Papers by An Yang

MOOSE-Copilot: A Web-Based Interactive Assistant for Unified Exploratory and Fine-Grained Scientific Hypothesis Discovery (2026.acl-demo)

Challenge:	Existing role-play fine-tuning techniques improve role adaptability but may degrade safety performance, especially for villainous characters.
Approach:	They propose safety-aware Role-Play Fine-Tuning (SaRFT) to balance role-playing capabilities and safety.
Outcome:	The proposed method outperforms state-of-the-art baselines under both LoRA and full-parameter fine-tuning settings.

Copied to clipboard

Hongran An, Zonglin Yang

Challenge:	Existing approaches to scientific hypothesis discovery operate autonomously with little to no human guidance.
Approach:	They propose a framework that empowers scientists to steer the generative process via explicit signals.
Outcome:	The proposed framework outperforms existing models in terms of quality and accuracy of the input signals.

Hello Again! LLM-powered Personalized Agent for Long-term Dialogue (2025.naacl-long)

Copied to clipboard

Hao Li, Chenghao Yang, An Zhang, Yang Deng, Xiang Wang, Tat-Seng Chua

Natural Language Interfaces

Challenge:	Existing dialogue systems focus on brief single-session interactions, neglecting real-world needs for long-term companionship and personalized interactions.
Approach:	They propose a model-agnostic framework for long-term dialogue agents . they use event summary and persona management to enable reasoning .
Outcome:	The proposed framework incorporates three independently tunable modules dedicated to event perception, persona extraction, and response generation.

LongSpec: Long-Context Lossless Speculative Decoding with Efficient Drafting and Verification (2026.acl-long)

Copied to clipboard

Penghui Yang, Cunxiao Du, Fengzhuo Zhang, Haonan Wang, Tianyu Pang, Chao Du, Bo An

Green & Sustainable NLP

Challenge:	Large Language Models (LLMs) can process extremely long contexts, requiring efficient inference over extended inputs.
Approach:	They propose a model that uses a constant-sized key-value cache to train long-context models.
Outcome:	Experimental results show that LongSpec achieves 3.26x speedup over strong Flash Attention baselines and 2.34x wall clock time on four math reasoning tasks.

Evaluation Metrics for Headline Generation Using Deep Pre-Trained Embeddings (2020.lrec-1)

Copied to clipboard

Abdul Moeed, Yang An, Gerhard Hagerer, Georg Groh

A Reinforcement Learning Framework for Robust and Secure LLM Watermarking (2026.eacl-long)

Challenge:	Recent generative language models have shown promise in abstractive summarization tasks.
Approach:	They propose to use Fr echet embedding distance and angular embeddable similarity to evaluate the performance of generative language models in abstractive summarization tasks.
Outcome:	The proposed metric shows close relation with human judgments and has overall better correlations with them.

Copied to clipboard

Li An, Yujian Liu, Yepeng Liu, Yuheng Bu, Yang Zhang, Shiyu Chang

Robustness in NLP

Challenge:	Existing watermarking algorithms rely on heuristic green/red token lists . however, these lists are inconsistent and can be compromised .
Approach:	They propose a framework for robust and secure LLM watermarking using reinforcement learning.
Outcome:	The proposed method achieves state-of-the-art trade-off across all criteria with notable improvements in resistance to spoofing attacks without degrading other criteria.

RMTBench: Benchmarking LLMs Through Multi-Turn User-Centric Role-Playing (2025.findings-emnlp)

Copied to clipboard

Hao Xiang, Tianyi Tang, Yang Su, Bowen Yu, An Yang, Fei Huang, Yichang Zhang, Yaojie Lu, Hongyu Lin, Xianpei Han, Jingren Zhou, Junyang Lin, Le Sun

Dialogue Systems & Conversational Agents

Challenge:	Existing benchmarks focus on character-centric approach and fail to reflect real-world applications.
Approach:	RMTBench is a user-centric bilingual role-playing benchmark featuring 80 diverse characters and over 8,000 dialogue rounds.
Outcome:	RMTBench features 80 diverse characters and over 8,000 dialogue rounds.

