stoyanov, Sophia Ananiadou, Preslav Nakov. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Most recommendation benchmarks evaluate how well a model imitates user behavior. In financial advisory, however, observed actions can be noisy or short-sighted under market volatility and may conflict with a user's long-term goals. Treating what users chose as the sole ground truth, therefore, conflates behavioral imitation with decision quality. We introduce Conv-FinRe, a conversational and longitudinal benchmark for stock recommendation that evaluates LLMs beyond behavior matching. Given an onboarding interview, step-wise market context, and advisory dialogues, models must generate rankings over a fixed investment horizon. Crucially, Conv-FinRe provides multi-view references that distinguish descriptive behavior from normative utility grounded in investor-specific risk preferences, enabling diagnosis of whether an LLM follows rational analysis, mimics user noise, or is driven by market momentum. We build the benchmark from real market data and human decision trajectories, instantiate controlled advisory conversations, and evaluate a suite of state-of-the-art LLMs. Results reveal a persistent tension between rational decision quality and behavioral alignment: models that perform well on utility-based ranking often fail to match user choices, whereas behaviorally aligned models can overfit short-term noise. The dataset is publicly released on Hugging Face. https://huggingface.co/collections/TheFinAI/conv-finre, and the codebase is available on GitHub. https://github.com/The-FinAI/Conv-FinRe.

Going beyond simple text processing, financial auditing requires detecting semantic, structural, and numerical inconsistencies across large-scale disclosures. As financial reports are filed in XBRL, a structured XML format governed by accounting standards, auditing becomes a structured information extraction and reasoning problem involving concept alignment, taxonomy-defined relations, and cross-document consistency. Although large language models (LLMs) show promise on isolated financial tasks, their capability in professional-grade auditing remains unclear. We introduce FinAuditing, a taxonomy-aligned, structure-aware benchmark built from real XBRL filings. It contains 1,102 annotated instances averaging over 33k tokens and defines three tasks: Financial Semantic Matching (FinSM), Financial Relationship Extraction (FinRE), and Financial Mathematical Reasoning (FinMR). Evaluations of 13 state-of-the-art LLMs reveal substantial gaps in concept retrieval, taxonomy-aware relation modeling, and consistent cross-document reasoning. These findings highlight the need for realistic, structure-aware benchmarks. We release the evaluation code1 and dataset2 publicly, and the task currently serves as the official benchmark of an ongoing public evaluation contest3.

OBJECTIVES: To assess the performance, generalizability, and computational efficiency of instruction-tuned Large Language Model Meta AI (LLaMA)-2 and LLaMA-3 models compared to bidirectional encoder representations from transformers (BERT) for clinical information extraction (IE) tasks, specifically named entity recognition (NER) and relation extraction (RE). MATERIALS AND METHODS: We developed a comprehensive annotated corpus of 1588 clinical notes from 4 data sources-UT Physicians (UTP) (1342 notes), Transcribed Medical Transcription Sample Reports and Examples (MTSamples) (146), Medical Information Mart for Intensive Care (MIMIC)-III (50), and Informatics for Integrating Biology and the Bedside (i2b2) (50), capturing 4 clinical entities (problems, tests, medications, other treatments) and 16 modifiers (eg, negation, certainty). Large Language Model Meta AI-2 and LLaMA-3 were instruction-tuned for clinical NER and RE, and their performance was benchmarked against BERT. RESULTS: Large Language Model Meta AI models consistently outperformed BERT across datasets. In data-rich settings (eg, UTP), LLaMA achieved marginal gains (approximately 1% improvement for NER and 1.5%-3.7% for RE). Under limited data conditions (eg, MTSamples, MIMIC-III) and on the unseen i2b2 dataset, LLaMA-3-70B improved F1 scores by over 7% for NER and 4% for RE. However, performance gains came with increased computational costs, with LLaMA models requiring more memory and Graphics Processing Unit (GPU) hours and running up to 28 times slower than BERT. DISCUSSION: While LLaMA models offer enhanced performance, their higher computational demands and slower throughput highlight the need to balance performance with practical resource constraints. Application-specific considerations are essential when choosing between LLMs and BERT for clinical IE. CONCLUSION: Instruction-tuned LLaMA models show promise for clinical NER and RE tasks. However, the tradeoff between improved performance and increased computational cost must be carefully evaluated. We release our Kiwi package (https://kiwi.clinicalnlp.org/) to facilitate the application of both LLaMA and BERT models in clinical IE applications.

