Papers by Shan Zhang
SILO-BENCH: A Scalable Environment for Evaluating Distributed Coordination in Multi-Agent LLM Systems (2026.acl-long)
Copied to clipboard
Yuzhe Zhang, Feiran Liu, Yi Shan, Xinyi Huang, Xin Yang, Yueqi Zhu, Xuxin Cheng, Cao Liu, Ke Zeng, Terry Jingchen Zhang, Wenyuan Jiang
| Challenge: | Existing benchmarks conflate coordination ability with role-based priors. |
| Approach: | They propose a role-free benchmark for evaluating free-form collaboration under information silos. |
| Outcome: | The proposed benchmark systematically probes coordination capabilities under information silos using 54 configurations and 3 frontier LLMs. |
DentalGPT: Incentivizing Multimodal Reasoning in Dentistry (2026.findings-acl)
Copied to clipboard
Zhenyang Cai, Jiaming Zhang, Junjie Zhao, Ziyi Zeng, Yanchao Li, Liang Jingyi, Junying Chen, Yunjin Yang, Jiajun You, Shuzhi Deng, null Xieruiqiii, Yuanting Chen, Xiangyi Feng, Jianquan Li, Liangyi Chen, Junwen Wang, Shan Jiang, Benyou Wang
| Challenge: | Current multimodal large language models (MLLMs) show limited understanding of dental images. |
| Approach: | They propose a dental-specialized multimodal large language model trained via staged multimodal alignment and reinforcement learning. |
| Outcome: | The proposed model outperforms state-of-the-art models on disease classification and dental VQA tasks. |
Leveraging Unit Language Guidance to Advance Speech Modeling in Textless Speech-to-Speech Translation (2025.findings-acl)
Copied to clipboard
Yuhao Zhang, Xiangnan Ma, Kaiqi Kou, Peizhuo Liu, Weiqiao Shan, Benyou Wang, Tong Xiao, Yuxin Huang, Zhengtao Yu, JingBo Zhu
| Challenge: | Existing textless speech-to-speech translation models have two main challenges: 1) learning cross-modal features and 2) learning alignment of difference languages in long sequences. |
| Approach: | They propose a unit language to overcome two main modeling challenges . they propose task prompt modeling to utilize the unit language in guiding the modeling process. |
| Outcome: | The proposed language improves over a strong baseline and achieves comparable performance to models trained with text. |
N24News: A New Dataset for Multimodal News Classification (2022.lrec-1)
Copied to clipboard
| Challenge: | Current news datasets focus on text features and rarely leverage the feature of images. |
| Approach: | They propose a news dataset that uses both images and text to achieve better news classification. |
| Outcome: | The proposed model improves on the existing dataset N24News with text and image information. |
Decorrelate Irrelevant, Purify Relevant: Overcome Textual Spurious Correlations from a Feature Perspective (2022.coling-1)
Copied to clipboard
| Challenge: | Existing methods to debiase samples with biased features obstructs the model in learning from non-biased parts of the samples. |
| Approach: | They propose to eliminate spurious correlations in a fine-grained manner from a feature space perspective by using Random Fourier Features and weighted re-sampling to decorrelate dependencies between features. |
| Outcome: | The proposed method eliminates spurious correlations in a fine-grained manner from a feature space perspective. |
Enhancing Speech Large Language Models with Prompt-Aware Mixture of Audio Encoders (2025.emnlp-main)
Copied to clipboard
Weiqiao Shan, Yuang Li, Yuhao Zhang, Yingfeng Luo, Chen Xu, Xiaofeng Zhao, Long Meng, Yunfei Lu, Min Zhang, Hao Yang, Tong Xiao, JingBo Zhu
| Challenge: | Existing work on integrating audio encoders with large language models (LLMs) has focused on semantic understanding tasks, but different tasks may require distinct features that emphasize either semantic or acoustic aspects. |
| Approach: | They propose to use a prompt-aware mixture to enhance the Speech LLM that uses multiple audio encoders to extract different features based on the prompt. |
| Outcome: | The proposed approach outperforms all single-encoder Speech LLMs on ASR, speaker number verification, and AC tasks. |
A Confidence-based Partial Label Learning Model for Crowd-Annotated Named Entity Recognition (2023.findings-acl)
Copied to clipboard
