Self-Organizing Prompt Maps for Lightweight Prompt Tuning

• Samuel Henrique Silva Universidade de Pernambuco (UPE) https://orcid.org/0009-0002-0242-2832
• Allan Miller Silva Lima Universidade de Pernambuco (UPE) https://orcid.org/0009-0007-3838-8986
• Denis Mayr Lima Martins Universidade de São Paulo (USP) https://orcid.org/0000-0002-8262-2369
• Fernando Buarque de Lima Neto Universidade de Pernambuco (UPE) https://orcid.org/0000-0003-1200-225X

DOI: https://doi.org/10.5753/sbbd_estendido.2026.249354

Resumo

Apresentamos Self-Organizing Prompt Maps (SOPM), um sistema inteligente que une agrupamento topológico com capacidades generativas para permitir adaptação de prompts sem ajuste de parâmetros em Grandes Modelos de Linguagem (LLMs). Ao organizar prompts históricos em uma grade topológica 2D via Self-Organizing Map (SOM), o SOPM cria um espaço semântico contínuo onde a adaptação de prompts ocorre via busca por vizinhança, garantindo transparência espacial e validação humana. Resultados demonstram que integrar recuperação de prompts via SOM com arquiteturas generativas melhora acurácia em tarefas subsequentes. O SOPM oferece um caminho prático para ajuste de prompts de LLM interpretável e eficiente.

Palavras-chave: Self-Organizing Maps, Prompt Engineering, Large Language Models, In-Context Learning, Retrieval-Augmented Generation

Referências

Arora, S., Narayan, A., Chen, M. F., Orr, L., Guha, N., Bhatia, K., Chami, I., and Re, C. (2023). Ask me anything: A simple strategy for prompting language models. In The Eleventh International Conference on Learning Representations.

Han, Z., Gao, C., Liu, J., Zhang, J., and Zhang, S. Q. (2024). Parameter-efficient fine-tuning for large models: A comprehensive survey.

Hu, T. and Zhou, X.-H. (2024). Unveiling llm evaluation focused on metrics: Challenges and solutions.

Huijben, I. A., Nijdam, A. A., Overeem, S., Van Gilst, M. M., and Van Sloun, R. (2023). Som-cpc: Unsupervised contrastive learning with self-organizing maps for structured representations of high-rate time series. In International Conference on Machine Learning, pages 14132–14152. PMLR.

Kohonen, T. (1990). The self-organizing map. Proceedings of the IEEE, 78(9):1464–1480.

Kohonen, T., Schroeder, M. R., and Huang, T. S. (1996). Engineering applications of the self-organizing map. Proceedings of the IEEE, 84(10):1358–1384.

Li, W., Wang, X., Li, W., and Jin, B. (2025). A survey of automatic prompt engineering: An optimization perspective.

Manduchi, L., Hüser, M., Vogt, J. E., Rätsch, G., and Horn, M. (2019). Deep probabilistic clustering with self-organizing maps. arXiv preprint arXiv:1910.01590.

Mao, W., Wu, J., Chen, W., Gao, C., Wang, X., and He, X. (2025). Reinforced prompt personalization for recommendation with large language models. ACM Trans. Inf. Syst., 43(3).

Papineni, K., Roukos, S., Ward, T., and Zhu, W.-J. (2002). Bleu: a method for automatic evaluation of machine translation. In Proceedings of the 40th Annual Meeting on Association for Computational Linguistics, ACL ’02, page 311–318, USA. Association for Computational Linguistics.

Wang, W., Wei, F., Dong, L., Bao, H., Yang, N., and Zhou, M. (2020). Minilm: Deep self-attention distillation for task-agnostic compression of pre-trained transformers. Advances in neural information processing systems, 33:5776–5788.

Yang, A., Li, A., Yang, B., Zhang, B., Hui, B., Zheng, B., Yu, B., Gao, C., Huang, C., Lv, C., et al. (2025). Qwen3 technical report. arXiv preprint arXiv:2505.09388.

Yazan, M., Verberne, S., and Situmeang, F. (2025). Improving rag for personalization with author features and contrastive examples. In European Conference on Information Retrieval, pages 408–416. Springer.

Zhang, X., Zhao, J., and LeCun, Y. (2015). Character-level convolutional networks for text classification. In Cortes, C., Lawrence, N., Lee, D., Sugiyama, M., and Garnett, R., editors, Advances in Neural Information Processing Systems, volume 28. Curran Associates, Inc.