BACKGROUND: Acupuncture, a key modality in traditional Chinese medicine, is gaining global recognition as a complementary therapy and a subject of increasing scientific interest. However, fragmented and unstructured acupuncture knowledge spread across diverse sources poses challenges for semantic retrieval, reasoning, and in-depth analysis. To address this gap, we developed AcuKG, a comprehensive knowledge graph that systematically organizes acupuncture-related knowledge to support sharing, discovery, and artificial intelligence-driven innovation in the field. METHODS: AcuKG integrates data from multiple sources, including online resources, guidelines, PubMed literature, ClinicalTrials.gov, and multiple ontologies (SNOMED CT, UBERON, and MeSH). We employed entity recognition, relation extraction, and ontology mapping to establish AcuKG, with human-in-the-loop to ensure data quality. Two cases evaluated AcuKG's usability: (1) how AcuKG advances acupuncture research for obesity and (2) how AcuKG enhances large language model (LLM) application on acupuncture question-answering. RESULTS: AcuKG comprises 1839 entities and 11 527 relations, mapped to 1836 standard concepts in 3 ontologies. Two use cases demonstrated AcuKG's effectiveness and potential in advancing acupuncture research and supporting LLM applications. In the obesity use case, AcuKG identified highly relevant acupoints (eg, ST25, ST36) and uncovered novel research insights based on evidence from clinical trials and literature. When applied to LLMs in answering acupuncture-related questions, integrating AcuKG with GPT-4o and LLaMA 3 significantly improved accuracy (GPT-4o: 46% → 54%, P = .03; LLaMA 3: 17% → 28%, P = .01). CONCLUSION: AcuKG is an open dataset that provides a structured and computational framework for acupuncture applications, bridging traditional practices with acupuncture research and cutting-edge LLM technologies.

Haohang Li, Yupeng Cao, Yangyang Yu, Shashidhar Reddy Javaji, Zhiyang Deng, Yueru He, Yuechen Jiang, Zining Zhu, K.p. Subbalakshmi, Jimin Huang, Lingfei Qian, Xueqing Peng, Jordan W. Suchow, Qianqian Xie. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Xueqing Peng, Triantafillos Papadopoulos, Efstathia Soufleri, Polydoros Giannouris, Ruoyu Xiang, Yan Wang, Lingfei Qian, Jimin Huang, Qianqian Xie, Sophia Ananiadou. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

Typically, in dynamic biological processes, there is a critical state or tipping point that marks the transition from one stable state to another, surpassing which a considerable qualitative shift takes place. Identifying this tipping point and its driving network is essential to avert or delay disastrous outcomes. However, most traditional approaches built upon undirected networks still suffer from a lack of robustness and effectiveness when implemented based on high-dimensional small-sample data, especially for single-cell data. To address this challenge, we develop a directed network flow entropy (DNFE) method, which can transform measured omics data into a directed network. This method is applicable to both single-cell RNA-sequencing (scRNA-seq) and bulk data. Applying this algorithm to six real datasets, including three single-cell datasets, two bulk tumor datasets, and a blood dataset, the method is proved to be effective not only in identifying critical states, as well as their dynamic network biomarkers, but also in helping explore regulatory relationships between genes. Numerical simulation results demonstrate that the DNFE algorithm is robust across various noise levels and outperforms existing methods in detecting tipping points. Furthermore, the numerical simulations for 100-node and 1000-node gene regulatory networks illustrate the method's application for large-scale data. The DNFE method predicts active transcription factors, and further identified "dark genes", which are usually overlooked with traditional methods.

