OntoCons: A Method for Intelligent Construction of Domain Ontology Models for Large Language Models
Yi Li, Wenxin Lu, Xiuzhen Xiang · 2025
With the development of large language models technology, its application in the field of education has been steadily expanding. However, due to limitations in understanding specialized literature in vertical fields, large language models face serious issues of “hallucination” and prior bias when responding to teachers' questions about specialized domains. To address this, this paper proposes an intelligent construction framework for ontology models based on knowledge tuple extraction, called OntoCons, which introduces entity extraction and relation extraction methods into the field of ontology model construction. OntoCons transforms the ontology model construction problem into a knowledge tuple extraction problem, enhancing the transparency and interpretability of ontology model construction methods, and introduces subgraph encoding strategies to improve the accuracy of relation extraction. This research provides a new method for the intelligent construction of ontology models in vertical fields, improving the understanding and response quality of large language models in specialized domains by combining machine learning and manual analysis.