Overcoming Language Disparity in Online Content Classification with Multimodal Learning

crown jewel figure
We use multimodal (image + text) learning to overcome the language disparity that exists between English and non-English languages. The figure illustrates an example of a social media post that is correctly classified in English but misclassified in Spanish. Including the corresponding image leads to correct classification in Spanish as well as other non-English languages.
Demo Video
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/.
Overcoming Language Disparity in Online Content Classification with Multimodal Learning
  title = {Overcoming {{Language Disparity}} in {{Online Content Classification}} with {{Multimodal Learning}}},
  booktitle = {Proceedings of the {{International AAAI Conference}} on {{Web}} and {{Social Media}}},
  author = {Verma, Gaurav and Mujumdar, Rohit and Wang, Zijie J. and De Choudhury, Munmun and Kumar, Srijan},
  year = {2022},
  month = may