Cross-lingual Transfer Learning for Check-worthy Claim Identification over Twitter

TL;DR

This paper explores cross-lingual transfer learning using Multilingual BERT to identify check-worthy claims on Twitter across multiple languages, demonstrating that zero-shot transfer can be effective even with limited target language data.

Contribution

It systematically compares six cross-lingual approaches for claim verification in social media, showing zero-shot transfer can match or surpass monolingual models in some cases.

Findings

Zero-shot transfer performs comparably to monolingual models.

Some language pairs benefit significantly from cross-lingual transfer.

Cross-lingual methods outperform or match state-of-the-art models in certain languages.

Abstract

Misinformation spread over social media has become an undeniable infodemic. However, not all spreading claims are made equal. If propagated, some claims can be destructive, not only on the individual level, but to organizations and even countries. Detecting claims that should be prioritized for fact-checking is considered the first step to fight against spread of fake news. With training data limited to a handful of languages, developing supervised models to tackle the problem over lower-resource languages is currently infeasible. Therefore, our work aims to investigate whether we can use existing datasets to train models for predicting worthiness of verification of claims in tweets in other languages. We present a systematic comparative study of six approaches for cross-lingual check-worthiness estimation across pairs of five diverse languages with the help of Multilingual BERT (mBERT) model. We run our experiments using a state-of-the-art multilingual Twitter dataset. Our results show that for some language pairs, zero-shot cross-lingual transfer is possible and can perform as good as monolingual models that are trained on the target language. We also show that in some languages, this approach outperforms (or at least is comparable to) state-of-the-art models.