Fusion-Based Automated Domain Ontology Construction Method
Dongsheng Yang, Chang Liu · 2024
Ontology is a crucial component of the Semantic Web, enabling knowledge sharing and reuse across data sources by organizing information into a knowledge network. Currently, domain ontologies are primarily constructed manually by domain experts. However, due to the typically unstructured or semi-structured nature of data sources, combined with large data volumes and the difficulty of content evaluation, the construction process is inefficient. Furthermore, relying on a single data source often fails to ensure the completeness of an ontology. To address these challenges, this paper proposes a fusion-based method for automated domain ontology construction. This approach integrates two types of data sources: relational databases and domain-specific texts (such as research papers and books). The method involves converting relational databases into domain ontology using the RDB2RDF standard, training a domain concept extraction model to identify concepts from texts, and employing pattern matching to extract relationships between concepts for ontology construction. Finally, ontology merging techniques are used to combine data from both sources, generating the final ontology. Using the field of electronic components as a case study, the proposed method was evaluated by domain experts, who confirmed the ontology's comprehensiveness and validated the approach's effectiveness.