Papers by Shu Yang
Flattery in Motion: Benchmarking and Analyzing Sycophancy in Video-LLMs (2026.acl-long)
Copied to clipboard
| Challenge: | Current sycophancy research has largely overlooked its specific manifestations in the video-language domain. |
| Approach: | They propose a video-LLM sycophancy benchmarking and evaluation to evaluate scophancies in video-LLMs. |
| Outcome: | The proposed benchmark evaluates sycophantic behavior in state-of-the-art Video-LLMs across diverse question formats, prompt biases, and visual reasoning tasks. |
Understanding and Mitigating Political Stance Cross-topic Generalization in Large Language Models (2026.acl-long)
Copied to clipboard
| Challenge: | Recent studies have focused on the internal representations of large language models and the mechanisms that lead to unintended cross-topic generalization. |
| Approach: | They propose a method that uses inhibition to localize political neurons and a technique that uses topic-specific blocking to mitigate the cross-topic generalization. |
| Outcome: | The proposed method reduces cross-topic generalization by 20% while preserving topic-specific performance. |
DALK: Dynamic Co-Augmentation of LLMs and KG to answer Alzheimer’s Disease Questions with Scientific Literature (2024.findings-emnlp)
Copied to clipboard
Dawei Li, Shu Yang, Zhen Tan, Jae Baik, Sukwon Yun, Joseph Lee, Aaron Chacko, Bojian Hou, Duy Duong-Tran, Ying Ding, Huan Liu, Li Shen, Tianlong Chen
| Challenge: | Recent advances in large language models have achieved promising performances across various applications, but the challenge of integrating long-tail knowledge continues to impede the seamless adoption of LLMs in specialized domains. |
| Approach: | They propose a dynamic co-augmentation framework for the refinement of large language models and knowledge graphs in the context of Alzheimer's Disease. |
| Outcome: | The proposed framework can be used to study Alzheimer's Disease (AD) using LLMs and KGs. |
Understanding How Value Neurons Shape the Generation of Specified Values in LLMs (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Current approaches to interpret value representations are limited by superficial judgments over mechanistic analysis. |
| Approach: | They propose a mechanistic interpretability framework that uses the Schwartz Values Survey to interpret value . they use a dataset that operationalizes four dimensions of universal value through behavioral contexts . |
| Outcome: | The proposed method bridges psychological value frameworks with neuron analysis in large language models. |
Forget the Unneeded: Backdooring Large Language Models via Contrastive-enhanced Machine Unlearning (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods for prompt tuning for Large Language Models find backdoor attacks to be significant in data-rich scenarios. |
| Approach: | They propose a backdoor attacks through contrastive-enhanced machine unlearning in data-limited scenarios . they use a machine un learning method to capture precise backdoor patterns . |
| Outcome: | The proposed method captures precise backdoor patterns without association between triggers and backdoors, reducing side effects. |
Visual Self-Fulfilling Alignment: Shaping Safety-Oriented Personas via Threat-Related Images (2026.acl-long)
Copied to clipboard
| Challenge: | Existing methods require explicit safety labels or contrastive data, yet visual inputs enable harmful outputs. |
| Approach: | They propose a visual self-fulfilling alignment mechanism that fine-tunes vision-language models on neutral VQA tasks without any safety labels. |
| Outcome: | The proposed approach reduces attack success rate, improves response quality, and mitigates over-refusal while preserving general capabilities. |
Taxonomy-Guided Zero-Shot Recommendations with LLMs (2025.coling-main)
Copied to clipboard
| Challenge: | Existing approaches to deploy large language models (LLMs) into RecSys have limited prompt length, unstructured item information, and un-constrained generation of recommendations. |
