Findings of Factify 2: Multimodal Fake News Detection

TL;DR

This paper reports on the Factify 2 shared task, which developed a multimodal dataset and approach for fake news detection using social media claims, supporting documents, text, and images, achieving an F1 score of 81.82%.

Contribution

It introduces a new multimodal dataset and benchmark for fake news detection, and evaluates various models, highlighting the effectiveness of DeBERTa, Swinv2, and CLIP.

Findings

Best model achieved an F1 score of 81.82%.

Multimodal approach improves fake news detection accuracy.

Over 60 participants contributed to the shared task.

Abstract

With social media usage growing exponentially in the past few years, fake news has also become extremely prevalent. The detrimental impact of fake news emphasizes the need for research focused on automating the detection of false information and verifying its accuracy. In this work, we present the outcome of the Factify 2 shared task, which provides a multi-modal fact verification and satire news dataset, as part of the DeFactify 2 workshop at AAAI'23. The data calls for a comparison based approach to the task by pairing social media claims with supporting documents, with both text and image, divided into 5 classes based on multi-modal relations. In the second iteration of this task we had over 60 participants and 9 final test-set submissions. The best performances came from the use of DeBERTa for text and Swinv2 and CLIP for image. The highest F1 score averaged for all five classes was 81.82%.