Time: 11:00 - 12:00
Fact-checking in technical terms is the process of analyzing textual content for claim veracity detection. To mitigate the time and the human burden of fact-checking and to allow for more fact-checked articles, the use of ML and NLP to automate the fact-checking procedure has attracted notable interest in the recent decade. Automated fact-checking can be decomposed into three major sub-tasks including claim detection, evidence retrieval/ranking, and claim verification. Factual verification is the term used to refer to evidence retrieval and verification, integrated into a single task. Stance detection has also been considered as a distinct component of the pipeline, but it is basically part of the verification procedure. Another integrant of the verification sometimes addressed as a separate task is justification/explanation generation. The proposed solutions either result in end-to-end fact-checking systems or tackle certain tasks in the pipeline. This seminar covers the scope of the automated fact-checking problem in recent literature focusing on several approaches based on traditional machine learning techniques, deep learning models, information retrieval systems, and knowledge graphs.
Ghazaal Sheikhi is a Postdoctoral Fellow in the Department of Information Science and Media Studies, University of Bergen, Norway. Her research interests revolve around machine learning, natural language processing and content analysis. Ghazaal investigates (semi) automated methods and tools to support fact-checking in newsrooms. In MediaFutures, she studies AI systems, NLP models and, ML methods to support fact-checking in newsrooms, particularly aiming at enhancing claim detection, verification, and justification.
Before, Ghazaal was an Assistant Professor in the Faculty of Engineering, Final International University, North Cyprus. She holds a PhD in Computer Engineering (Machine Learning) from Eastern Mediterranean University, North Cyprus and a master’s degree in Biomedical Engineering from the Amirkabir University of Technology, Tehran, Iran.
TITLE: Automated Fact-checking: the scope of the problem in the state-of-the-art
WHEN: Friday 14 January, 11:00-12:00
Meeting ID: 683 1048 2035