Qiang Sheng's Publications

FactGuard: Agentic Video Misinformation Detection via Reinforcement Learning
Zehao Li, Hongwei Yu, Hao Jiang, Qiang Sheng, Yilong Xu, Baolong Bi, Yang Li, Zhenlong Yuan, Yujun Cai, and Zhaoqi Wang
Preprint

PhantomHunter: Detecting Unseen Privately-Tuned LLM-Generated Text via Family-Aware Learning
Yuhui Shi, Yehan Yang, Qiang Sheng, Hao Mi, Beizhe Hu, Chaoxi Xu, and Juan Cao
Preprint / Media Coverage: Unite.AI

IEEE TVCG Interactive Visual Assessment for Text-to-Image Generation Models
Xiaoyue Mi, Fan Tang, Juan Cao, Qiang Sheng, Ziyao Huang, Peng Li, Yang Liu, Tong-Yee Lee
IEEE Transactions on Visualization and Computer Graphics
Preprint / Paper / BibTeX
TL;DR: We build an interactive visual assessment tool for exposing T2I models' vulnerability.

EACL'26 Tailoring Rumor Debunking to You: Diversifying Chinese Rumor-Debunking Passages with an LLM-Driven Simulated Feedback-Enhanced Framework
Xinle Pang, Danding Wang, Qiang Sheng, Yifan Sun, Beizhe Hu, and Juan Cao
Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics
Paper (TBA) / BibTeX
TL;DR: We build an LLM-driven framework that generates rumor-debunking passages for various groups of people.

AAAI'26 Reasoning About the Unsaid: Misinformation Detection with Omission-Aware Graph Inference
Zhengjia Wang, Danding Wang, Qiang Sheng, Jiaying Wu, and Juan Cao
Proceedings of the 40th AAAI Conference on Artificial Intelligence (Acceptance Rate: 4167/23680=17.60%)
Preprint / BibTeX
TL;DR: We consider the omitted information to better reason the creator's intent for misinformation detection.

NeurIPS'25 From Judgment to Interference: Early Stopping LLM Harmful Outputs via Streaming Content Monitoring
Yang Li, Qiang Sheng, Yehan Yang, Xueyao Zhang, and Juan Cao
Proceedings of the 39th Annual Conference on Neural Information Processing Systems (Acceptance Rate: 5290/21575=24.52%)
Preprint / Paper (TBA) / Project Page / Github Repo / Chinese Blog / Chinese Video / BibTeX
TL;DR: We build a content moderator that can early stop LLMs' harmful outputs with low latency.

EMNLP'25 The Staircase of Ethics: Probing LLM Value Priorities through Multi-Step Induction to Complex Moral Dilemmas
Ya Wu, Qiang Sheng, Danding Wang, Guang Yang, Yifan Sun, Zhengjia Wang, Yuyan Bu, and Juan Cao
Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing (Acceptance Rate: 1811/8174=22.16%)
Preprint / Paper / Project Page / Chinese Blog / BibTeX
TL;DR: We build the MMD benchmark to probe the LLMs' value preferences in complex moral dilemmas.

EMNLP'25 Findings Forewarned is Forearmed: Pre-Synthesizing Jailbreak-like Instructions to Enhance LLM Safety Guardrail to Potential Attacks
Sheng Liu, Qiang Sheng, Danding Wang, Yang Li, Guang Yang, and Juan Cao
Preprint / Paper / Chinese Blog / BibTeX
Findings of the Association for Computational Linguistics: EMNLP 2025 (Acceptance Rate: 1417/8174=17.34%)
TL;DR: We enhance LLM resistance to jailbreaking by synthesizing instructions that may fall into the potentially risky space.

CIKM'25 Bridging Thoughts and Words: Graph-Based Intent-Semantic Joint Learning for Fake News Detection
Zhengjia Wang, Qiang Sheng, Danding Wang, Beizhe Hu, and Juan Cao
Proceedings of the 34th ACM International Conference on Information and Knowledge Management (Acceptance Rate: 443/1627=27.2%)
Preprint / Paper / Chinese Blog / BibTeX
TL;DR: We propose to inject intent information with graph-based joint learning into fake news detection.

CIKM'25 Enhancing Fake News Video Detection via LLM-Driven Creative Process Simulation
Yuyan Bu, Qiang Sheng, Juan Cao, Shaofei Wang, Peng Qi, Yuhui Shi, and Beizhe Hu
Proceedings of the 34th ACM International Conference on Information and Knowledge Management (Acceptance Rate: 185/604=30.6%)
Preprint / Paper / Media Coverage: 52CV / BibTeX
TL;DR: We propose AgentAug, which simulates the creative process of fake news videos using LLMs to mitigate the data scarcity issue.

