Supporting the Curation of Educational Texts: Using Computational Methods for Violent Content Screening

• Lara da Silva Dias Universidade Federal de Juiz de Fora (UFJF)
• Lucas Larcher Universidade Federal de Juiz de Fora (UFJF)
• João Augusto Pilato de Castro Universidade Federal de Juiz de Fora (UFJF)
• João Vítor de Castro Martins F. Nogueira Universidade Federal de Juiz de Fora (UFJF)
• Jairo Francisco de Souza Universidade Federal de Juiz de Fora (UFJF)

DOI: https://doi.org/10.5753/sbie.2026.27240

Resumo

Curating texts for educational use is costly but essential for producing learning materials, item tests, and other resources. In addition to aligning with a skill intended to be taught or assessed, the material must be non-harmful to students, especially free of violent content, inappropriate language, or any other form of hate speech. This study analyzes computational methods for automatically detecting and grading violence in Portuguese children's texts. We introduce the Gru, a corpus annotated by experts based on a taxonomy derived from the Brazilian Indicative Rating System. Results reveal that traditional models miss implicit violence, whereas LLMs excel in contextual reasoning.

Palavras-chave: Content Moderation, Large Language Models, Educational Text Curation

Referências

Barbosa, G., Batista, H., Miranda, P., Santos, J., Isotani, S., Cordeiro, T., Bittencourt, I. I., and Mello, R. F. (2022). Aprendizagem de máquina para classificação de tipos textuais: Estudo de caso em textos escritos em português brasileiro. In Anais do XXXIII Simpósio Brasileiro de Informática na Educação, pages 920–931, Porto Alegre, RS, Brasil. SBC.

Battiston, T. (2025). Classificação de manchetes de cunho violento por meio de mineração de texto. Trabalho de Conclusão de Curso (Graduação em Engenharia de Controle e Automação) – Universidade Federal de Santa Catarina, Campus Blumenau. Acesso em: 22 abr. 2026.

Brennan, R. L. (2023). Educational measurement. Bloomsbury Publishing PLC.

Burnap, P. and Williams, M. (2015). Cyber hate speech on twitter: An application of machine classification and statistical modeling for policy and decision making: Machine classification of cyber hate speech. Policy & Internet, 7.

Chico, V., Regino, A., Bonacin, R., and Reis, J. (2025). Llm-based automatic generation of multiple-choice questions with meaningful distractors. In Anais do XXXVI Simpósio Brasileiro de Informática na Educação, pages 814–828, Porto Alegre, RS, Brasil. SBC.

Clarke, M. and Luna-Bazaldua, D. (2021). Primer on Large-Scale Assessments of Educational Achievement. World Bank Publications, Washington, DC.

da Justiça e Segurança Pública, B. M. (2018). Guia prático de classificação indicativa. Acesso em: 22 abr. 2026.

da Justiça e Segurança Pública, B. M. (s.d.). Descritores de conteúdos. Documento institucional, Ministério da Justiça e Segurança Pública, Brasília.

da Silva, F. R. and Ramos, E. M. L. S. (2023). Diagrama e taxonomia da tipologia das violências. Relatório técnico, Programa de Pós-Graduação em Segurança Pública, Instituto de Filosofia e Ciências Humanas, Universidade Federal do Pará. PPGSP/IFCH/UFPA.

de Oliveira Trajano, D. (2023). Detecção de linguagem tóxica aplicada a textos em português. Dissertação de mestrado, Pontifícia Universidade Católica do Rio Grande do Sul (PUCRS), Porto Alegre, Brasil. Programa de Pós-Graduação em Ciência da Computação.

de Pelle, R. and Moreira, V. (2017). Offensive comments in the brazilian web: a dataset and baseline results. In Anais do VI Brazilian Workshop on Social Network Analysis and Mining, pages 510–519, Porto Alegre, RS, Brasil. SBC.

Devlin, J., Chang, M.-W., Lee, K., and Toutanova, K. (2019). BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), pages 4171–4186, Minneapolis, Minnesota. Association for Computational Linguistics.

dos Santos Rodrigues, R. C. (2023). Evidências de validade de escala para rastreio da violência contra a pessoa idosa. Master's thesis, Universidade Federal da Paraíba (UFPB).

Educational Testing Service (2022). Ets guidelines for fair tests and communications. Technical report, Educational Testing Service (ETS), Princeton, NJ.

Fortuna, P. and Nunes, S. (2018). A survey on automatic detection of hate speech in text. ACM Comput. Surv., 51(4).

Furniturewala, S., Jandial, S., Java, A., Banerjee, P., Shahid, S., Bhatia, S., and Jaidka, K. (2024). Thinking fair and slow: On the efficacy of structured prompts for debiasing language models.

