Rethinking fake news detection for the real world

• Grant number: 22/05002-0
• Support Opportunities: Scholarships abroad - Research Internship - Doctorate
• Start date: September 04, 2022
• End date: July 31, 2023
• Field of knowledge: Physical Sciences and Mathematics - Computer Science
• Principal Investigator: Anderson de Rezende Rocha
• Grantee: Jing Yang
• Supervisor: Iryna Gurevych
• Host Institution: Instituto de Computação (IC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Institution abroad: Technische Universität Darmstadt (TU Darmstadt), Germany
• Associated to the scholarship: 19/04053-8 - Event reconstruction from heterogeneous visual data, BP.DR

Abstract

Due to the significant impact of fake news, many researchers are devoted to studying automated fact-checking. However, fact-checkers are concerned with machine-based black-box solutions that try to automate the entire fact-checking pipeline. Nonetheless, humans cannot do the job alone either, as the billions of news pieces and social media posts are generated every day. This project aims to bridge the gap between human fact-checkers and automated algorithms, so they can complement each other to achieve scalability and explainability. For scalability, we aim to generate and rank claims from massive disorganized social media data, exploring their relationship in graph settings. For explainability, we attempt to generate high quality questions based on claim syntactic structure and semantic meaning; then we perform question-based rationale alignment by extract precise evidence from documents using question answering. We will rely on a human-in-the-loop formulation for both qualitative evaluation and data annotation in both directions. The solutions developed in the project can provide fact-checkers useful tools to increase checking efficiency and more transparency for automated fact-checking. (AU)