There generally exists a critical state or tipping point from a stable state to another in the development of colorectal cancer (CRC) beyond which a significant qualitative transition occurs. Gut microbiome sequencing data can be collected non-invasively from fecal samples, making it more convenient to obtain. Furthermore, intestinal microbiome sequencing data contain phylogenetic information at various levels, which can be used to reliably identify critical states, thereby providing early warning signals more accurately and effectively. Yet, pinpointing the critical states using gut microbiome data presents a formidable challenge due to the high dimension and strong noise of gut microbiome data. To address this challenge, we introduce a novel approach termed the specific network information gain (SNIG) method to detect CRC's critical states at various taxonomic levels via gut microbiome data. The numerical simulation indicates that the SNIG method is robust under different noise levels and that it is also superior to the existing methods on detecting the critical states. Moreover, utilizing SNIG on two real CRC datasets enabled us to discern the critical states preceding deterioration and to successfully identify their associated dynamic network biomarkers at different taxonomic levels. Notably, we discovered certain 'dark species' and pathways intimately linked to CRC progression. In addition, we accurately detected the tipping points on an individual dataset of type I diabetes.

MOTIVATION: Large Language Models (LLMs) have the potential to revolutionize the field of Natural Language Processing, excelling not only in text generation and reasoning tasks but also in their ability for zero/few-shot learning, swiftly adapting to new tasks with minimal fine-tuning. LLMs have also demonstrated great promise in biomedical and healthcare applications. However, when it comes to Named Entity Recognition (NER), particularly within the biomedical domain, LLMs fall short of the effectiveness exhibited by fine-tuned domain-specific models. One key reason is that NER is typically conceptualized as a sequence labeling task, whereas LLMs are optimized for text generation and reasoning tasks. RESULTS: We developed an instruction-based learning paradigm that transforms biomedical NER from a sequence labeling task into a generation task. This paradigm is end-to-end and streamlines the training and evaluation process by automatically repurposing pre-existing biomedical NER datasets. We further developed BioNER-LLaMA using the proposed paradigm with LLaMA-7B as the foundational LLM. We conducted extensive testing on BioNER-LLaMA across three widely recognized biomedical NER datasets, consisting of entities related to diseases, chemicals, and genes. The results revealed that BioNER-LLaMA consistently achieved higher F1-scores ranging from 5% to 30% compared to the few-shot learning capabilities of GPT-4 on datasets with different biomedical entities. We show that a general-domain LLM can match the performance of rigorously fine-tuned PubMedBERT models and PMC-LLaMA, biomedical-specific language model. Our findings underscore the potential of our proposed paradigm in developing general-domain LLMs that can rival SOTA performances in multi-task, multi-domain scenarios in biomedical and health applications. AVAILABILITY AND IMPLEMENTATION: Datasets and other resources are available at https://github.com/BIDS-Xu-Lab/BioNER-LLaMA.

Complex biological systems often undergo sudden qualitative changes during their dynamic evolution. These critical transitions are typically characterized by a catastrophic progression of the system. Identifying the critical point is critical to uncovering the underlying mechanisms of complex biological systems. However, the system may exhibit minimal changes in its state until the critical point is reached, and in the face of high throughput and strong noise data, traditional biomarkers may not be effective in distinguishing the critical state. In this study, we propose a novel approach, mutual information weighted entropy (MIWE), which uses mutual information between genes to build networks and identifies critical states by quantifying molecular dynamic differences at each stage through weighted differential entropy. The method is applied to one numerical simulation dataset and four real datasets, including bulk and single-cell expression datasets. The critical states of the system can be recognized and the robustness of MIWE method is verified by numerical simulation under the influence of different noises. Moreover, we identify two key transcription factors (TFs), CREB1 and CREB3, that regulate downstream signaling genes to coordinate cell fate commitment. The dark genes in the single-cell expression datasets are mined to reveal the potential pathway regulation mechanism.