CSLM: A Framework for Question Answering Dataset Generation through Collaborative Small Language Models (2024.findings-emnlp)

Copied to clipboard

Yiming Wang, Yang Liu, Lingchen Wang, An Xiao

Musical Score Understanding Benchmark: Evaluating Large Language Models’ Comprehension of Complete Musical Scores (2026.acl-long)

Challenge:	Collecting high-quality question-answer (QA) pairs is vital for training large language models, but computational demands and associated costs often render such approaches prohibitive for the average researcher.
Approach:	They propose a small-scaled, open-source solution that generates QA pairs from documents or raw corpora using large-scale models like Llama-70B.
Outcome:	Experiments on domain-specific datasets show that the proposed model can generate high-quality QA pairs, making it accessible to a broader range of researchers.

Copied to clipboard

Congren Dai, Yue Yang, Krinos Li, Huichi Zhou, Shijie Liang, Zhang Bo, Enyang Liu, Ge Jin, Hongran An, Haosen Zhang, Peiyuan Jing, KinHei Lee, Zhenxuan Zhang, Xiaobing Li, Maosong Sun

MPO: Multilingual Safety Alignment via Reward Gap Optimization (2025.acl-long)

Challenge:	Existing benchmarks for musical score understanding are narrow in scope, focusing on isolated fragments, short excerpts, or multiple-choice formulations, rather than supporting holistic reasoning over entire scores.
Approach:	They propose a benchmark for score-level musical understanding across textual and visual modalities.
Outcome:	The musical score understanding benchmark contains 1,800 question-answer pairs from works by Bach, Beethoven, Chopin, Debussy, and others.

Copied to clipboard

Weixiang Zhao, Yulin Hu, Yang Deng, Tongtong Wu, Wenxuan Zhang, Jiahe Guo, An Zhang, Yanyan Zhao, Bing Qin, Tat-Seng Chua, Ting Liu

Multilinguality

Challenge:	Existing preference learning methods for safety alignment are monolingual and struggle with noisy multilingual data.
Approach:	They propose a multilingual reward gaP optimization approach that leverages the well-aligned safety capabilities of the dominant language to improve safety alignment across multiple languages.
Outcome:	Extensive experiments on three LLMs, LLaMA-3.1, Gemma-2 and Qwen2.5, validate MPO’s efficacy in multilingual safety alignment without degrading general multilingual utility.

PACE: Prefix-Protected and Difficulty-Aware Compression for Efficient Reasoning (2026.findings-acl)

Copied to clipboard

Ruixiang Feng, Yuntao Wen, Silin Zhou, Ke Shi, Yifan Wang, Ran Le, Zhenwei An, Zongchao Chen, Chen Yang, Guangyue Peng, Yiming Jia, Dongsheng Wang, Tao Zhang, Lisi Chen, Yang Song, Shen Gao, Shuo Shang

ThinkAnswer Loss: Balancing Semantic Similarity and Exact Matching for LLM Reasoning Enhancement (2025.findings-emnlp)

Challenge:	Existing LRMs often suffer from "overthinking" and excessively long reasoning traces . a dual-level framework for length compression of LRM is proposed .
Approach:	They propose a framework for prefix-protected and difficulty-aware compression under hierarchical supervision.
Outcome:	The proposed framework reduces token usage while improving accuracy on math benchmarks.

Copied to clipboard

Shan Yang, Kun Wu, Zeju Li, Linlin Zhang, Xiangyu Pei, Leike An, Yu Liu

Reasoning

Challenge:	Existing methods for knowledge distillation use Chain-of-Thought (CoT) and answer pairs, but they lack appropriate supervision signals.
Approach:	They propose a framework that decouples CoT and answer supervision . the framework applies semantic similarity constraints while maintaining strict literal matching for the answer .
Outcome:	The proposed framework decouples CoT and answer supervision while maintaining strict literal matching for the answer.

