PhD Dissertation Defense by ZENG Fengzhu | Computational Fact-Checking With Limited Resources

Computational Fact-Checking With Limited Resources

ABOUT THE TALK

The rapid spread of misinformation online has raised concerns, but manual fact-checking is time-consuming and lacks scalability. Computational fact-checking, leveraging natural language processing (NLP) techniques, offers a potential solution but faces challenges under limited resource, due to data scarcity and high computational demands. Data scarcity often arises because emerging events on social media outpace the creation of training data for fact-checking, manifesting in two key ways: the limited availability of early-stage data during misinformation propagation and the lack of high-quality, labeled training data for new events. While Large Language Models (LLMs) perform well on various NLP tasks with limited data, their substantial computational requirements create another significant barrier. Effectively addressing fact-checking with limited resources remains a critical yet underexplored area.

To tackle these challenges, this thesis proposes methodologies that can make the most of the available data and computing resources, striking a balance between performance and resource availability. It introduces the stabilization process of early rumor detection to ensure timely, accurate and stable predictions with limited early-stage data. Further, it proposes the first few-shot early rumor detection method for scarce labeled training instances. For fact verification and multimodal misinformation detection, this thesis proposes enabling medium-sized pre-trained models to effectively compete with LLMs and multimodal LLMs, in both few-shot and zero-shot settings. It also presents the first solution for justification generation with evidence retrieval in the few-shot setting, using a medium-sized model. Experiments show these methodologies achieve more effective performance as compared to strong baselines, including LLMs. Finally, this thesis outlines potential future research directions to advance computational fact-checking with limited resources.

SPEAKER BIOGRAPHY

ZENG Fengzhu is a Ph.D. candidate in Computer Science at the School of Computing and Information Systems, Singapore Management University (SMU), under the supervision of Prof. GAO Wei. Her research focuses on Natural Language Processing (NLP) and Large Language Models (LLMs), with a particular interest in enhancing computational fact-checking in limited-resources environments.