Detecção de Viés Ideológico em Artigos de Notícias Utilizando Aprendizagem Métrica Profunda e Representações Contextuais
Resumo
Este estudo investiga a detecção de viés ideológico em artigos de notícias por meio de aprendizagem métrica profunda com DistilRoBERTa. Na melhor configuração experimental, o modelo atingiu Macro F1-Score de 0,83 no conjunto ABP em random split. Sob o cenário mais rigoroso de media-bias split, o desempenho caiu para 0,43, evidenciando os desafios de generalização entre fontes. Na comparação com LLMs em regime zero-shot, a abordagem proposta superou GPT-3.5 e Llama-2 em Macro F1, indicando melhor equilíbrio entre classes para esta tarefa. Em síntese, os resultados sustentam a aprendizagem métrica como estratégia competitiva para detecção de viés ideológico, ao mesmo tempo em que expõem limitações relevantes de adaptação de domínio entre ecossistemas de notícias.Referências
Baly, R., Da San Martino, G., Glass, J., and Nakov, P. (2020). We can detect your bias: Predicting the political ideology of news articles. In Proc. of EMNLP, pages 4982–4991.
Brink, H., Richards, J., and Fetherolf, M. (2016). Real-World Machine Learning. Manning Publications.
Chen, W., Wachsmuth, H., Khatib, K., and Stein, B. (2018). Learning to flip the bias of news headlines. In Proc. of INLG, pages 79–88.
Chiang, C. and Knight, B. (2011). Media bias and influence: Evidence from newspaper endorsements. The Review of Economic Studies, 78:795–820.
Dallmann, A., Lemmerich, F., Zoller, D., and Hotho, A. (2015). Media bias in german online newspapers. In Proc. of HT, pages 133–137.
Devlin, J., Chang, M., Lee, K., and Toutanova, K. (2018). Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:1810.04805.
Efron, M. (2004). The liberal media and right-wing conspiracies. In Proc. of ACM CIKM, pages 390–398.
Gangula, R., Duggenpudi, S., and Mamidi, R. (2019). Detecting political bias in news articles using headline attention. In Proc. of ACL BlackboxNLP, pages 77–84.
Gentzkow, M. and Shapiro, J. M. (2006). Media bias and reputation. Journal of Political Economy, 114:280–316.
Goodfellow, I., Bengio, Y., and Courville, A. (2016). Deep Learning. MIT Press.
Kertész, G. (2021). Different triplet sampling techniques for lossless triplet loss on metric similarity learning. In Proc. of IEEE SAMI, pages 449–454.
Lin, L., Wang, L., Guo, J., and Wong, K. (2025). Investigating bias in llm-based bias detection: Disparities between llms and human perception. In Proc. of COLING, pages 10634–10649.
Lin, L., Wang, L., Zhao, X., Li, J., and Wong, K. (2024). Indivec: An exploration of leveraging llms for media bias detection. arXiv preprint arXiv:2402.00345.
Lin, Y., Bagrow, J., and Lazer, D. (2011). More voices than ever? quantifying media bias in networks. In Proc. of ICWSM, volume 5.
Liu, Y. et al. (2019). Roberta: A robustly optimized bert pretraining approach. arXiv preprint arXiv:1907.11692.
Liu, Y., Zhang, X., Wegsman, D., Beauchamp, N., and Wang, L. (2022). Politics: Pre-training with same-story article comparison for ideology prediction. In Findings of NAACL, pages 1354–1374.
Patricia Aires, V., Nakamura, F., and Nakamura, E. (2019). A link-based approach to detect media bias in news websites. In Companion Proc. of WWW, pages 742–745.
Spinde, T. et al. (2021). Neural media bias detection using distant supervision with babe. In Findings of EMNLP, pages 1166–1177.
Volf, M. and Simko, J. (2025). Political learning and politicalness classification of texts. arXiv preprint 2507.13913.
Zhang, Y. and Rao, Z. (2020). Deep neural networks with pre-train model bert for aspect-level sentiments classification. In Proc. of IEEE ITOEC, pages 923–927.
Brink, H., Richards, J., and Fetherolf, M. (2016). Real-World Machine Learning. Manning Publications.
Chen, W., Wachsmuth, H., Khatib, K., and Stein, B. (2018). Learning to flip the bias of news headlines. In Proc. of INLG, pages 79–88.
Chiang, C. and Knight, B. (2011). Media bias and influence: Evidence from newspaper endorsements. The Review of Economic Studies, 78:795–820.
Dallmann, A., Lemmerich, F., Zoller, D., and Hotho, A. (2015). Media bias in german online newspapers. In Proc. of HT, pages 133–137.
Devlin, J., Chang, M., Lee, K., and Toutanova, K. (2018). Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:1810.04805.
Efron, M. (2004). The liberal media and right-wing conspiracies. In Proc. of ACM CIKM, pages 390–398.
Gangula, R., Duggenpudi, S., and Mamidi, R. (2019). Detecting political bias in news articles using headline attention. In Proc. of ACL BlackboxNLP, pages 77–84.
Gentzkow, M. and Shapiro, J. M. (2006). Media bias and reputation. Journal of Political Economy, 114:280–316.
Goodfellow, I., Bengio, Y., and Courville, A. (2016). Deep Learning. MIT Press.
Kertész, G. (2021). Different triplet sampling techniques for lossless triplet loss on metric similarity learning. In Proc. of IEEE SAMI, pages 449–454.
Lin, L., Wang, L., Guo, J., and Wong, K. (2025). Investigating bias in llm-based bias detection: Disparities between llms and human perception. In Proc. of COLING, pages 10634–10649.
Lin, L., Wang, L., Zhao, X., Li, J., and Wong, K. (2024). Indivec: An exploration of leveraging llms for media bias detection. arXiv preprint arXiv:2402.00345.
Lin, Y., Bagrow, J., and Lazer, D. (2011). More voices than ever? quantifying media bias in networks. In Proc. of ICWSM, volume 5.
Liu, Y. et al. (2019). Roberta: A robustly optimized bert pretraining approach. arXiv preprint arXiv:1907.11692.
Liu, Y., Zhang, X., Wegsman, D., Beauchamp, N., and Wang, L. (2022). Politics: Pre-training with same-story article comparison for ideology prediction. In Findings of NAACL, pages 1354–1374.
Patricia Aires, V., Nakamura, F., and Nakamura, E. (2019). A link-based approach to detect media bias in news websites. In Companion Proc. of WWW, pages 742–745.
Spinde, T. et al. (2021). Neural media bias detection using distant supervision with babe. In Findings of EMNLP, pages 1166–1177.
Volf, M. and Simko, J. (2025). Political learning and politicalness classification of texts. arXiv preprint 2507.13913.
Zhang, Y. and Rao, Z. (2020). Deep neural networks with pre-train model bert for aspect-level sentiments classification. In Proc. of IEEE ITOEC, pages 923–927.
Publicado
19/07/2026
Como Citar
JANUÁRIO, Jailson Pereira; CARVALHO, André Luiz da Costa.
Detecção de Viés Ideológico em Artigos de Notícias Utilizando Aprendizagem Métrica Profunda e Representações Contextuais. In: BRAZILIAN WORKSHOP ON SOCIAL NETWORK ANALYSIS AND MINING (BRASNAM), 15. , 2026, Gramado/RS.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2026
.
p. 109-121.
ISSN 2595-6094.
DOI: https://doi.org/10.5753/brasnam.2026.21950.