Multimodal Invariant Sentiment Representation Learning (2025.findings-acl)

Copied to clipboard

Aoqiang Zhu, Min Hu, Xiaohua Wang, Jiaoyun Yang, Yiming Tang, Ning An

PLAWBENCH: A Rubric-Based Benchmark for Evaluating LLMs in Real-World Legal Practice (2026.acl-long)

Challenge:	Existing methods for multimodal sensing ignore significant sentiment distribution imbalances and cross-modal sentiment conflicts, hindering performance improvement.
Approach:	They propose a method to learn stable multimodal invariant sentiment representations by incorporating distributional discrepancies and sentiment conflicts into the model training.
Outcome:	The proposed method improves MSA performance and achieves new state-of-the-art.

Copied to clipboard

Yuzhen Shi, Huanghai Liu, Yiran HU, Song Gaojie, Xu Xinran, Yubo Ma, Tianyi Tang, Li Zhang, Qingjing Chen, Feng Di, Wenbo Lv, Weiheng Wu, Kexin Yang, Sen Yang, Wei Wang, Rongyao Shi, Qiu Yuanyang, Yuemeng Qi, Zhang Jingwen, Sui Xiaoyu, Yifan Chen, Zhang Yi, An Yang, Bowen Yu, Dayiheng Liu, Junyang Lin, Weixing Shen, Bing Zhao, Charles L. A. Clarke, HU Wei

Reasoning

Challenge:	Existing benchmarks for large language models (LLMs) are coarse, single-dimensional metrics and do not explicitly assess fine-grained legal reasoning.
Approach:	They propose a Practical Law Benchmark to evaluate large language models in real-world legal practice scenarios.
Outcome:	The proposed model is based on 850 questions and 13 scenarios with expert-designed evaluation rubrics.

Refining Corpora from a Model Calibration Perspective for Chinese Spelling Correction (2024.findings-acl)

Copied to clipboard

Dingyao Yu, Yang An, Wei Ye, Xiongfeng Xiao, Shaoguang Mao, Tao Ge, Shikun Zhang

Harder Task Needs More Experts: Dynamic Routing in MoE Models (2024.acl-long)

Challenge:	Chinese Spelling Correction (CSC) lacks large-scale high-quality corpora due to labor-intensive labeling of spelling errors in real-life writing or typing scenarios.
Approach:	They propose to use OCR/ASR-based generation to refine Chinese Spelling Correction models on random replacement-based corpora and filter them based on prediction confidence.
Outcome:	The proposed model outperforms existing models on three widely-used benchmarks while significantly alleviating over-correction.

Copied to clipboard

Quzhe Huang, Zhenwei An, Nan Zhuang, Mingxu Tao, Chen Zhang, Yang Jin, Kun Xu, Kun Xu, Liwei Chen, Songfang Huang, Yansong Feng

SWE-Swiss: A Multi-Task Fine-Tuning and RL Recipe for High-Performance Issue Resolution (2026.findings-acl)

Challenge:	Unlike existing MoE approaches that rely on fixed TopK Routing, our dynamic expert selection framework dynamically allocates experts based on the confidence level in expert selection for each input.
Approach:	They propose a dynamic expert selection framework that dynamically allocates experts based on the confidence level in expert selection for each input.
Outcome:	The proposed method achieves an average improvement of 0.7% with less than 90% activated parameters and outperforms dense models in QA and machine translation tasks.

Copied to clipboard

Zhenyu He, Qingping Yang, Wei Shen, Xiaojian Zhong, Kechi Zhang, Chenxin An, Wenlei Shi, Tianle Cai, Di He, Jiaze Chen, Jingjing Xu

MorphoBench: A Benchmark with Difficulty Adaptive to Model Reasoning (2026.findings-acl)

Challenge:	SWE-Swiss-32B demonstrates strong generalization to other common LLM benchmarks.
Approach:	They propose a two-phase training recipe that decomposes issue resolution into three core skills: Localization, Repair, and Unit Test Generation.
Outcome:	The proposed model achieves a 60.2% score on the SWE-bench Verified benchmark and is in the top-tier performance bracket of much larger models.