IP&M Exploring News Intent and Its Application: A Theory-Driven Approach
Zhengjia Wang, Danding Wang, Qiang Sheng, Juan Cao, Siyuan Ma, and Haonan Cheng
Information Processing and Management, 2025
Preprint / Paper / Chinese Blog / BibTeX
TL;DR: We, for the first time, conceptualize and computerize news intent modeling and showcase its application on fake news detection, popularity prediction, and propaganda detection.

ACL'25 Findings Enhancing the Comprehensibility of Text Explanations via Unsupervised Concept Discovery
Yifan Sun, Danding Wang, Qiang Sheng, Juan Cao, and Jintao Li
Findings of the Association for Computational Linguistics: ACL 2025
Preprint / Paper / Chinese Blog / BibTeX
TL;DR: We exploit LLMs to automatically discover key concepts for text classification to enhance the comprehensibility of text explanations.

SIGIR'25 LLM-Generated Fake News Induces Truth Decay in News Ecosystem: A Case Study on Neural News Recommendation
Beizhe Hu, Qiang Sheng, Juan Cao, Yang Li, and Danding Wang
Proceedings of The 48th International ACM SIGIR Conference on Research and Development in Information Retrieval (Acceptance Rate: 239/1071=22.3%)
Preprint / Paper / Project Page / BibTeX
TL;DR: We reveal the truth-decay phenomenon where real news gradually loses its top-ranked advantage against fake news when LLM-generated news penetrates.

CAS Bulletin Challenges on Public Security System in AI Era
Juan Cao, Qiang Sheng, Guojie Li
Bulletin of Chinese Academy of Sciences
Paper / Media Coverage: 中国教育报, 新华文摘 / BibTeX
TL;DR: We discussed new challenges and countermeasures in public security brought by AIGC techniques.

FCS Exploiting User Comments for Early Detection of Fake News Prior to Users' Commenting
Qiong Nan, Qiang Sheng, Juan Cao, Yongchun Zhu, Danding Wang, Guang Yang, and Jintao Li
Frontiers of Computer Science
Preprint / Paper / BibTeX
TL;DR: We explored how to let a comment-based model effectively help a content-only one in fake news early detection.

CIKM'24 Let Silence Speak: Enhancing Fake News Detection with Generated Comments from Large Language Models
Qiong Nan, Qiang Sheng, Juan Cao, Beizhe Hu, Danding Wang, and Jintao Li
Proceedings of the 33rd ACM International Conference on Information and Knowledge Management (Acceptance Rate: 347/1496=23.2%)
Preprint / Paper / GitHub Repo / Chinese Blog / BibTeX
TL;DR: We prompt LLMs to role-play social media users to obtain generated comments for enhancing fake news detection.

MM'24 FakingRecipe: Detecting Fake News on Short Video Platforms from the Perspective of Creative Process
Yuyan Bu, Qiang Sheng, Juan Cao, Peng Qi, Danding Wang, and Jintao Li
Proceedings of the 32nd ACM International Conference on Multimedia (Acceptance Rate: 1149/4385=26.2%)
Preprint / Paper / GitHub Repo / Chinese Blog / CSIG MM论文导读 / BibTeX
TL;DR: We detect fake news on short video platforms by modeling videos from the faking process perspective and constructed a FakeSV's sister dataset in English, namely FakeTT.

IJCAI'24 Ten Words Only Still Help: Improving Black-Box AI-Generated Text Detection via Proxy-Guided Efficient Re-Sampling
Yuhui Shi, Qiang Sheng, Juan Cao, Hao Mi, Beizhe Hu, and Danding Wang
Proceedings of the 33rd International Joint Conference on Artificial Intelligence (Acceptance Rate: 791/5651=14.0%)
Preprint / Paper / GitHub Repo / Slides / Chinese Blog / BibTeX
TL;DR: To detect and attribute text generated by black-box LMs, we estimate their generation probabilities of representative words guided by a white-box proxy LM to obtain a strong feature.

IJCAI'24 Vision-fused Attack: Advancing Aggressive and Stealthy Adversarial Text against Neural Machine Translation
Yanni Xue, Haojie Hao, Jiakai Wang, Qiang Sheng, Renshuai Tao, Yu Liang, Pu Feng, and Xianglong Liu
Proceedings of the 33rd International Joint Conference on Artificial Intelligence (Acceptance Rate: 791/5651=14.0%)
Paper / GitHub Repo / Chinese Blog / BibTeX
TL;DR: We propose a vision-fused attack framework to acquire powerful adversarial text (i.e., more aggressive and stealthy) in neural machine translation.