Limao Xiong, Jie Zhou, Qunxi Zhu, Xiao Wang, Yuanbin Wu, Qi Zhang, Tao Gui, Xuanjing Huang, Jin Ma, Ying Shan
| Challenge: | Existing models for named entity recognition (NER) are based on large-scale labeled datasets, which always obtain using crowdsourcing. |
| Approach: | They propose a CONfidence-based partial Label Learning method to integrate prior and posterior confidences for crowd-annotated named entity recognition models. |
| Outcome: | The proposed model improves on real-world and synthetic datasets compared with baselines. |
INFACT: A Diagnostic Benchmark for Induced Faithfulness and Factuality Hallucinations in Video-LLMs (2026.acl-long)
Copied to clipboard
| Challenge: | Existing benchmarks only evaluate models in clean settings due to hallucinations . |
| Approach: | They propose a diagnostic benchmark that evaluates models in four modes for faithfulness and factuality. |
| Outcome: | The proposed benchmark evaluates models in four modes: Base (clean), Visual Degradation, Evidence Corruption, and Temporal Intervention for order-sensitive items. |
From Paraphrasing to Semantic Parsing: Unsupervised Semantic Parsing via Synchronous Semantic Decoding (2021.acl-long)
Copied to clipboard
Shan Wu, Bo Chen, Chunlei Xin, Xianpei Han, Le Sun, Weipeng Zhang, Jiansong Chen, Fan Yang, Xunliang Cai
| Challenge: | Experimental results show that Synchronous Semantic Decoding (SSD) can achieve state-of-the-art unsupervised semantic parsing performance on multiple datasets. |
| Approach: | They propose an unsupervised method which solves the semantic gap and the structure gap by leveraging paraphrasing and grammar-constrained decoding. |
| Outcome: | The proposed method can solve the semantic gap and structure gap on multiple datasets. |
TruthReader: Towards Trustworthy Document Assistant Chatbot with Reliable Attribution (2024.emnlp-demo)
Copied to clipboard
| Challenge: | Document assistant chatbots are empowered with extensive capabilities by Large Language Models (LLMs) however, they suffer from hallucinations that are difficult to verify in the context of given documents. |
| Approach: | They propose a document assistant chatbot with reliable attribution that enables users to seek relevant information from given documents. |
| Outcome: | The proposed system generates answers with detailed inline citations, which can be attributed to the original document paragraphs, facilitating verification of factual consistency of the generated text. |
Time-for-Accuracy: Formalizing Chain-of-Thought as an Expansion of Logical Depth (2026.findings-acl)
Copied to clipboard
| Challenge: | Chain-of-thought (CoT) prompting can improve multi-step reasoning, but it is unclear what kind of additional sequential computation longer traces actually enable. |
| Approach: | They propose a deletion-based measure of step necessity under a specified inference interface to operationalize realized depth beyond raw length. |
| Outcome: | The proposed method combines effective logical depth with Bennett's logical depth to show that it is more efficient than a linear model. |
Query Structure Modeling for Inductive Logical Reasoning Over Knowledge Graphs (2023.acl-long)
Copied to clipboard
| Challenge: | Existing methods for inductive reasoning over knowledge graphs lack the ability to model the logical structures of complex queries. |
| Approach: | They propose a structure-modeled textual encoding framework for inductive logical reasoning over KGs that encodes linearized query structures and entities using pre-trained language models to find answers. |
| Outcome: | The proposed framework encodes query structures and entities using pre-trained language models to find answers. |
Task-adaptive Label Dependency Transfer for Few-shot Named Entity Recognition (2023.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for Named Entity Recognition (NER) ignore label dependency, resulting in suboptimal performance. |
| Approach: | They propose a meta-learning method to make label dependency transferable by learning general initialization and individual parameter update rule for label dependency. |