Copied to clipboard

Xukai Wang, Xuanbo Liu, Mingrui Chen, Haitian Zhong, Xuanlin Yang, Bohan Zeng, Jinbo Hu, Hao Liang, Junbo Niu, Xuchen Li, Ruitao Wu, Ruichuan An, Yang Shi, Liu Liu, Qiang Liu, Zhouchen Lin, Xu-Yao Zhang, Wentao Zhang, Bin Dong

Reasoning

Challenge:	Existing benchmarks designed to evaluate the reasoning capabilities of large models are limited in scope and lack flexibility to adapt difficulty according to evolving reasoning capacities of models.
Approach:	They propose a benchmark that incorporates multidisciplinary questions to evaluate the reasoning capabilities of large models and can adjust and update question difficulty based on the reasoning abilities of advanced models.
Outcome:	The proposed benchmark incorporates multidisciplinary questions to evaluate the reasoning capabilities of large models and can adjust and update question difficulty based on the reasoning abilities of advanced models.

AgentOCR: Reimagining Agent History via Optical Self-Compression (2026.acl-long)

Copied to clipboard

Lang Feng, Fuchao Yang, Feng Chen, Xin Cheng, Haiyang Xu, Zhenglin Wan, Ming Yan, Bo An

Visual Data in NLP

Challenge:	Recent advances in large language models (LLMs) have enabled agentic systems trained with reinforcement learning over multi-turn interaction, but practical deployment is bottlenecked by rapidly growing textual histories that inflate token and memory costs.
Approach:	They propose a framework that represents the accumulated observation-action history as a compact rendered image.
Outcome:	The proposed framework preserves over 95% of text-based agent performance while significantly reducing token consumption (>50%), yielding consistent token and memory efficiency.

Sketch and Refine: Towards Faithful and Informative Table-to-Text Generation (2021.findings-acl)

Copied to clipboard

Peng Wang, Junyang Lin, An Yang, Chang Zhou, Yichang Zhang, Jingren Zhou, Hongxia Yang

Structured Data in NLP

Challenge:	Existing methods for table-to-text generation suffer from poor faithfulness and low coverage.
Approach:	They propose a method that combines Autoregressive and Non-Autoregressive generation to generate a table-to-text from a key-value table using a skeleton and an edit-based non-autoregressively generation model.
Outcome:	The proposed method outperforms the existing methods on WikiPerson and WikiBio datasets on coverage and faithfulness.

HMoE: Heterogeneous Mixture of Experts for Language Modeling (2025.emnlp-main)

Copied to clipboard

An Wang, Xingwu Sun, Ruobing Xie, Shuaipeng Li, Jiaqi Zhu, Zhen Yang, Pinxue Zhao, Weidong Han, Zhanhui Kang, Di Wang, Naoaki Okazaki, Cheng-zhong Xu

EmoRes: Toward Adaptive Psychological Support via User-Agnostic Benchmark and Topic-Mining Agent (2026.findings-acl)

Challenge:	Mixture of Experts (MoE) models use homogeneous experts with diverse capacities, resulting in a lack of expert specialization and parameter utilization.
Approach:	They propose a framework where experts differ in size and possess diverse capacities . they propose HMoE to encourage frequent activation of smaller experts .
Outcome:	The proposed framework outperforms homogeneous homogenous MoE models on evaluation benchmarks and achieves lower loss rate with fewer activated parameters.

Copied to clipboard

Zhengwei Zou, Xuanming Jiang, Baoyi An, Dingyu Nie, Zhengxing Fang, Qingyu Liu, Xueming Qian, Guoshuai Zhao, Zhongyu Yang

Purging the Gray Zone: Latent-Geometric Denoising for Precise Knowledge Boundary Awareness (2026.findings-acl)

Challenge:	Large language models generate fragmented and emotionally inconsistent dialogues lacking the therapeutic structure necessary for reliable assessment.
Approach:	They propose a framework that boosts psychological reasoning via a Topic-Mining Emotional Agent and a multi-perspective Self-Reflection Agent.
Outcome:	The proposed framework improves topic continuity, emotional coherence, and clinical interpretability over baselines and validated by ablation studies and human evaluations.