| Approach: | They propose a taxonomy-guided recommendation framework that empowers LLMs with category information in a systematic approach. |
| Outcome: | The proposed framework significantly improves recommendation quality compared to zero-shot approaches. |
AutoMonitor-Bench: Evaluating the Reliability of LLM-Based Misbehavior Monitor (2026.findings-acl)
Copied to clipboard
| Challenge: | AutoMonitor-Bench evaluates the reliability of LLM-based misbehavior monitors across diverse tasks and failure modes. |
| Approach: | They introduce AutoMonitor-Bench, a benchmark designed to evaluate misbehavior monitors across diverse tasks and failure modes. |
| Outcome: | The new benchmark evaluates the reliability of LLM-based misbehavior monitors across tasks and failure modes. |
RAVEN++: Pinpointing Fine-Grained Violations in Advertisement Videos with Active Reinforcement Reasoning (2025.emnlp-industry)
Copied to clipboard
Deyi Ji, Yuekui Yang, Liqun Liu, Peng Shu, Haiyang Wu, Shaogang Tang, Xudong Chen, Shaoping Ma, Tianrun Chen, Lanyun Zhu
| Challenge: | Recent advances in large language models have improved the detection of non-compliant content, but critical gaps persist in fine-grained understanding, explainability, and generalization. |
| Approach: | They propose a framework that combines active reinforcement learning, fine-grained violation understanding and progressive multi-stage training. |
| Outcome: | The proposed framework outperforms general-purpose LLMs and specialized models in fine-grained violation understanding, explainability, and generalization. |
JARVIS or Ultron? A Survey on the Safety and Security Threats of Computer-Using Agents (2026.acl-long)
Copied to clipboard
Ada Chen, Yongjiang Wu, Junyuan Zhang, Jingyu Xiao, Shu Yang, Jen-tse Huang, Kun Wang, Wenxuan Wang, Shuai Wang
| Challenge: | Recent advances in computer-using agents have created new safety and security risks . despite the impressive capabilities of CUAs, there are still significant security risks. |
| Approach: | They propose a systematization of knowledge on the safety and security threats of Computer-Using Agents. |
| Outcome: | The proposed framework provides a framework for assessing the safety and security risks of computer-using agents. |
Who Wrote This? The Key to Zero-Shot LLM-Generated Text Detection Is GECScore (2025.coling-main)
Copied to clipboard
| Challenge: | Existing methods for detecting LLM-generated text require no training data. |
| Approach: | They propose a black-box zero-shot detection approach that calculates the Grammar Error Correction Score for a given text to differentiate between human-written and LLM-generated texts. |
| Outcome: | The proposed method outperforms current state-of-the-art zero-shot and supervised methods, achieving an average AUROC of 98.62% across XSum and Writing Prompts datasets. |
Real-time Ad Retrieval via LLM-generative Commercial Intention for Sponsored Search Advertising (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods for retrieving documents and ads use one-to-few mappings and time-consuming content extraction. |
| Approach: | They propose a framework that leverages LLM-generated commercial intents as an intermediate semantic representation to directly retrieve ads for queries in real-time. |
| Outcome: | The proposed framework has been implemented in a real-world online system, handling daily search volumes in billions. |
MedDialog: Large-scale Medical Dialogue Datasets (2020.emnlp-main)
Copied to clipboard
Guangtao Zeng, Wenmian Yang, Zeqian Ju, Yue Yang, Sicheng Wang, Ruisi Zhang, Meng Zhou, Jiaqi Zeng, Xiangyu Dong, Ruoyu Zhang, Hongchao Fang, Penghui Zhu, Shu Chen, Pengtao Xie
| Challenge: | telemedicine is a medical practice that provides patient care remotely using video conferencing tools. |
| Approach: | They build large-scale medical dialogue datasets to facilitate research . they pretrain several models on the Chinese MedDialog dataset and compare their performance . |
| Outcome: | The proposed datasets show that models trained on MedDialog can generate doctor-like medical dialogues. |
Fraud-R1 : A Multi-Round Benchmark for Assessing the Robustness of LLM Against Augmented Fraud and Phishing Inducements (2025.findings-acl)