IMPORTANCE: The study highlights the potential of large language models, specifically GPT-3.5 and GPT-4, in processing complex clinical data and extracting meaningful information with minimal training data. By developing and refining prompt-based strategies, we can significantly enhance the models' performance, making them viable tools for clinical NER tasks and possibly reducing the reliance on extensive annotated datasets. OBJECTIVES: This study quantifies the capabilities of GPT-3.5 and GPT-4 for clinical named entity recognition (NER) tasks and proposes task-specific prompts to improve their performance. MATERIALS AND METHODS: We evaluated these models on 2 clinical NER tasks: (1) to extract medical problems, treatments, and tests from clinical notes in the MTSamples corpus, following the 2010 i2b2 concept extraction shared task, and (2) to identify nervous system disorder-related adverse events from safety reports in the vaccine adverse event reporting system (VAERS). To improve the GPT models' performance, we developed a clinical task-specific prompt framework that includes (1) baseline prompts with task description and format specification, (2) annotation guideline-based prompts, (3) error analysis-based instructions, and (4) annotated samples for few-shot learning. We assessed each prompt's effectiveness and compared the models to BioClinicalBERT. RESULTS: Using baseline prompts, GPT-3.5 and GPT-4 achieved relaxed F1 scores of 0.634, 0.804 for MTSamples and 0.301, 0.593 for VAERS. Additional prompt components consistently improved model performance. When all 4 components were used, GPT-3.5 and GPT-4 achieved relaxed F1 socres of 0.794, 0.861 for MTSamples and 0.676, 0.736 for VAERS, demonstrating the effectiveness of our prompt framework. Although these results trail BioClinicalBERT (F1 of 0.901 for the MTSamples dataset and 0.802 for the VAERS), it is very promising considering few training samples are needed. DISCUSSION: The study's findings suggest a promising direction in leveraging LLMs for clinical NER tasks. However, while the performance of GPT models improved with task-specific prompts, there's a need for further development and refinement. LLMs like GPT-4 show potential in achieving close performance to state-of-the-art models like BioClinicalBERT, but they still require careful prompt engineering and understanding of task-specific knowledge. The study also underscores the importance of evaluation schemas that accurately reflect the capabilities and performance of LLMs in clinical settings. CONCLUSION: While direct application of GPT models to clinical NER tasks falls short of optimal performance, our task-specific prompt framework, incorporating medical knowledge and training samples, significantly enhances GPT models' feasibility for potential clinical applications.

OBJECTIVE: In acupuncture therapy, the accurate location of acupoints is essential for its effectiveness. The advanced language understanding capabilities of large language models (LLMs) like Generative Pre-trained Transformers (GPTs) and Llama present a significant opportunity for extracting relations related to acupoint locations from textual knowledge sources. This study aims to explore the performance of LLMs in extracting acupoint-related location relations and assess the impact of fine-tuning on GPT's performance. MATERIALS AND METHODS: We utilized the World Health Organization Standard Acupuncture Point Locations in the Western Pacific Region (WHO Standard) as our corpus, which consists of descriptions of 361 acupoints. Five types of relations ("direction_of", "distance_of", "part_of", "near_acupoint", and "located_near") (n = 3174) between acupoints were annotated. Four models were compared: pre-trained GPT-3.5, fine-tuned GPT-3.5, pre-trained GPT-4, as well as pretrained Llama 3. Performance metrics included micro-average exact match precision, recall, and F1 scores. RESULTS: Our results demonstrate that fine-tuned GPT-3.5 consistently outperformed other models in F1 scores across all relation types. Overall, it achieved the highest micro-average F1 score of 0.92. DISCUSSION: The superior performance of the fine-tuned GPT-3.5 model, as shown by its F1 scores, underscores the importance of domain-specific fine-tuning in enhancing relation extraction capabilities for acupuncture-related tasks. In light of the findings from this study, it offers valuable insights into leveraging LLMs for developing clinical decision support and creating educational modules in acupuncture. CONCLUSION: This study underscores the effectiveness of LLMs like GPT and Llama in extracting relations related to acupoint locations, with implications for accurately modeling acupuncture knowledge and promoting standard implementation in acupuncture training and practice. The findings also contribute to advancing informatics applications in traditional and complementary medicine, showcasing the potential of LLMs in natural language processing.