Guide, B. F. (2022). Detecção automática de discurso de ódio punitivista em redes sociais. Tese de doutorado, Universidade de São Paulo, São Paulo, Brasil. Faculdade de Filosofia, Letras e Ciências Humanas.

Haladyna, T. and Rodriguez, M. (2013). Developing and Validating Test Items. Routledge, Abingdon, UK.

Hee, M. S., Sharma, S., Cao, R., Nandi, P., Nakov, P., Chakraborty, T., and Lee, R. K.-W. (2024). Recent advances in online hate speech moderation: Multimodality and the role of large models. In Al-Onaizan, Y., Bansal, M., and Chen, Y.-N., editors, Findings of the Association for Computational Linguistics: EMNLP 2024, pages 4407–4419, Miami, Florida, USA. Association for Computational Linguistics.

Heru, S., Hidayah, I., and Kusumawardani, S. (2022). Automatic item generation with reading passages: A systematic literature review. In 8th International Conference on Education and Technology (ICET), pages 250–255.

Krug, E. G., Dahlberg, L. L., Mercy, J. A., Zwi, A. B., and Lozano, R., editors (2002). World Report on Violence and Health. World Health Organization, Geneva.

Laitz, T., Almeida, T. S., Abonizio, H., Junior, R. M., Bonás, G. K., Piau, M., Larcher, C., Pires, R., and Nogueira, R. (2026). Sabiá-4 technical report.

Leite, J. A., Silva, D., Bontcheva, K., and Scarton, C. (2020). Toxic language detection in social media for Brazilian Portuguese: New dataset and multilingual analysis. In Kam-Fai Wong, K. K. and Wu, H., editors, Proceedings of the 1st Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 10th International Joint Conference on Natural Language Processing, pages 914–924, Suzhou, China. Association for Computational Linguistics.

Liu, B. (2012). Sentiment Analysis and Opinion Mining, volume 5 of Synthesis Lectures on Human Language Technologies. Morgan & Claypool Publishers, San Rafael, CA.

Mantini, L. A., de Arajo da Silva, Y. P., and Meireles, M. R. G. (2021). Reconhecimento automático de narrativas violentas contra a mulher utilizando classificadores neurais. Pista: Periódico Interdisciplinar [Sociedade Tecnologia Ambiente], 3(2). Seção Discente – Temática Livre.

Nogueira, J., Castro, J., Larcher, L., Ferreira, R., Barbosa, B., and Souza, J. (2025). Uma análise de qualidade do uso de grandes modelos de linguagem para geração automática de itens avaliativos em português. In Anais do XXXVI Simpósio Brasileiro de Informática na Educação, pages 235–248, Porto Alegre, RS, Brasil. SBC.

Olweus, D. (1993). Bullying at school: What we know and what we can do. Wiley-Blackwell, USA.

Palla, K., García, J. L. R., Hauff, C., Fabbri, F., Damianou, A., Lindström, H., Taber, D., and Lalmas, M. (2025). Policy-as-prompt: Rethinking content moderation in the age of large language models. In Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency, FAccT '25, page 840–854, New York, NY, USA. Association for Computing Machinery.

Razavi, A., Inkpen, D., Uritsky, S., and Matwin, S. (2010). Offensive language detection using multi-level classification. In Proceedings of the 23rd Canadian Conference on Artificial Intelligence, pages 16–27.

Schmidt, A. and Wiegand, M. (2017). A survey on hate speech detection using natural language processing. In Ku, L.-W. and Li, C.-T., editors, Proceedings of the Fifth International Workshop on Natural Language Processing for Social Media, pages 1–10, Valencia, Spain. Association for Computational Linguistics.

Spertus, E. (1997). Smokey: automatic recognition of hostile messages. In Proceedings of the Fourteenth National Conference on Artificial Intelligence and Ninth Conference on Innovative Applications of Artificial Intelligence, AAAI'97/IAAI'97, page 1058–1065. AAAI Press.

Trajano, Douglas, Bordini, H., R., and Vieira, R. (2024). OLID-BR: offensive language identification dataset for brazilian portuguese. Language Resources and Evaluation, 58(4):1263–1289.

Vargas, F., Carvalho, I., de Góes, F. R., Pardo, T., and Benevenuto, F. (2022). HateBR: A large expert annotated corpus of Brazilian Instagram comments for offensive language and hate speech detection. In Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 7174–7183, Marseille, France. European Language Resources Association.

Zampieri, M., Malmasi, S., Nakov, P., Rosenthal, S., Farra, N., and Kumar, R. (2019). Predicting the type and target of offensive posts in social media. In Doran, J. B. C. and Solorio, T., editors, Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), pages 1415–1420, Minneapolis, Minnesota. Association for Computational Linguistics.