Recuperando Perguntas em vez de Trechos: Uma Abordagem Question-Augmented RAG para Perguntas sobre Medicamentos

• Claudio M.V. de Andrade UFMG
• Raiane Asevedo UFMG
• Isabella Schettini UFMG
• Zilma Reis UFMG
• Marcos A. Gonçalves UFMG

DOI: https://doi.org/10.5753/webmedia.2026.30968

Resumo

Large Language Models (LLMs) have been increasingly employed for healthcare question answering, particularly in applications requiring access to specialized knowledge. This paper investigates the impact of replacing traditional chunkbased indexing with synthetic question indexing on retrieval quality in Retrieval-Augmented Generation (RAG). Using Portuguese drug package inserts from the Brazilian Health Regulatory Agency (Anvisa), we compare three settings: zeroshot inference, conventional RAG, and Question-Augmented Retrieval-Augmented Generation (QA-RAG), in which retrieval is performed over synthetic questions generated from document chunks. The evaluation comprises 90 automatically generated questions covering the 30 most popular medications in Brazil, with responses assessed by both healthcare specialists and an LLM-as-a-Judge. Results show that context retrieval substantially improves answer quality over zeroshot inference. Moreover, QA-RAG consistently outperforms conventional RAG, achieving a 44% improvement in the human evaluation while producing higher semantic retrieval scores. Although question-based indexing introduces additional computational cost, the results indicate that it provides a favorable trade-off between retrieval effectiveness and computational overhead for pharmaceutical question answering in Portuguese.

Palavras-chave: Retrieval-Augmented Generation, Question Answering, Pharmaceutical Information, LLM-as-a-Judge

Referências

Claudio Moisés Valiense De Andrade, Washington Cunha, Guilherme Fonseca, Ana Clara Souza Pagano, Luana De Castro Santos, Adriana Silvina Pagano, Leonardo Chaves Dutra Da Rocha, and Marcos André Gonçalves. 2024. Explaining the Hardest Errors of Contextual Embedding Based Classifiers. In Proceedings of the 28th Conference on Computational Natural Language Learning, Libby Barak and Malihe Alikhani (Eds.). Association for Computational Linguistics, Miami, FL, USA, 419–434. DOI: 10.18653/v1/2024.conll-1.31

Washington Cunha, Celso França, Guilherme Fonseca, Leonardo Rocha, and Marcos André Gonçalves. 2023. An Effective, Efficient, and Scalable Confidence-Based Instance Selection Framework for Transformer-Based Text Classification. In Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval. ACM, 665–674. DOI: 10.1145/3539618.3591638

Washington Cunha, Felipe Viegas, Celso França, Guilherme Fonseca, Leonardo Rocha, and Marcos André Gonçalves. 2025. A Noise-Oriented and Redundancy-Aware Instance Selection Framework. ACM Transactions on Information Systems 43, 1 (2025). DOI: 10.1145/3701769

Claudio M.V. de Andrade, Fabiano M. Belém, Washington Cunha, Celso França, Felipe Viegas, Leonardo Rocha, and Marcos André Gonçalves. 2023. On the class separability of contextual embeddings representations – or “The classifier does not matter when the (text) representation is so good!”. Information Processing Management 60, 4 (2023), 103336. DOI: 10.1016/j.ipm.2023.103336

Claudio M. V. de Andrade, Fabiano Muniz Belém, Celso França, Marcos Carvalho, Marcelo Ganem, Gabriel Teixeira, Gabriel Jallais, Alberto H. F. Laender, and Marcos A. Gonçalves. 2025. PromptNER: An Automatic Prompt-Learning Data Labeling Approach for Named Entity Recognition on Sensitive Data. Journal of Information and Data Management 16, 1 (Jan. 2025), 62–71. DOI: 10.5753/jidm.2025.4298

Matthijs Douze, Alexandr Guzhva, Chengqi Deng, Jeff Johnson, Gergely Szilvasy, Pierre-Emmanuel Mazaré, Maria Lomeli, Lucas Hosseini, and Hervé Jégou. 2024. The Faiss library. (2024). arXiv:2401.08281 [cs.LG]

David Fernandes, Edleno Silva de Moura, Berthier A. Ribeiro-Neto, Altigran Soares da Silva, and Marcos André Gonçalves. 2007. Computing Block Importance for Searching on Web Sites. In Proceedings of the Sixteenth ACM Conference on Information and Knowledge Management. ACM, 165–174. DOI: 10.1145/1321440.1321466

Luyu Gao, Xueguang Ma, Jimmy Lin, and Jamie Callan. 2023. Precise Zero-Shot Dense Retrieval without Relevance Labels. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Anna Rogers, Jordan Boyd-Graber, and Naoaki Okazaki (Eds.). Association for Computational Linguistics, Toronto, Canada, 1762–1777. DOI: 10.18653/v1/2023.acl-long.99

Yunfan Gao et al. 2023. Retrieval-Augmented Generation for Large Language Models: A Survey. arXiv preprint arXiv:2312.10997 (2023). [link]

Seokgi Lee. 2025. Transforming Questions and Documents for Semantically Aligned Retrieval-Augmented Generation. arXiv:2508.09755 [cs.CL]

Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, and Douwe Kiela. 2020. Retrieval-augmented generation for knowledge-intensive NLP tasks. In Proceedings of the 34th International Conference on Neural Information Processing Systems. Curran Associates.