| Outcome: | The proposed method improves existing methods by learning general initialization and individual parameter update rule for label dependency. |
Re3Syn: A Dependency-Based Data Synthesis Framework for Long-Context Post-training (2025.acl-long)
Copied to clipboard
| Challenge: | Existing methods for constructing long-context data by concatenating short documents have overlooked a crucial characteristic of long-constituency data quality, semantic dependency. |
| Approach: | They propose a framework called Retrieval, Dependency Recognition, and Reorder for data synthesis which leverages semantic similarity to retrieve relevant documents and form several batches. |
| Outcome: | The proposed framework leverages semantic similarity to retrieve relevant documents and form several batches. |
FactVerse: A Benchmark for Factual Consistency in Interleaved Image–Text Generation (2026.acl-long)
Copied to clipboard
Yubo Shan, Kun Zhang, Qiming Xu, Liping Cao, Yingying Cao, Jian Zhang, Yu Wang, Jingyuan Li, Yuanzhuo Wang
| Challenge: | Existing benchmarks lack effective mechanisms to evaluate factual consistency in interleaved image-text generation. |
| Approach: | They propose a benchmark dedicated to evaluating factual consistency in interleaved image-text generation. |
| Outcome: | The proposed framework outperforms existing evaluation methods in evaluating factual consistency in interleaved image-text generation. |
RouteMoA: Dynamic Routing without Pre-Inference Boosts Efficient Mixture-of-Agents (2026.acl-long)
Copied to clipboard
Jize Wang, Han Wu, Zhiyuan You, Yiming Song, Yijun Wang, Zifei Shan, Yining Li, Songyang Zhang, Xinyi Le, Cailian Chen, Xinping Guan, Dacheng Tao
| Challenge: | Existing methods for mixing-of-agents (MoA) lack model selection criteria and struggle with large model pools. |
| Approach: | They propose a mixture-of-agents framework with dynamic routing that uses a lightweight scorer to perform initial screening and refines the model scores through self- and cross-assessment. |
| Outcome: | The proposed framework outperforms existing methods for large model pools and tasks . it reduces cost by 89.8% and latency by 63.6% in the large-scale model pool. |
ThinkAnswer Loss: Balancing Semantic Similarity and Exact Matching for LLM Reasoning Enhancement (2025.findings-emnlp)
Copied to clipboard
| 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. |
On the Perception Bottleneck of VLMs for Chart Understanding (2025.findings-emnlp)
Copied to clipboard
| Challenge: | a perception bottleneck in large vision-language models is critical for chart understanding . instruction tuning improves the extraction capability of LVLMs, but the vision encoder remains a bottleneck . |
| Approach: | They propose to decompose the perception bottleneck into two components . the vision encoder bottleneck is where visual representation fails to encapsulate the correct information . |
| Outcome: | The proposed approach significantly mitigates the vision encoder bottleneck and improves the ability of LVLMs to comprehend charts. |
Stephanie: Step-by-Step Dialogues for Mimicking Human Interactions in Social Conversations (2025.findings-naacl)
Copied to clipboard
Hao Yang, Hongyuan Lu, Xinhua Zeng, Yang Liu, Xiang Zhang, Haoran Yang, Yumeng Zhang, Shan Huang, Yiran Wei, Wai Lam
| Challenge: | a new paradigm for dialogue systems is being developed to mimic human interactions . the current single-step dialogue paradigm lacks the depth and fluidity of human interactions. |
| Approach: | They propose a step-by-step dialogue paradigm that mimics human interactions . they use a dataset to fine-tune existing language models . |
| Outcome: | The proposed system mimics the dynamic nature of human conversations . it is compared with existing paradigms and will be released later this year . |
Lexical Diversity-aware Relevance Assessment for Retrieval-Augmented Generation (2025.acl-long)
Copied to clipboard
| Challenge: | Extensive experiments on widely used benchmarks demonstrate the efficacy of our approach, yielding a 10.6% accuracy improvement on HotpotQA. |
