Knowledge Graph-Guided Entity Linking with LLMs for the Identification of Construction Cost Items

• Renato Ramos UFSC
• Renato Fileto UFSC
• Diego Calvetti UFSC

DOI: https://doi.org/10.5753/eniac.2026.31314

Resumo

Linking construction service descriptions to cost catalog items requires technical compatibility, not only textual similarity. This paper frames the task as catalog-constrained entity linking, using a Knowledge Graph (KG) derived from SINAPI, the Brazilian National System for Construction Costs and Indexes. The proposed pipeline applies this KG to constrain decisions and retrieve compatible items that are then disambiguated by a Large Language Model (LLM). On 354 annotated descriptions from three public-building projects, the pipeline improves final Hit@3 by 10.73–32.77 percentage points over a hybrid retrieval-and-disambiguation baseline across model and candidate-list configurations, reaching 96.61% in the best configuration.

Referências

Akanbi, T. and Zhang, J. (2021). Design information extraction from construction specifications to support cost estimation. Automation in Construction, 131:103835.

Boscariol, M., Meschini, S., and Tagliabue, L. C. (2024). A methodological approach to asset information management via knowledge graphs and large language models. In Proceedings of the 2024 European Conference on Computing in Construction.

Brasil (2021). Lei no 14.133, de 1o de abril de 2021. Lei de Licitações e Contratos Administrativos.

CAIXA (2026). SINAPI – sistema nacional de pesquisa de custos e Índices da construção civil. Página institucional. Acesso em: 12 jun. 2026.

Deza, J. I., Ihshaish, H., and Mahdjoubi, L. (2022). A machine learning approach to classifying construction cost documents into the international construction measurement standard. arXiv preprint arXiv:2211.07705.

Ding, Y., Ma, J., and Luo, X. (2022). Applications of natural language processing in construction. Automation in Construction, 136:104169.

Farghaly, K., Soman, R., and Whyte, J. (2024). cSite ontology for production control of construction sites. Automation in Construction, 158:105224.

Gatto, C., Mirarchi, C., and Pavan, A. (2025). Towards automation in cost estimation: LLM-based methodology for classifying and extracting cost data from tender docu ments. In Proceedings of the 2025 European Conference on Computing in Construction.

IBGE (2026). SINAPI – sistema nacional de pesquisa de custos e Índices da construção civil. Página institucional. Acesso em: 19 jun. 2026.

Iranmanesh, S., Saadany, H., and Vakaj, E. (2025). LLM-assisted Graph-RAG information extraction from IFC data. In Proceedings of the 2025 European Conference on Computing in Construction.

Qdrant (2026). Hybrid and multi-stage queries. Documentação técnica. Acesso em: 19 jun. 2026.

Rasmussen, M. H., Lefrançois, M., Pauwels, P., Hviid, C. A., and Karlshøj, J. (2019). Managing interrelated project information in AEC knowledge graphs. Automation in Construction, 108:102956.

Robertson, S. and Zaragoza, H. (2009). The probabilistic relevance framework: BM25 and beyond. Foundations and Trends in Information Retrieval, 3(4):333–389.

Sevgili, Ö., Shelmanov, A., Arkhipov, M., Panchenko, A., and Biemann, C. (2022). Neural entity linking: A survey of models based on deep learning. Semantic Web, 13(3):527–570.

Shamshiri, A., Ryu, K. R., and Park, J. Y. (2024). Text mining and natural language processing in construction. Automation in Construction, 158:105200.

Shen, W., Wang, J., and Han, J. (2015). Entity linking with a knowledge base: Issues, techniques, and solutions. IEEE Transactions on Knowledge and Data Engineering, 27(2):443–460.

Sousa, L. J. d., Martins, J. P., Sanhudo, L., and Silva, J. M. (2025). Portuguese construction dataset for AI: BOQ text extraction and synthetic data augmentation using LLMs. In Proceedings of the 2025 European Conference on Computing in Construction.

Stellmacher, D., Große Lögten, K., Theißen, S., Kelm, A., and Meins-Becker, A. (2025). Digital decision support in construction: Analysis of information requirements and data provision for AI-based selection of sustainable building products. In Proceedings of the 2025 European Conference on Computing in Construction.

Suárez-Figueroa, M. C., Gómez-Pérez, A., and Fernández-López, M. (2015). The NeOn methodology framework: A scenario-based methodology for ontology development. Applied Ontology, 10(2):107–145.