Copied to clipboard
Shu Yang, Shenzhe Zhu, Zeyu Wu, Keyu Wang, Junchi Yao, Junchao Wu, Lijie Hu, Mengdi Li, Derek F. Wong, Di Wang
| Challenge: | Existing fraud detection benchmarks focus on single-turn classification tasks, failing to capture dynamic nature of real-world fraud attempts. |
| Approach: | They propose a bilingual benchmark to assess LLMs' ability to resist fraud and phishing attacks across five key fraud categories: Fraudulent Services, Impersonation, Phishing Scams, Fake Job Postings, and Online Relationships. |
| Outcome: | The proposed model improves in role-play settings and in e-commerce and recommendation systems. |
Rethinking Prompt-based Debiasing in Large Language Model (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing prompt-based methods for debiasing are often superficial and lack a thorough understanding of complex bias concepts. |
| Approach: | They analyze a BBQ and stereoSet benchmarks to examine the assumption that large language models understand biases. |
| Outcome: | The proposed model misclassified 90% of unbiased content as biased despite high accuracy on BBQ dataset . the proposed model may have been flawed in previous attempts to debiase . |
Piecing It All Together: Verifying Multi-Hop Multimodal Claims (2025.coling-main)
Copied to clipboard
| Challenge: | Existing claim verification datasets often do not require systems to perform complex reasoning or effectively interpret multimodal evidence. |
| Approach: | They propose a task that requires models to reason over multiple pieces of evidence . they construct a large-scale dataset comprising 15k multi-hop claims paired with multimodal evidence - generated and refined using large language models with additional input from human feedback. |
| Outcome: | The proposed method is based on human performance benchmarks and human reasoning hops. |
Understanding the Repeat Curse in Large Language Models from a Feature Perspective (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models suffer from repetitive text generation, a phenomenon we refer to as the ”Repeat Curse”. |
| Approach: | They propose a method to induce and analyze the Repeat Curse in large language models by using mechanistic interpretability. |
| Outcome: | The proposed method induces and analyzes the Repeat Curse in large language models using mechanistic interpretability. |
M3TQA: Massively Multilingual Multitask Table Question Answering (2026.findings-acl)
Copied to clipboard
Daixin Shu, Jian Yang, Zhenhe Wu, Xianjie Wu, Xianfu Cheng, Guan Xiangyuan, Yanghai Wang, Pengfei Wu, Tingyang Yang, Hualei Zhu, Wei Zhang, Ge Zhang, Jiaheng Liu, Zhoujun Li
| Challenge: | Existing multilingual table benchmarks suffer from geolinguistic imbalance - overrepresenting certain languages and lacking sufficient scale for rigorous cross-lingual analysis. |
| Approach: | They propose a framework for massively multilingual table question answering that includes tables expanded to 97 languages from Chinese and English sources. |
| Outcome: | Experiments on state-of-the-art LLMs show that synthetically generated training data significantly boosts performance, especially for low-resource languages. |
Rejection-to-Acceptance Transition: Model Editing-Based Jailbreak Backdoor Injection Not Limited to Few Output Tokens (2026.findings-acl)
Copied to clipboard
Shiji Yang, Min Cai, Hao Xiong, Congyao Mei, Haodong Zou, Shicheng Tan, Jie Chen, Fulan Qian, Shu Zhao
| Challenge: | Existing methods for jailbreaking LLMs are implemented by binding backdoors to predefined phrases as first few output tokens, inducing the LLM’s next-token prediction to produce continuous responses. |
| Approach: | They propose a model editing-based jailbreak backdoor attack that hijacks LLM representations into a acceptance domain rather than binding to a few output tokens. |
| Outcome: | The proposed model editing method outperforms existing methods, showing stronger jailbreak capabilities across LLMs and datasets. |
On the Generation of Medical Dialogs for COVID-19 (2021.acl-short)
Copied to clipboard