| Approach: | They propose a Lexical Diversity-aware RAG method to address the biases in relevant information retrieval and utilization induced by lexical diversity. |
| Outcome: | Extensive experiments on widely used benchmarks show the proposed method yields a 10.6% accuracy improvement on HotpotQA. |
Evaluating Cognitive-Behavioral Fixation via Multimodal User Viewing Patterns on Social Media (2025.emnlp-main)
Copied to clipboard
| Challenge: | Digital media platforms often contribute to cognitive-behavioral fixation, a phenomenon in which users exhibit sustained and repetitive engagement with narrow content domains. |
| Approach: | They propose a multimodal topic extraction module and a cognitive-behavioral fixation quantification module that collaboratively enable adaptive, hierarchical, and interpretable assessment of user behavior. |
| Outcome: | The proposed framework lays the groundwork for scalable computational analysis of cognitive fixation. |
MMAU: A Holistic Benchmark of Agent Capabilities Across Diverse Domains (2025.findings-naacl)
Copied to clipboard
Guoli Yin, Haoping Bai, Shuang Ma, Feng Nan, Yanchao Sun, Zhaoyang Xu, Shen Ma, Jiarui Lu, Xiang Kong, Aonan Zhang, Dian Ang Yap, Yizhe Zhang, Karsten Ahnert, Vik Kamath, Mathias Berglund, Dominic Walsh, Tobias Gindele, Juergen Wiest, Zhengfeng Lai, Xiaoming Simon Wang, Jiulong Shan, Meng Cao, Ruoming Pang, Zirui Wang
| Challenge: | Existing benchmarks focus on specific application scenarios, emphasizing task completion but failing to dissect the underlying skills that drive these outcomes. |
| Approach: | They propose a Massive Multitask Agent Understanding benchmark that evaluates LLMs across five domains and offline tasks. |
| Outcome: | The Massive Multitask Agent Understanding (MMAU) benchmark evaluates models across five domains including Tool-use, Directed Acyclic Graph (DAG) QA, Data Science and Machine Learning coding, Contest-level programming and Mathematics. |
Characterizing the Impacts of Instances on Robustness (2023.findings-acl)
Copied to clipboard
Rui Zheng, Zhiheng Xi, Qin Liu, Wenbin Lai, Tao Gui, Qi Zhang, Xuanjing Huang, Jin Ma, Ying Shan, Weifeng Ge
| Challenge: | Existing defense approaches focus on developing new model structures or training algorithms, but they do little to tap the potential of training instances. |
| Approach: | They propose a method that can distinguish between robust and non-robust instances according to the model’s sensitivity to perturbations on individual instances during training. |
| Outcome: | The proposed method can distinguish between robust and non-robust instances according to the model’s sensitivity to perturbations on individual instances during training. |
Read Anywhere Pointed: Layout-aware GUI Screen Reading with Tree-of-Lens Grounding (2024.emnlp-main)
Copied to clipboard
| Challenge: | Existing models for GUI understanding ignore a key GUI-referring task: screen reading based on user-indicated points. |
| Approach: | They propose a Tree-of-Lens agent that constructs a Hierarchical Layout Tree based on user input points and a GUI screenshot. |
| Outcome: | The proposed agent can interpret the Screen Point-and-Read task on mobile, web, and operating systems. |
A Contextual Hierarchical Attention Network with Adaptive Objective for Dialogue State Tracking (2020.acl-main)
Copied to clipboard
| Challenge: | Existing methods for dialogue state tracking ignore the slot imbalance problem and treat all slots indiscriminately, which limits the learning of hard slots. |
| Approach: | They propose to employ a contextual hierarchical attention network to enhance the DST by learning contextual representations. |
| Outcome: | The proposed approach achieves 52.68% and 58.55% joint accuracy on multiWOZ 2.0 and MultiWOZ 2.1 datasets and significantly improves performance (+1.24% and +5.98%) |
Keep it Consistent: Topic-Aware Storytelling from an Image Stream via Iterative Multi-agent Communication (2020.coling-main)
Copied to clipboard
| Challenge: | Existing methods for visual storytelling construct text description independently for each image and roughly concatenate them as a story, which leads to the problem of generating semantically incoherent content. |
