Metadata-Driven Construction of Semantic Views in Enterprise Knowledge Graphs with LLM Agents

• Tulio Vidal Rolim Universidade Federal do Ceará (UFC) / Instituto Federal de Educação, Ciência e Tecnologia Piauí (IFPI)
• José Renato S. Freitas Universidade Federal do Ceará / Instituto Federal de Educação, Ciência e Tecnologia do Ceará
• Vânia Maria Ponte Vidal Universidade Federal do Ceará

DOI: https://doi.org/10.5753/sbbd_estendido.2025.248167

Resumo

An Enterprise Knowledge Graph (EKG) provides a robust foundation for knowledge management, data integration, and advanced analytics within organizations by offering a semantic view that unifies and semantically integrates heterogeneous data sources from the data lake. The data integration process remains complex and time-consuming due to schema mismatches, divergent terminologies, and inconsistencies in data collection practices. Recent advances in large language models (LLMs) have shown promise in addressing these challenges by assisting with various data integration tasks. This paper introduces a modular, agent-oriented architecture that supports the incremental and interactive construction of semantic views for EKGs. The architecture leverages LLM-powered agents in conjunction with a metadata graph that captures rich contextual information about each semantic view. This metadata graph plays a central role in enabling automation, enhancing explainability, and ensuring reusability throughout the construction process. By forming agent decisions in structured and trustworthy metadata, the proposed framework facilitates the development of semantic views of enterprise knowledge graphs.

Palavras-chave: Metadata, Semantic View, Agents, Architecture, Enterprise Knowledge Graph

Referências

Feuer, B., Liu, Y., Hegde, C., and Freire, J. (2023). Archetype: A novel framework for open-source column type annotation using large language models. arXiv preprint arXiv:2310.18208.

Galkin, M., Auer, S., Vidal, M.-E., and Scerri, S. (2017). Enterprise knowledge graphs: A semantic approach for knowledge management in the next generation of enterprise information systems. In International Conference on Enterprise Information Systems, volume 2, pages 88–98. SCITEPRESS.

He, J., Treude, C., and Lo, D. (2024). Llm-based multi-agent systems for software engineering: Literature review, vision and the road ahead. Proceedings of the ACM (forthcoming). Preprint available via arXiv.

Kayali, M., Lykov, A., Fountalis, I., Vasiloglou, N., Olteanu, D., and Suciu, D. (2023). Chorus: foundation models for unified data discovery and exploration. arXiv preprint arXiv:2306.09610.

Laurenzi, E., Mathys, A., and Martin, A. (2024). An llm-aided enterprise knowledge graph (ekg) engineering process. In AAAI Spring Symposium Series (SSS-24). Association for the Advancement of Artificial Intelligence.

Liu, Y., Pena, E., Santos, A., Wu, E., and Freire, J. (2024). Magneto: Combining small and large language models for schema matching. arXiv preprint arXiv:2412.08194.

Nuzzolese, A. G. (2025). Streamlining knowledge graph creation with pyrml. arXiv preprint arXiv:2505.20949.

Santos, A., Pena, E. H., Lopez, R., and Freire, J. (2025). Interactive data harmonization with llm agents. arXiv preprint arXiv:2502.07132.

Tu, J., Fan, J., Tang, N., Wang, P., Li, G., Du, X., Jia, X., and Gao, S. (2023). Unicorn: A unified multi-tasking model for supporting matching tasks in data integration. Proceedings of the ACM on Management of Data, 1(1):1–26.

Vidal, V., Freitas, R., Arruda, N., Casanova, M. A., and Renso, C. (2024). A data design pattern for building and exploring semantic views of enterprise knowledge graphs. In Anais do XXXIX Simpósio Brasileiro de Bancos de Dados, pages 1–13, Porto Alegre, RS, Brasil. SBC.

Xiao, G., Lanti, D., Kontchakov, R., Komla-Ebri, S., Güzel-Kalaycı, E., Ding, L., Corman, J., Cogrel, B., Calvanese, D., and Botoeva, E. (2020). The virtual knowledge graph system ontop. In International Semantic Web Conference, pages 259–277. Springer.

Zhu, Y., Wang, X., Chen, J., Qiao, S., Ou, Y., Yao, Y., Deng, S., Chen, H., and Zhang, N. (2024). Llms for knowledge graph construction and reasoning: Recent capabilities and future opportunities. arXiv preprint arXiv:2305.13168.