As you may know, DisAI is organizing a shared task for the SemEval 2025 workshop this year. The task, titled “Task 7: Multilingual and Crosslingual Fact-Checked Claim Retrieval”, focuses on a core challenge faced by fact-checkers today: given a claim, how can one efficiently determine whether it has already been fact-checked, keeping in mind that the fact-check might be in a different language and published on any of a number of different fact-checking platforms.
A special, updated version of the MultiClaim dataset was prepared for the task and two different tracks were designed to score approaches in both a monolingual and a cross-lingual retrieval setting. The task attracted a number of participants – a total of 179 researchers across 31 teams submitted 52 runs and 22 system papers, pushing the boundaries of multilingual fact-checked claim retrieval.
The results of the shared task are going to be presented 31st July – 1st August in Vienna, where the SemEval 2025 workshop is co-located with ACL 2025. We hope to meet some of you there!