AAAI'24 Bad Actor, Good Advisor: Exploring the Role of Large Language Models in Fake News Detection
Beizhe Hu, Qiang Sheng, Juan Cao, Yuhui Shi, Yang Li, Danding Wang, and Peng Qi
Proceedings of the 38th AAAI Conference on Artificial Intelligence
Preprint / Paper / GitHub Repo / English Video & Slides / BibTeX
TL;DR: Large LMs generally underperform fine-tuned Small LMs for fake news detection, but they can be good advisors by providing rationales.

CCCF The Safety in the AIGC era: Techniques Make the World More Credible (AIGC时代的安全：用技术让世界更可信)
Juan Cao, and Qiang Sheng
Communications of the China Computer Federation, October 2023
Paper / WeChat Release / BibTeX
TL;DR: Our perspective paper on the safety issues brought by AI-generated content.

MM'23 Combating Online Misinformation Videos: Characterization, Detection, and Future Directions
Yuyan Bu, Qiang Sheng, Juan Cao, Peng Qi, Danding Wang, and Jintao Li
Proceedings of the 31st ACM International Conference on Multimedia (Acceptance Rate: 902/3669=24.6%)
Preprint / Paper / GitHub Repo / Chinese Blog / BibTeX
TL;DR: We conduct the very first survey focusing on online misinformation video detection, to the best of our knowledge.

ACL'23 Learn over Past, Evolve for Future: Forecasting Temporal Trends for Fake News Detection
Beizhe Hu, Qiang Sheng, Juan Cao, Yongchun Zhu, Danding Wang, Zhengjia Wang, and Zhiwei Jin
Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics
Preprint / Paper / Code / Poster / Slides / English Video / Explanatory AI arXiv Weekly Roundup / BibTeX
TL;DR: We propose to address the temporal shift issue in real-world fake news detection systems via forecasting topic-level trends and accordingly adjusting the detector update strategy.

COLING'22 Improving Fake News Detection of Influential Domain via Domain- and Instance-Level Transfer
Qiong Nan, Danding Wang, Yongchun Zhu, Qiang Sheng, Yuhui Shi, Juan Cao, and Jintao Li
Proceedings of the 29th International Conference on Computational Linguistics (Acceptance Rate: 632/2253=28.1%)
Preprint / Paper / Code / English Video / Chinese Blog 1/ 2 / BibTeX
TL;DR: We propose a domain- and instance-level transfer framework for detecting fake news of target domains.

IEEE TKDE Memory-Guided Multi-View Multi-Domain Fake News Detection
Yongchun Zhu, Qiang Sheng, Juan Cao, Qiong Nan, Kai Shu, Minghui Wu, Jindong Wang, and Fuzhen Zhuang
IEEE Transactions on Knowledge and Data Engineering, 2022
Preprint / Paper / Code / Chinese Blog / BibTeX
TL;DR: We tackle the issue of domain shift and domain labeling incompleteness for simultaneously modeling multi-domain fake news via multi-view encoding and memory bank mechanism.

IP&MCharacterizing Multi-Domain False News and Underlying User Effects on Chinese Weibo
Qiang Sheng, Juan Cao, H. Russell Bernard, Kai Shu, Jintao Li, and Huan Liu
Information Processing and Management, 2022
Preprint / Paper / Code / Dataset / Chinese Brief / BibTeX
TL;DR: Based on the largest false news dataset on Chinese Weibo (44k+), we characterize the spread of false news in nine domains and analyze the underlying user effects.

SIGIR'22 Generalizing to the Future: Mitigating Entity Bias in Fake News Detection
Yongchun Zhu, Qiang Sheng, Juan Cao, Shuokai Li, Danding Wang, and Fuzhen Zhuang
Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 2022) (Acceptance Rate: 165/667=24.7%)
Preprint / Paper / Code / Chinese Blog / BibTeX
TL;DR: We reveal the entity bias in fake news detection datasets and propose a causal framework to mitigate such bias for better generalization to future data.

ACL'22 Zoom Out and Observe: News Environment Perception for Fake News Detection
Qiang Sheng, Juan Cao, Xueyao Zhang, Rundong Li, Danding Wang, and Yongchun Zhu
Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Acceptance Rate: 701/3378=20.75%)
Preprint / Paper / Poster / Code / Dataset / Chinese Video / Chinese Blog / English Video / English Blog / BibTeX
TL;DR: For the first time, we propose to perceive signals from the news environment for fake news detection.

