A Simple Voting Mechanism for Online Sexist Content Identification

TL;DR

This paper introduces a simple voting-based approach using multiple BERT models for detecting sexist content in social media, demonstrating improved performance and robustness across languages.

Contribution

The paper proposes a straightforward voting mechanism that enhances sexist content detection by combining multiple BERT models, showing its effectiveness in multilingual social media data.

Findings

Voting mechanism improves model accuracy

System is robust across languages and data sources

Achieved best results among tested models

Abstract

This paper presents the participation of the MiniTrue team in the EXIST 2021 Challenge on the sexism detection in social media task for English and Spanish. Our approach combines the language models with a simple voting mechanism for the sexist label prediction. For this, three BERT based models and a voting function are used. Experimental results show that our final model with the voting function has achieved the best results among our four models, which means that our voting mechanism brings an extra benefit to our system. Nevertheless, we also observe that our system is robust to data sources and languages.