Meng Zhou, Zechen Li, Bowen Tan, Guangtao Zeng, Wenmian Yang, Xuehai He, Zeqian Ju, Subrato Chakravorty, Shu Chen, Xingyi Yang, Yichen Zhang, Qingyang Wu, Zhou Yu, Kun Xu, Eric Xing, Pengtao Xie
| Challenge: | under the pandemic of COVID-19, people experiencing COVI D19-related symptoms have a pressing need to consult doctors. |
| Approach: | They develop a medical dialog system that can provide COVID19-related consultations . they use two dialog datasets containing conversations between doctors and patients . |
| Outcome: | The proposed system can provide COVID19-related consultations, but is too small compared with general-domain dialog datasets. |
Do Vision-Language Models Have Internal World Models? Towards an Atomic Evaluation (2025.findings-acl)
Copied to clipboard
Qiyue Gao, Xinyu Pi, Kevin Liu, Junrong Chen, Ruolan Yang, Xinqi Huang, Xinyu Fang, Lu Sun, Gautham Kishore, Bo Ai, Stone Tao, Mengyang Liu, Jiaxi Yang, Chao-Jung Lai, Chuanyang Jin, Jiannan Xiang, Benhao Huang, Zeming Chen, David Danks, Hao Su, Tianmin Shu, Ziqiao Ma, Lianhui Qin, Zhiting Hu
| Challenge: | Recent studies have evaluated and shown limitations in specific capabilities such as visual understanding, but a systematic evaluation of VLMs’ fundamental WM abilities remains absent. |
| Approach: | They propose a framework that assesses perception and prediction to provide an atomic evaluation of VLMs as WMs. |
| Outcome: | The proposed framework assesses perception and prediction abilities on 15 latest VLMs and compares them to human-level models. |
Can Large Language Models Identify Implicit Suicidal Ideation? An Empirical Evaluation (2025.findings-emnlp)
Copied to clipboard
Tong Li, Shu Yang, Junchao Wu, Jiyao Wei, Lijie Hu, Mengdi Li, Derek F. Wong, Joshua R. Oltmanns, Di Wang
| Challenge: | Existing data on suicidal ideation in private conversations are limited . a new dataset of 1,200 test cases is presented to address this gap . |
| Approach: | They propose a dataset of 1,200 test cases simulating implicit suicidal ideation in private contexts. |
| Outcome: | The proposed dataset includes 1,200 test cases simulating implicit suicidal ideation in dialogue scenarios. |
TableLLM: Enabling Tabular Data Manipulation by LLMs in Real Office Usage Scenarios (2025.findings-acl)
Copied to clipboard
Xiaokang Zhang, Sijia Luo, Bohan Zhang, Zeyao Ma, Jing Zhang, Yang Li, Guanlin Li, Zijun Yao, Kangli Xu, Jinchang Zhou, Daniel Zhang-Li, Jifan Yu, Shu Zhao, Juanzi Li, Jie Tang
| Challenge: | TableLLM is a robust large language model capable of handling tabular data manipulation tasks. |
| Approach: | They propose a distant supervision method for training which includes a reasoning process extension strategy and a cross-way validation strategy. |
| Outcome: | The proposed model has 8 billion parameters and is capable of handling tabular data tasks. |
FinDABench: Benchmarking Financial Data Analysis Ability of Large Language Models (2025.coling-main)
Copied to clipboard
Shu Liu, Shangqing Zhao, Chenghao Jia, Xinlin Zhuang, Zhaoguang Long, Jie Zhou, Aimin Zhou, Man Lan, Yang Chong
| Challenge: | Large Language Models (LLMs) have demonstrated impressive capabilities across a wide range of tasks, but their proficiency and reliability in the specialized domain of financial data analysis remain uncertain. |
| Approach: | FinDABench is a benchmark designed to evaluate the financial data analysis capabilities of Large Language Models (LLMs) it comprises 15,200 training instances and 8,900 test instances, all meticulously crafted by human experts. |
| Outcome: | FinDABench measures the financial data analysis capabilities of large language models (LLMs) across three dimensions: 1) Core Ability; 2) Analytical Ability; 3) Technical Ability. |
Execution as Verification: Fine-Grained Self-Correcting Reasoning for Complex KBQA (2026.acl-long)
Copied to clipboard
| Challenge: | Existing knowledge base question answering methods are limited by syntactic constraints and are prone to structural deviations that render queries unexecutable. |
| Approach: | They propose a framework that reframes semantic parsing as an iterative reasoning process driven by execution feedback. |