| Approach: | They propose a topic description task to detect the global semantic context of an image stream and a story is then constructed with the guidance of the topic description. |
| Outcome: | The proposed framework can generate stories with higher quality compared to state-of-the-art methods on a VIST dataset. |
KCL: Few-shot Named Entity Recognition with Knowledge Graph and Contrastive Learning (2024.lrec-main)
Copied to clipboard
| Challenge: | Named Entity Recognition (NER) is a key subtask in natural language processing but is limited to a few labeled samples. |
| Approach: | They propose a few-shot method that harnesses the power of Knowledge Graph and Contrastive Learning to improve the prototypical semantic space learning. |
| Outcome: | The proposed method improves the prototypical semantic space learning by using knowledge graphs and contrastive learning to learn the label semantic representation. |
On the Universal Adversarial Perturbations for Efficient Data-free Adversarial Detection (2023.findings-acl)
Copied to clipboard
| Challenge: | Existing adversarial detection methods require access to training data, which brings noteworthy concerns regarding privacy leakage and generalizability. |
| Approach: | They propose a data-agnostic adversarial detection framework which induces different responses between normal and adversarials to UAPs. |
| Outcome: | The proposed framework achieves competitive detection performance on various text classification tasks, and maintains equivalent time consumption to normal inference. |
StepCoder: Improving Code Generation with Reinforcement Learning from Compiler Feedback (2024.acl-long)
Copied to clipboard
Shihan Dou, Yan Liu, Haoxiang Jia, Enyu Zhou, Limao Xiong, Junjie Shan, Caishuang Huang, Xiao Wang, Xiaoran Fan, Zhiheng Xi, Yuhao Zhou, Tao Ji, Rui Zheng, Qi Zhang, Tao Gui, Xuanjing Huang
| Challenge: | Existing work integrates reinforcement learning with compiler feedback to enhance code generation quality but the long code generated by LLMs makes RL exploration ineffective. |
| Approach: | They propose a framework that integrates reinforcement learning and compiler feedback to enhance code generation quality. |
| Outcome: | The proposed framework outperforms state-of-the-art approaches in corresponding benchmarks and integrates reinforcement learning with compiler feedback to improve code generation quality. |
Entity Concept-enhanced Few-shot Relation Extraction (2021.acl-short)
Copied to clipboard
| Challenge: | Existing FSRE methods fail to classify relations based on information of sentences and entity pairs due to limited samples and lack of knowledge. |
| Approach: | They propose a concept-sentence attention module to select the most appropriate concept from multiple concepts of each entity by calculating the semantic similarity between sentences and concepts. |
| Outcome: | The proposed scheme outperforms existing methods on a few-shot relation extraction dataset. |
LaERC-S: Improving LLM-based Emotion Recognition in Conversation with Speaker Characteristics (2025.coling-main)
Copied to clipboard
| Challenge: | Emotion recognition in conversation (ERC) is a task of discerning human emotions for each utterance within a conversation. |
| Approach: | They propose a framework that uses large language models to analyze speaker characteristics . they use two-stage learning to make the models reason speaker characteristics and track emotion of the speaker . |
| Outcome: | The proposed framework outperforms existing methods on three benchmark datasets. |
A Survey of Large Language Model-Based Search Agents (2026.acl-long)
Copied to clipboard
Yunjia Xi, Jianghao Lin, Yongzhao Xiao, Zheli Zhou, Rong Shan, Te Gao, Jiachen Zhu, Weiwen Liu, Yong Yu, Weinan Zhang
| Challenge: | Large Language Models (LLMs) have revolutionized web search, but their integration is static and cannot handle complex contexts. |
| Approach: | They analyze existing research and analyze existing work from the perspectives of architecture, optimization, application, and evaluation. |
