Applying Text Generation AI to Assist Categorical Database Construction from Institutional Regulations
DOI:
https://doi.org/10.52731/lir.v005.419Keywords:
Database, Categorical database, Text generation AIAbstract
We demonstrated that a text generation AI (ChatGPT) can support the construction of categorical databases from institutional documents. By extracting concepts, formulating functional relationships, and verifying the semantic validity of composite morphisms consistency through natural language, this approach supports the creation of database schemas that accurately reflect the meaning of the original documents. It also makes it easier for non-experts to take part in schema building, providing a clear way to transform written regulations into structured data. This semi-automated method not only reduces manual workload but also improves clarity and maintainability of database structures. Our findings highlight the potential of language models to bridge formal data modeling and natural language logic in educational and administrative domains.
References
D.I. Spivak, “Functorial data migration,” Information and Computation 217, 2012, pp. 31-51.
D.I. Spivak, R. Wisnesky, “Relational Foundations for Functorial Data Migration,” arXiv:1212.5303v7 [cs.DB] 24 Jul 2015, 2015.
D.I. Spivak, R.E. Kent, “Ologs: A categorical framework for knowledge represen-tation,” arXiv:1102.1889v2 [cs.Lo] 7 aug 2011, 2011.
T. Inakura, S. Imai, K. Takamatsu, S. Matsumoto, M. Mori, “Application of Category Theory to Database Construction in Institutional Research,” Proceedings of the 12th Meeting on Japanese Institutional Research, 2023, pp. 170-175, in Japanese.
T. Inakura, S. Imai, K. Takamatsu, S. Matsumoto, M. Mori, “A database that ensures data integrity based on category theory,” Proceedings of the 40th Annual Meeting of Japanese Society of Educational Information, 2024, pp. 66-69, in Japanese.