| Outcome: | The proposed method achieves significant improvements in query executability and answer accuracy on the WebQSP and CWQ datasets. |
A Good Plan is Hard to Find: Aligning Models with Preferences is Misaligned with What Helps Users (2025.emnlp-main)
Copied to clipboard
Nishant Balepur, Matthew Shu, Yoo Yeon Sung, Seraphina Goldfarb-Tarrant, Shi Feng, Fumeng Yang, Rachel Rudinger, Jordan Lee Boyd-Graber
| Challenge: | We test alignment methods to ensure LLMs are helpful, but they train or evaluate on what users prefer . |
| Approach: | They test alignment methods to ensure LLMs generate plans that help users . they get 4388 plan executions and 5584 comparisons to measure user preferences . |
| Outcome: | The proposed approach can be applied to the problem of user preferences and helpfulness. |
CompTab: A Comprehensive Benchmark for Real-World TableQA with Complex Reasoning and Irregular Tables (2026.acl-long)
Copied to clipboard
Zhen Yang, Wei Du, Jie Wang, Wenze Zhou, Xiangfeng Meng, Zhengyang Wang, Suping Sun, Ziwei Du, Haodong Zou, Jie Chen, Yongbin Liu, Shicheng Tan, Jiahao Ying, Shu Zhao
| Challenge: | Existing benchmarks focus on well-structured tables and fail to reflect irregular structures and complex reasoning commonly encountered in real-world scenarios. |
| Approach: | They propose a benchmark to evaluate TableQA under complex reasoning and irregular table conditions. |
| Outcome: | The proposed framework improves generalization and realism of large language models under complex and irregular table conditions. |
Thinking Beyond the Local: Multi-View Instructed Adaptive Reasoning in KG-Enhanced LLMs (2026.findings-eacl)
Copied to clipboard
| Challenge: | Existing methods for large language models adopt query-driven iterative reasoning from a local perspective, limiting efficiency and accuracy for complex multi-hop tasks. |
| Approach: | They propose a multi-view instructed adaptive reasoning of LLM on Knowledge Graphs that allows LLMs to plan, evaluate, and adapt reasoning paths from a global perspective. |
| Outcome: | The proposed model overcomes the limitations of local exploration by enabling LLMs to plan, evaluate, and adapt reasoning paths from a global perspective. |
SepSeq: A Training-Free Framework for Long Numerical Sequence Processing in LLMs (2026.findings-acl)
Copied to clipboard
Jie Sun, Yu Liu, Lu Han, Qiwen Deng, Xiang Shu, Yang Xiao, Lintao Ma, Xingyu Lu, Jun Zhou, Pengfei Liu, Jiancan Wu, Xiang Wang
| Challenge: | Existing large-scale large-context models suffer from performance degradation when processing long numerical sequences. |
| Approach: | They propose a framework to mitigate attention dispersion by strategically inserting separator tokens into the model to recalibrat attention to local segments while preserving global context. |
| Outcome: | The proposed framework improves accuracy and reduces inference token consumption by 16.4% on 9 widely-adopted LLMs. |
Self-Reflective Generation at Test Time (2026.acl-long)
Copied to clipboard
| Challenge: | Existing self-reflection mechanisms are reactive and inefficient for large language models . a fundamental tension persists between the ability to execute complex multi-step reasoning and the ability of the model to generate coherent traces. |
| Approach: | They propose a test-time framework that reflects before generating at uncertain points . SRGen utilizes dynamic entropy thresholding to identify high-uncertainty tokens . |
| Outcome: | The proposed framework can significantly strengthen large language models' reasoning process. |
RLShield: Dynamic Jailbreak Detection for LLMs via Reinforced Adaptive Learning (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to detect jailbreak prompts rely on static model components or fixed decision thresholds. |
| Approach: | They propose a dynamic jailbreak detection framework that employs reinforcement learning for adaptive threshold selection. |
| Outcome: | Experimental results show that the framework outperforms baselines in detection performance while maintaining high computational efficiency. |