Overcoming Language Disparity in Online Content Classification with Multimodal Learning

TL;DR

This paper investigates how multimodal learning, incorporating images alongside text, can reduce language disparities in online content classification, especially improving performance on non-English languages compared to traditional language-only models.

Contribution

It demonstrates that multimodal learning with images helps bridge the performance gap between English and non-English languages in content detection tasks.

Findings

Multimodal models outperform text-only models on non-English languages.

Including images reduces the performance disparity between English and other languages.

Multimodal approaches improve detection accuracy across crisis, fake news, and emotion recognition tasks.

Abstract

Advances in Natural Language Processing (NLP) have revolutionized the way researchers and practitioners address crucial societal problems. Large language models are now the standard to develop state-of-the-art solutions for text detection and classification tasks. However, the development of advanced computational techniques and resources is disproportionately focused on the English language, sidelining a majority of the languages spoken globally. While existing research has developed better multilingual and monolingual language models to bridge this language disparity between English and non-English languages, we explore the promise of incorporating the information contained in images via multimodal machine learning. Our comparative analyses on three detection tasks focusing on crisis information, fake news, and emotion recognition, as well as five high-resource non-English languages, demonstrate that: (a) detection frameworks based on pre-trained large language models like BERT and multilingual-BERT systematically perform better on the English language compared against non-English languages, and (b) including images via multimodal learning bridges this performance gap. We situate our findings with respect to existing work on the pitfalls of large language models, and discuss their theoretical and practical implications. Resources for this paper are available at https://multimodality-language-disparity.github.io/.