Copied to clipboard

Hao An, Yibin Lou, Jiayi Guo, Yang Xu

Proxy-Driven Robust Multimodal Sentiment Analysis with Incomplete Data (2025.acl-long)

Challenge:	Existing abstention fine-tuning methods cause models to suffer from label noise near the decision boundaries.
Approach:	They propose a latent space representation perspective for abstention fine-tuning . they propose 'geometric denoising' framework that constructs a truth hyperplane .
Outcome:	The proposed framework significantly enhances model truthfulness and demonstrates strong generalization in out-of-distribution scenarios.

Copied to clipboard

Aoqiang Zhu, Min Hu, Xiaohua Wang, Jiaoyun Yang, Yiming Tang, Ning An

Robustness in NLP

Challenge:	Existing studies focus on optimizing model structures to handle uncertain missingness, but models still face challenges when dealing with uncertain missing data.
Approach:	They propose a data-centric robust multimodal sentiment analysis method, Proxy-Driven Robust Multimodal Fusion, which maps unimodal data to the latent space of Gaussian distributions to capture core features and structure.
Outcome:	The proposed method outperforms existing models in noise resistance and achieves state-of-the-art performance on multiple benchmark datasets.

SeDev: Structured Semantic Exploration for LLM-Driven Code Generation (2026.acl-long)

Copied to clipboard

Ronghui Yang, Jie Liu, Jiajie Zeng, Jiexin Wang, Jiuchuan Jiang, Bo An, Yi Cai, Mengchen Zhao

Programming Languages in NLP

Challenge:	Large Language Models (LLMs) have shown remarkable capabilities in automating code generation, but they suffer from insufficient exploration of the vast solution space.
Approach:	They propose a large-scale LLM-driven code generation framework that efficiently finds high-quality solutions in only a few iterations.
Outcome:	The proposed framework outperforms baselines while maintaining reasonable time and computational costs.

Prompt Tuning for Unified Multimodal Pretrained Models (2023.findings-acl)

Copied to clipboard

Hao Yang, Junyang Lin, An Yang, Peng Wang, Chang Zhou

AdaSteer: Your Aligned LLM is Inherently an Adaptive Jailbreak Defender (2025.emnlp-main)

Challenge:	Prompt tuning has demonstrated success in natural language pretraining and even vision pretraining.
Approach:	They propose to apply prompt tuning to a unified sequence-to-sequence pretrained model by adding a sequence of learnable embeddings to each layer and finetuning the pretrained models on downstream tasks.
Outcome:	The proposed method outperforms other parameter-efficient tuning methods on multimodal models and is robust against adversarial attacks.

Copied to clipboard

Weixiang Zhao, Jiahe Guo, Yulin Hu, Yang Deng, An Zhang, Xingyu Sui, Xinyang Han, Yanyan Zhao, Bing Qin, Tat-Seng Chua, Ting Liu

Gradient-Attention Guided Dual-Masking Synergetic Framework for Robust Text-based Person Retrieval (2025.emnlp-main)

Challenge:	Activation steering offers training-free defense but relies on fixed steering coefficients, resulting in suboptimal protection and increased false rejections of benign inputs.
Approach:	They propose an adaptive activation steering method that dynamically adjusts model behavior based on input characteristics.
Outcome:	The proposed method outperforms baseline methods across multiple jailbreak attacks with minimal impact on utility.

Copied to clipboard

Tianlu Zheng, Yifan Zhang, Xiang An, Ziyong Feng, Kaicheng Yang, Qichuan Ding

SciDTB: Discourse Dependency TreeBank for Scientific Abstracts (P18-2)

Challenge:	a large-scale visionlanguage pre-training framework is limited by the scarcity of large-sized annotated vision-language data . noise-resistant data construction pipeline is needed to filter and caption web-sourced images . noisy text tokens can be a problem for fine-grained representation learning .
Approach:	They develop a noise-resistant data construction pipeline that leverages in-context learning capabilities of MLLMs to automatically filter and caption web-sourced images.
Outcome:	The proposed framework improves cross-modal alignment by masking noisy textual tokens based on the gradient-attention similarity score.

