An Approach for the Automatic Detection of Prejudice in Instant Messaging Applications

Melissa Sousa, Fernanda Nascimento, Gustavo Martins, José Maria Monteiro, Javam Machado

2025

Abstract

Instant messaging applications have revolutionized communication, making it more accessible and efficient. However, they have also facilitated the widespread dissemination of prejudiced media content. In this context, the rapid and effective detection of prejudice in texts shared via messaging apps is crucial for promoting a healthy, diverse, and tolerant communicative environment. Few prejudice detection methods have been specifically developed for instant messaging platforms. Moreover, the development of effective methods requires labeled datasets containing prejudiced messages disseminated on these platforms, as user expressions differ significantly from those on other social networks like Facebook, Instagram, and X. However, we have not found any datasets containing prejudiced messages extracted from WhatsApp or Telegram. This work presents two publicly available labeled datasets, named PrejudiceWhatsApp.Br and PrejudiceTelegram.Br, consisting of Brazilian Portuguese (PT-BR) messages collected from public groups on WhatsApp and Telegram, respectively. Additionally, we developed a dictionary of prejudiced words for Brazilian Portuguese, named PrejudicePT-br, comprising 842 words organized into nine categories. Finally, we built a dictionary-based machine learning model to automatically detect prejudice in WhatsApp and Telegram messages. We conducted a series of text classification experiments, combining two feature extraction methods, three distinct token generation strategies, two preprocessing approaches, and nine classification algorithms to classify texts into two categories: prejudiced and non-prejudiced. Our best results achieved an F1-score of 0.86 for both datasets, demonstrating the feasibility of the proposed approach.

Paper Citation

in Harvard Style

Sousa M., Nascimento F., Martins G., Monteiro J. and Machado J. (2025). An Approach for the Automatic Detection of Prejudice in Instant Messaging Applications. In Proceedings of the 14th International Conference on Data Science, Technology and Applications - Volume 1: DATA; ISBN 978-989-758-758-0, SciTePress, pages 108-119. DOI: 10.5220/0013555100003967

in Bibtex Style

@conference{data25,
author={Melissa Sousa and Fernanda Nascimento and Gustavo Martins and José Monteiro and Javam Machado},
title={An Approach for the Automatic Detection of Prejudice in Instant Messaging Applications},
booktitle={Proceedings of the 14th International Conference on Data Science, Technology and Applications - Volume 1: DATA},
year={2025},
pages={108-119},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013555100003967},
isbn={978-989-758-758-0},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 14th International Conference on Data Science, Technology and Applications - Volume 1: DATA
TI - An Approach for the Automatic Detection of Prejudice in Instant Messaging Applications
SN - 978-989-758-758-0
AU - Sousa M.
AU - Nascimento F.
AU - Martins G.
AU - Monteiro J.
AU - Machado J.
PY - 2025
SP - 108
EP - 119
DO - 10.5220/0013555100003967
PB - SciTePress