Nikolaos Livathinos, Christoph Auer, Maksym Lysak, Ahmed Nassar, Michele Dolfi, Panos Vagenas, Cesar Berrospi Ramis, Matteo Omenetti, Kasper Dinkla, Yusik Kim, Shubham Gupta, Rafael Teixeira de Lima, Valery Weber, Lucas Morin, Ingmar Meijer, Viktor Kuropiatnyk, and Peter W. J. Staar. 2025. Docling: An Efficient Open-Source Toolkit for AI-driven Document Conversion. arXiv:2501.17887 [cs.CL]

Vatsal Raina and Mark Gales. 2024. Question-Based Retrieval using Atomic Units for Enterprise RAG. In Proceedings of the Seventh Fact Extraction and VERification Workshop (FEVER), Michael Schlichtkrull, Yulong Chen, Chenxi Whitehouse, Zhenyun Deng, Mubashara Akhtar, Rami Aly, Zhijiang Guo, Christos Christodoulopoulos, Oana Cocarascu, Arpit Mittal, James Thorne, and Andreas Vlachos (Eds.). Association for Computational Linguistics, Miami, Florida, USA, 219–233. [link]

Stephen Edward Robertson, Steve Walker, Susan Jones, Micheline M Hancock-Beaulieu, Mike Gatford, et al. 1994. Okapi at TREC. (1994).

Arun James Thirunavukarasu, Darren Shu Jeng Ting, Kabilan Elangovan, Laura Gutierrez, Ting Fang Ting, and Daniel Shu Wei Ting. 2023. Large language models in medicine. Nature medicine 29, 8 (2023), 1930–1940.

Felipe Viegas, Sergio Canuto, Washington Cunha, Celso França, Claudio Valiense, Leonardo Rocha, and Marcos Gonçalves. 2023. CluSent – Combining Semantic Expansion and De-Noising for Dataset-Oriented Sentiment Analysis of Short Texts. In Proceedings of the 29th Brazilian Symposium on Multimedia and the Web (Ribeirão Preto/SP). SBC, Porto Alegre, RS, Brasil.

Felipe Viegas, Sérgio Canuto, Christian Gomes, Washington Cunha, Thierson Rosa, Sabir Ribas, Leonardo Rocha, and Marcos André Gonçalves. 2019. CluWords: Exploiting Semantic Word Clustering Representation for Enhanced Topic Modeling. In Proceedings of the Twelfth ACM International Conference on Web Search and Data Mining. ACM, 753–761. DOI: 10.1145/3289600.3291032

Felipe Viegas, Washington Cunha, Christian Gomes, Antônio Pereira, Leonardo Rocha, and Marcos André Gonçalves. 2020. CluHTM: Semantic Hierarchical Topic Modeling Based on CluWords. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. Association for Computational Linguistics, 8138–8150.

Felipe Viegas, Washington Cunha, Christian Gomes, Leonardo Rocha, and Marcos André Gonçalves. 2025. Exploiting Contextual Embeddings in Hierarchical Topic Modeling and Investigating the Limits of the Current Evaluation Metrics. Computational Linguistics (2025).

Liang Wang, Nan Yang, and Furu Wei. 2023. Query2doc: Query Expansion with Large Language Models. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, Houda Bouamor, Juan Pino, and Kalika Bali (Eds.). Association for Computational Linguistics, Singapore, 9414–9423. DOI: 10.18653/v1/2023.emnlp-main.585

Cyril Zakka, Akash Chaurasia, Rohan Shad, Alex R Dalal, Jennifer L Kim, Michael Moor, Kevin Eng, Christopher Fong, William Phillips, Michelle M Mello, et al. 2024. Almanac: retrievalaugmented language models for clinical medicine. NEJM AI 1, 2 (2024), AIoa2300068.

Lianmin Zheng, Wei-Lin Chiang, Ying Sheng, Siyuan Hao, Zhanghao Wu, Sen Ba, Zi Lin Jiang, Hao Wu, Yonghao Zhuang, Zi Lin, et al. 2024. Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena. In Advances in Neural Information Processing Systems, Vol. 36.