BOU-Guard: Uma Abordagem para Detecção de Conteúdo Impróprio na Internet
Abstract
Amidst the ever-expanding digital landscape, exposure to inappropriate content such as racism, homophobia, and sexism has become an increasingly pressing concern. Despite the existing literature on online hate speech, significant limitations persist, including a lack of automation and effective warning mechanisms. This article proposes an innovative approach, introducing the BOU-Guard (Behavior Observation Unit - Guard), based on GPT-3.5-Turbo technology, to detect and filter prejudiced or offensive content. Through a proof of concept, we demonstrated that the proposed mechanism applied to 30 web pages can detect offensive content with a high F1-Score, on average, to content related to homophobia (94.69%), racism (98.45%), and sexism (98.09%).
References
Chiu, K.-L., Collins, A., and Alexander, R. (2021). Detecting hate speech with gpt-3. arXiv preprint arXiv:2103.12407.
Gámez-Guadix, M. and Incera, D. (2021). Homophobia is online: Sexual victimization and risks on the internet and mental health among bisexual, homosexual, pansexual, asexual, and queer adolescents. Computers in human behavior, 119:106728.
Martins, M. V., Formiga, A., Santos, C., Sousa, D., Resende, C., Campos, R., Nogueira, N., Carvalho, P., and Ferreira, S. (2020). Adolescent internet addiction–role of parental control and adolescent behaviours. International Journal of Pediatrics and Adolescent Medicine, 7(3):116–120.
Sheth, A., Shalin, V. L., and Kursuncu, U. (2022). Defining and detecting toxicity on social media: context and knowledge are key. Neurocomputing, 490:312–318.
Wang, H., Hee, M. S., Awal, M. R., Choo, K. T. W., and Lee, R. K.-W. (2023). Evaluating gpt-3 generated explanations for hateful content moderation. arXiv preprint arXiv:2305.17680.
Yenala, H., Jhanwar, A., Chinnakotla, M. K., and Goyal, J. (2018). Deep learning for detecting inappropriate content in text. International Journal of Data Science and Analytics, 6:273–286.
