A Transparent Multiclass Classification Architecture for Credit Variable Categorization

• André S. G. Moreira UFRN
• Caio V. A. Justino UFRN
• Igor W. C. Fernandes UFRN
• Marcos Murguel ICT Itaú
• Luana Oliveira ICT Itaú
• Anna M. C. Araujo Itaú-Unibanco S.A.
• Luana F. Alvarenga UFRN
• Renan C. Moioli UFRN
• Cesar Renno-Costa UFRN

DOI: https://doi.org/10.5753/eniac.2026.31383

Resumo

Integrating Alternative data in credit scoring risks revealing sensitive consumer traits. To prevent privacy violations, this study proposes an automated screening tool that categorizes textual variable descriptions into an eleven category taxonomy. To overcome labeled data scarcity, the methodology generates a synthetic dataset using the Gemma4 31B model. Because regulations demand transparency, the architecture relies exclusively on explainable machine learning models. Validation on empirical data showed high clustering cohesion, and Logistic Regression achieved 73.02% accuracy. This workflow guarantees the interpretability required for safe financial governance.

Referências

Alvarenga, L. F., Ferreira, N. F., Peres, A. S. C., Moioli, R. C., and Rennó-Costa, C. (2026). Legal and ethical considerations for translating federated learning into crossborder healthcare innovation. IEEE Journal of Biomedical and Health Informatics, (1).

Amazon Web Services (2026). Amazon macie documentation. AWS Official Documentation.

Anomah, S. (2026). Bridging the interpretability gap: integrating explainable ai into public sector audit systems in emerging economies. Accounting Research Journal, pages 1–27.

Becker, B. and Kohavi, R. (1996). Adult. UCI Machine Learning Repository. DOI: 10.24432/C5XW20.

Björkegren, D. and Grissen, D. (2018). The Potential of Digital Credit to Bank the Poor. AEA Papers and Proceedings, 108:68–71.

Brasil (2018). Lei nº 13.709, de 14 de agosto de 2018. lei geral de proteção de dados pessoais (lgpd). Accessed: Jun 21, 2026.

Brown, T., Mann, B., Ryder, N., Subbiah, M., Kaplan, J. D., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020). Language models are few-shot learners. Advances in neural information processing systems, 33:1877–1901.

Caliński, T. and Harabasz, J. (1974). A dendrite method for cluster analysis. Communications in Statistics-theory and Methods, 3(1):1–27.

Çallı, B. A. and Coşkun, E. (2021). A longitudinal systematic review of credit risk assessment and credit default predictors. Sage Open, 11(4).

Davies, D. L. and Bouldin, D. W. (1979). A cluster separation measure. IEEE transactions on pattern analysis and machine intelligence, (2):224–227.

European Union (2016). Regulation (eu) 2016/679 of the european parliament and of the council of 27 april 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data (general data protection regulation). Official Journal of the European Union.

Fabian, P. (2011). Scikit-learn: Machine learning in python. Journal of machine learning research 12, page 2825.

Federal Financial Institutions Examination Council (2025). Home Mortgage Disclosure Act (HMDA) Data. FFIEC / CFPB.

Google (2026). Cloud data loss prevention documentation. Google Cloud Official Documentation.

Guo, G., Zhu, F., Chen, E., Liu, Q., Wu, L., and Guan, C. (2016). From Footprint to Evidence: An Exploratory Study of Mining Social Data for Credit Scoring. ACM Trans. Web, 10(4):22:1–22:38.

Hermans, A., Beyer, L., and Leibe, B. (2017). In defense of the triplet loss for person re-identification. arXiv preprint arXiv:1703.07737.

Hurley, M. and Adebayo, J. (2017). Credit scoring in the era of big data. Yale Journal of Law and Technology.

Kurshan, E., Shen, H., and Chen, J. (2021). Towards self-regulating ai: challenges and opportunities of ai model governance in financial services. In Proceedings of the First ACM International Conference on AI in Finance, ICAIF ’20, New York, NY, USA. Association for Computing Machinery.

McInnes, L., Healy, J., and Melville, J. (2020). UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction. arXiv:1802.03426 [stat.ML].

Microsoft (2026). Microsoft presidio: Data protection and anonymization sdk. Microsoft GitHub Repository.

Mwesigwa, D., J. Jackson, S., and Csikszentmihalyi, C. (2026a). Risk, data, alignment: Making credit scoring work in kenya. In Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems, CHI ’26, New York, NY, USA. Association for Computing Machinery.

Mwesigwa, D., Jackson, S. J., and Csikszentmihalyi, C. (2026b). Risk, Data, Alignment: Making Credit Scoring Work in Kenya. Version Number: 1.

Prosper Marketplace (2014). Prosper Loan Data. Kaggle.

Ribeiro, M., Malcorra, B., Mota, N. B., Wilkens, R., Villavicencio, A., Hubner, L. C., and Rennó-Costa, C. (2024). A methodology for explainable large language models with integrated gradients and linguistic analysis in text classification. arXiv, (arXiv:2410.00250).

Rousseeuw, P. J. (1987). Silhouettes: a graphical aid to the interpretation and validation of cluster analysis. Journal of computational and applied mathematics, 20:53–65.

Salton, G., Wong, A., and Yang, C.-S. (1975). A vector space model for automatic indexing. Communications of the ACM, 18(11):613–620.

Schwarcz, D. and Prince, A. E. R. Proxy Discrimination in the Age of Artificial Intelligence and Big Data.

Team, G., Mesnard, T., Hardin, C., Dadashi, R., Bhupatiraju, S., Pathak, S., Sifre, L., Rivière, M., Kale, M. S., Love, J., et al. (2024). Gemma: Open models based on gemini research and technology. arXiv preprint arXiv:2403.08295.

Tounsi, Y., Hassouni, L., and Anoun, H. (2017). Credit scoring in the age of Big Data – A State-of-the-Art. 15(7).

Xiao, S., Liu, Z., Zhang, P., Muennighoff, N., Lian, D., and Nie, J.-Y. (2024). C-pack: Packed resources for general chinese embeddings. In Proceedings of the 47th international ACM SIGIR conference on research and development in information retrieval, pages 641–649.

Yang, A., Yang, B., Zhang, B., Hui, B., Zheng, B., Yu, B., Li, C., Liu, D., Huang, F., Dong, G., et al. (2024). Qwen2.5 technical report. arXiv preprint arXiv:2412.15115.