HyLEX-RAG: Bridging the Lexical-Semantic Gap in Regulatory Compliance Retrieval
Resumo
Regulatory information retrieval requires balancing semantic understanding with exact lexical precision. Pure semantic retrieval captures contextual meaning but suffers from normative ambiguity, while lexical approaches fail under paraphrases and synonym variations. This paper proposes HyLEX-RAG, a hybrid retrieval approach that combines hypothetical 5W1H question indexing with weighted lexical-semantic fusion for regulatory question answering. Experiments conducted on BACEN regulatory norms show that HyLEX-RAG achieved the best overall results, reaching 66.7% faithfulness, 70.0% answer rate, and 0% hallucination on out-of-scope questions, achieving better overall results than the evaluated baselines (p < 0.05). The results demonstrate that combining semantic and lexical retrieval enhances the reliability of RAG systems for compliance and auditing applications.Referências
Antoniou, G., Atkinson, K., Baryannis, G., Batsakis, S., Caro, L. D., Governatori, G., Robaldo, L., Siragusa, G., and Tachmazidis, I. (2022). Explainable reasoning with legal big data: A layered framework. Journal of Applied Logics — IfCoLog Journal of Logics and their Applications, 9(4):1155–1170.
Ashley, K. D. (2017). Artificial Intelligence and Legal Analytics: New Tools for Law Practice in the Digital Age. Cambridge University Press.
Devlin, J., Chang, M.-W., Lee, K., and Toutanova, K. (2019). Bert: Pre-training of deep bidirectional transformers for language understanding. Proceedings of NAACL-HLT, pages 4171–4186.
Es, S., James, J., Espinosa Anke, L., and Schockaert, S. (2024). RAGAs: Automated evaluation of retrieval augmented generation. In Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics: System Demonstrations, pages 150–158, Stroudsburg, PA, USA. Association for Computational Linguistics.
Gao, L., Ma, X., Lin, J., and Callan, J. (2023). Precise zero-shot dense retrieval without relevance labels. Findings of the Association for Computational Linguistics: ACL 2023, pages 1762–1777.
Karpukhin, V., Oguz, B., Min, S., Lewis, P., Wu, L., Edunov, S., Chen, D., and Yih, W.-t. (2020). Dense passage retrieval for open-domain question answering. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 6769–6781.
Kokina, J., Blanchette, S., Davenport, T. H., and Pachamanova, D. (2025). Challenges and opportunities for artificial intelligence in auditing: Evidence from the field. International Journal of Accounting Information Systems, 56:100734.
Lewis, P., Perez, E., Piktus, A., Petroni, F., Karpukhin, V., Goyal, N., K”uttler, H., Lewis, M., Yih, W.-t., Rockt”aschel, T., Riedel, S., and Kiela, D. (2020). Retrieval-augmented generation for knowledge-intensive nlp tasks. In Advances in Neural Information Processing Systems, volume 33, pages 9459–9474.
Lin, J., Ma, X., Lin, S.-C., Yang, J.-H., Pradeep, R., and Nogueira, R. (2021). Pyserini: A python toolkit for reproducible information retrieval research with sparse and dense representations. In Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 2356–2362.
Manning, C. D., Raghavan, P., and Sch”utze, H. (2008). Introduction to Information Retrieval. Cambridge University Press.
Petrov, E. and Stocken, P. C. (2022). Auditor specialization and information spillovers. The Accounting Review, 97(7):401–428.
Reimers, N. and Gurevych, I. (2019). Sentence-bert: Sentence embeddings using siamese bert-networks. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, pages 3982–3992.
Robertson, S. and Zaragoza, H. (2009). The probabilistic relevance framework: Bm25 and beyond. Foundations and Trends in Information Retrieval, 3(4):333–389.
Thakur, N., Reimers, N., R”uckl’e, A., Srivastava, A., and Gurevych, I. (2021). Beir: A heterogeneous benchmark for zero-shot evaluation of information retrieval models. In NeurIPS Datasets and Benchmarks Track.
Vake, D. et al. (2025). Bridging the question-answer gap in retrieval-augmented generation: Hypothetical prompt embeddings. IEEE Access.
Ashley, K. D. (2017). Artificial Intelligence and Legal Analytics: New Tools for Law Practice in the Digital Age. Cambridge University Press.
Devlin, J., Chang, M.-W., Lee, K., and Toutanova, K. (2019). Bert: Pre-training of deep bidirectional transformers for language understanding. Proceedings of NAACL-HLT, pages 4171–4186.
Es, S., James, J., Espinosa Anke, L., and Schockaert, S. (2024). RAGAs: Automated evaluation of retrieval augmented generation. In Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics: System Demonstrations, pages 150–158, Stroudsburg, PA, USA. Association for Computational Linguistics.
Gao, L., Ma, X., Lin, J., and Callan, J. (2023). Precise zero-shot dense retrieval without relevance labels. Findings of the Association for Computational Linguistics: ACL 2023, pages 1762–1777.
Karpukhin, V., Oguz, B., Min, S., Lewis, P., Wu, L., Edunov, S., Chen, D., and Yih, W.-t. (2020). Dense passage retrieval for open-domain question answering. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 6769–6781.
Kokina, J., Blanchette, S., Davenport, T. H., and Pachamanova, D. (2025). Challenges and opportunities for artificial intelligence in auditing: Evidence from the field. International Journal of Accounting Information Systems, 56:100734.
Lewis, P., Perez, E., Piktus, A., Petroni, F., Karpukhin, V., Goyal, N., K”uttler, H., Lewis, M., Yih, W.-t., Rockt”aschel, T., Riedel, S., and Kiela, D. (2020). Retrieval-augmented generation for knowledge-intensive nlp tasks. In Advances in Neural Information Processing Systems, volume 33, pages 9459–9474.
Lin, J., Ma, X., Lin, S.-C., Yang, J.-H., Pradeep, R., and Nogueira, R. (2021). Pyserini: A python toolkit for reproducible information retrieval research with sparse and dense representations. In Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 2356–2362.
Manning, C. D., Raghavan, P., and Sch”utze, H. (2008). Introduction to Information Retrieval. Cambridge University Press.
Petrov, E. and Stocken, P. C. (2022). Auditor specialization and information spillovers. The Accounting Review, 97(7):401–428.
Reimers, N. and Gurevych, I. (2019). Sentence-bert: Sentence embeddings using siamese bert-networks. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, pages 3982–3992.
Robertson, S. and Zaragoza, H. (2009). The probabilistic relevance framework: Bm25 and beyond. Foundations and Trends in Information Retrieval, 3(4):333–389.
Thakur, N., Reimers, N., R”uckl’e, A., Srivastava, A., and Gurevych, I. (2021). Beir: A heterogeneous benchmark for zero-shot evaluation of information retrieval models. In NeurIPS Datasets and Benchmarks Track.
Vake, D. et al. (2025). Bridging the question-answer gap in retrieval-augmented generation: Hypothetical prompt embeddings. IEEE Access.
Publicado
19/10/2026
Como Citar
CAMPOS, Márcia Rodrigues; PEDROSA, Glauco Vitor.
HyLEX-RAG: Bridging the Lexical-Semantic Gap in Regulatory Compliance Retrieval. In: ENCONTRO NACIONAL DE INTELIGÊNCIA ARTIFICIAL E COMPUTACIONAL (ENIAC), 23. , 2026, Cuiabá/MT.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2026
.
p. 587-598.
ISSN 2763-9061.
DOI: https://doi.org/10.5753/eniac.2026.30059.