| Outcome: | The proposed models can comprehend user intentions and context and execute multi-turn retrieval with dynamic planning, extending search capabilities far beyond the web. |
A Comprehensive Survey of Process Reward Models: Data Generation, Model Construction, and Usage (2026.acl-long)
Copied to clipboard
Congmin Zheng, Jiachen Zhu, Zhuoying Ou, Yuxiang Chen, Kangning Zhang, Rong Shan, Zeyu Zheng, Mengyue Yang, Jianghao Lin, Yong Yu, Weinan Zhang
| Challenge: | Large Language Models (LLMs) have advanced reasoning ability, yet conventional alignment remains dominated by outcome reward models that judge only final answers. |
| Approach: | They summarize applications across math, code, text, multimodal reasoning, robotics, and agents . goal is to clarify design spaces, reveal open challenges, and guide future research toward fine-grained, robust reasoning alignment. |
| Outcome: | The proposed model enables finer credit assignment, richer diagnostics, and improved robustness. |
Distill Visual Chart Reasoning Ability from LLMs to MLLMs (2025.findings-emnlp)
Copied to clipboard
Wei He, Zhiheng Xi, Wanxu Zhao, Xiaoran Fan, Yiwen Ding, Zifei Shan, Tao Gui, Qi Zhang, Xuanjing Huang
| Challenge: | a new method for generating chart annotations is proposed to improve visual reasoning in multimodal large language models. |
| Approach: | They propose a code-as-intermediary translation method for distilling visual reasoning abilities from LLMs to MLLMs. |
| Outcome: | The proposed method is cost-effective, efficient and scalable. |
TextFlint: Unified Multilingual Robustness Evaluation Toolkit for Natural Language Processing (2021.acl-demo)
Copied to clipboard
Xiao Wang, Qin Liu, Tao Gui, Qi Zhang, Yicheng Zou, Xin Zhou, Jiacheng Ye, Yongxin Zhang, Rui Zheng, Zexiong Pang, Qinzhuo Wu, Zhengyan Li, Chong Zhang, Ruotian Ma, Zichu Fei, Ruijian Cai, Jun Zhao, Xingwu Hu, Zhiheng Yan, Yiding Tan, Yuan Hu, Qiyuan Bian, Zhihua Liu, Shan Qin, Bolin Zhu, Xiaoyu Xing, Jinlan Fu, Yue Zhang, Minlong Peng, Xiaoqing Zheng, Yaqian Zhou, Zhongyu Wei, Xipeng Qiu, Xuanjing Huang
| Challenge: | Existing approaches to textual robustness evaluation focus on slightly modifying the input data, which maintains the original meaning and results in a different prediction. |
| Approach: | They propose a multilingual robustness evaluation toolkit for NLP that integrates universal text transformations, task-specific transformations and adversarial attack. |
| Outcome: | The toolkit includes universal text transformation, task-specific transformation, adversarial attack, subpopulation, and their combinations to provide comprehensive robustness analyses. |
WildFeedback: Aligning LLMs With In-situ User Interactions And Feedback (2026.acl-long)
Copied to clipboard
Taiwei Shi, Zhuoer Wang, Longqi Yang, Ying-Chun Lin, Zexue He, Mengting Wan, Pei Zhou, Sujay Kumar Jauhar, Sihao Chen, Shan Xia, Hongfei Zhang, Jieyu Zhao, Xiaofeng Xu, Xia Song, Jennifer Neville
| Challenge: | Traditional alignment methods rely on human annotations and are subjective and misalignment with real-world user preferences. |
| Approach: | They propose a framework that leverages in-situ user feedback during conversations with LLMs to create preference datasets automatically. |
| Outcome: | The proposed framework identifies and classifies user feedback to LLM responses between conversation turns and creates examples of preferred and dispreferred responses according to user preferences. |
DSRM: Boost Textual Adversarial Training with Distribution Shift Risk Minimization (2023.acl-long)
Copied to clipboard
| Challenge: | Existing adversarial training methods require multi-step gradient ascents or word substitutions to obtain adversarials, which impairs the effectiveness of adversariarial training. |
| Approach: | They propose a procedure for instead adversarial training with only clean data that estimates the adversarials loss by perturbing the input data’s probability distribution rather than their embeddings. |
| Outcome: | The proposed procedure reduces time consumption by up to 70% compared to current best-performing adversarial training methods. |