ExU: AI Models for Examining Multilingual Disinformation Narratives and Understanding their Spread

TL;DR

ExU addresses multilingual disinformation by developing stance detection and claim retrieval capabilities across 20+ languages, with an emphasis on explainability for end users. The approach combines multilingual transformers and retrieval-augmented generation to support fact-checkers, leveraging the large-scale MultiClaim dataset (39 languages) and user-centric evaluation. A pilot survey indicates strong demand for accurate translations, automatic stance explanations, and concise summaries of fact-checks. The work aims to deliver scalable, explainable tools that improve cross-language disinformation analysis and support journalism and fact-checking workflows.

Abstract

Addressing online disinformation requires analysing narratives across languages to help fact-checkers and journalists sift through large amounts of data. The ExU project focuses on developing AI-based models for multilingual disinformation analysis, addressing the tasks of rumour stance classification and claim retrieval. We describe the ExU project proposal and summarise the results of a user requirements survey regarding the design of tools to support fact-checking.

Figures (1)

• Figure 1: Counts of languages from responses to the survey question "Which languages do you encounter most often in your work?". The "Other" category is comprised of Czech, Hindi, Polish, Portuguese, Russian, Sinhala, Slovak, and Turkish, all of which had a count of one.