AAAI'22 DRAG: Dynamic Region-Aware GCN for Privacy-Leaking Image Detection
Guang Yang, Juan Cao, Qiang Sheng, Peng Qi, Xirong Li, and Jintao Li
Proceedings of the 36th AAAI Conference on Artificial Intelligence (Acceptance Rate: 1349/9020=15.0%)
Preprint / Paper / English Video / BibTeX
TL;DR: We propose a GCN-based method to find out crucial privacy-indicative regions for privacy-leaking image detection dynamically.

MM'21 Improving Fake News Detection by Using an Entity-enhanced Framework to Fuse Diverse Multimodal Clues
Peng Qi, Juan Cao, Xirong Li, Huan Liu, Qiang Sheng, Xiaoyue Mi, Qin He, Yongbiao Lv, Chenyang Guo, and Yingchao Yu
Proceedings of the 29th ACM International Conference on Multimedia
Preprint / Paper / Chinese Blog 1/ 2 / BibTeX
TL;DR: We point out three valuable text-image correlations in multimodal fake news and propose an entity-enhanced fusion framework for cross-modal correlation modeling in fake news detection.

CIKM'21 Integrating Pattern- and Fact-based Fake News Detection via Model Preference Learning
Qiang Sheng*, Xueyao Zhang*, Juan Cao, and Lei Zhong (*:Equal Contribution)
Proceedings of the 30th ACM International Conference on Information and Knowledge Management (Acceptance Rate:271/1251=21.7%)
Preprint / Paper / Poster / Code / Datasets / Chinese Blog 1/ 2/ 3 / BibTeX
TL;DR: We propose a graph-based model preference learning framework to separately handle the pattern and fact indicators in fake news detection.

ACL'21 Article Reranking by Memory-enhanced Key Sentence Matching for Detecting Previously Fact-checked Claims
Qiang Sheng, Juan Cao, Xueyao Zhang, Xirong Li, and Lei Zhong
Proceedings of the Joint Conference of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Acceptance Rate:571/2327=24.5%)
Preprint / Paper / Poster / Code / Chinese Dataset / English Video / Chinese Blog 1/ 2/ 3 / BibTeX
TL;DR: We detect previously fact-checked claims by matching them against the key sentences in fact-checking articles.

J-CRAD Semantics-Enhanced Multi-modal Fake News Detection (语义增强的多模态虚假新闻检测)
Peng Qi, Juan Cao, and Qiang Sheng
Journal of Computer Research and Development, 2021
Paper / BibTeX
TL;DR: We focus on the role of explicit semantics for fake news detection via visual semantic extraction and knowledge-guided multi-modal semantic interaction.

WWW'21 Mining Dual Emotion for Fake News Detection
Xueyao Zhang, Juan Cao, Xirong Li, Qiang Sheng, Lei Zhong, and Kai Shu
Proceedings of the 30th Web Conference (Acceptance Rate:357/1736=20.6%)
Preprint / Paper / Code / Chinese Video / BibTeX
TL;DR: We leverage both publisher emotion and social emotion for fake news detection.

ACL'20 Integrating Semantic and Structural Information with Graph Convolutional Network for Controversy Detection
Lei Zhong, Juan Cao, Qiang Sheng, Junbo Guo, and Ziang Wang
Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics (Acceptance Rate:571/2244=25.4%)
Paper / Chinese Dataset / BibTeX
TL;DR: We propose a GCN-based method for post-level controversy detection. Also, we release the first Chinese dataset for this task.

CCCF Fake News and Disinformation Detection on the Internet: Recent Advances and Future Prospects (互联网虚假信息检测进展与展望)
Juan Cao, Qiang Sheng, and Peng Qi
Communications of the China Computer Federation, March 2020
Webpage Version / WeChat Release / BibTeX
TL;DR: We review the recent advances of detecting fake news and disinformation and propose potential directions.

Springer Book Chapter Exploring the Role of Visual Content in Fake News Detection
Juan Cao, Peng Qi, Qiang Sheng, Tianyun Yang, Junbo Guo, and Jintao Li
Disinformation, Misinformation, and Fake News in Social Media: Emerging Research Challenges and Opportunities, Lecture Notes in Social Network (LNSN)
Preprint / Springer Link / BibTeX
TL;DR: A survey chapter that focuses on the visual information for fake news detection.