Fake News Detection: generating training examples for NLI and TNLI models -- JF Bussotti, P Papotti

Fake News Detection: generating training examples for NLI and TNLI models JF Bussotti, P Papotti
Wimmics Seminar - 16/11/2023 ‪@wimmicsinria3092‬

Abstract: Computational fact-checking relies on supervised models to verify claims based on given evidence, requiring a resource-intensive process to annotate large volumes of training data. We introduce Unown, a novel framework that generates training instances for fact-checking systems automatically using both textual and tabular content. Unown selects relevant evidence and generates supporting and refuting claims with advanced negation artifacts. Designed to be flexible, Unown accommodates various strategies for evidence selection and claim generation, offering unparalleled adaptability. We comprehensively evaluate Unown on both text-only and table+text benchmarks, including Feverous, SciFact, and MMFC, a new multi-modal fact-checking dataset. Our results demonstrate that Unown examples are of comparable quality to expert-labeled data, even enabling models to achieve up to 5% higher accuracy.

The code, data, and models are available at https://github.com/JeFlBu/unown.git.