Copied to clipboard

An Yang, Sujian Li

Syntactic Text Processing

Challenge:	Discourse relations are annotated on scientific articles.
Approach:	They propose a domain-specific discourse treebank annotated on scientific articles . they use dependency trees to represent discourse structure, which is flexible and simplified .
Outcome:	The proposed treebank is a benchmark for evaluating discourse dependency parsers.

Rationales Are Not Silver Bullets: Measuring the Impact of Rationales on Model Performance and Reliability (2025.findings-acl)

Copied to clipboard

Chiwei Zhu, Benfeng Xu, An Yang, Junyang Lin, Quan Wang, Chang Zhou, Zhendong Mao

DaNet: Dual-Aware Enhanced Alignment Network for Multimodal Aspect-Based Sentiment Analysis (2025.findings-acl)

Challenge:	Existing studies have shown that training language models with rationales augmentation is beneficial, but this view does not hold consistently.
Approach:	They conduct comprehensive investigations to thoroughly inspect the impact of rationales on model performance and a novel perspective of model reliability.
Outcome:	The proposed method outperforms untrained models in several areas and provides informative regulations on the broad utilization of rationales.

Copied to clipboard

Aoqiang Zhu, Min Hu, Xiaohua Wang, Jiaoyun Yang, Yiming Tang, Ning An

Sentiment Analysis

Challenge:	Existing methods assume a direct alignment between images and aspects, matching the entire image with a corresponding aspect. Existing algorithms assume 'direct alignment' between images, introducing noise.
Approach:	They propose a Dual-Aware Enhanced Alignment Network (DaNet) that can enhance fine-grained multimodal aspect-image alignment and denoising.
Outcome:	The proposed system outperforms existing methods in three subtasks and is available on https://github.com/***/DaNet.

Learning Relation Alignment for Calibrated Cross-modal Retrieval (2021.acl-long)

Copied to clipboard

Shuhuai Ren, Junyang Lin, Guangxiang Zhao, Rui Men, An Yang, Jingren Zhou, Xu Sun, Hongxia Yang

RWKV-CLIP: A Robust Vision-Language Representation Learner (2024.emnlp-main)

Challenge:	despite advances in multimodal pre-training, cross-modal retrieval remains challenging . lack of relation consistency impairs contextualized representation of image-text pairs .
Approach:	They propose a new metric to quantify the relation consistency by measuring the semantic distance between linguistic and visual relations.
Outcome:	The proposed method boosts the performance of prevailing models on Flickr30k and MS COCO datasets by a considerable margin.

Copied to clipboard

Tiancheng Gu, Kaicheng Yang, Xiang An, Ziyong Feng, Dongnan Liu, Weidong Cai, Jiankang Deng

Enhancing Pre-Trained Language Representations with Rich Knowledge for Machine Reading Comprehension (P19-1)

Challenge:	Using large image-text datasets, large-scale image-data sets have been used for visionlanguage pre-training.
Approach:	They propose a framework that leverages Large Language Models to combine and refine information from web-based image-text pairs, synthetic captions, and detection tags.
Outcome:	The proposed framework can combine and refine information from web-based image-text pairs, synthetic captions, and detection tags.

Copied to clipboard

An Yang, Quan Wang, Jing Liu, Kai Liu, Yajuan Lyu, Hua Wu, Qiaoqiao She, Sujian Li

Transferring General Multimodal Pretrained Models to Text Recognition (2023.findings-acl)

Challenge:	Recent results show pre-trained language models (LMs) can improve machine reading comprehension (MRC) Experimental results indicate that KT-NET offers significant and consistent improvements over BERT .
Approach:	They propose a method that leverages external knowledge bases to improve machine reading comprehension (MRC) KT-NET employs an attention mechanism to select desired knowledge from KBs and fuses selected knowledge with BERT to enable context- and knowledge-aware predictions.
Outcome:	The proposed model outperforms baseline models on ReCoRD and SQuAD1.1 benchmarks and ranks 1st on the ReCoDR and SQUAD1.1 leaderboards.

Copied to clipboard

Junyang Lin, Xuancheng Ren, Yichang Zhang, Gao Liu, Peng Wang, An Yang, Chang Zhou