Exploring logical and hierarchical information to map relational databases into ontologies

Abstract

Ontologies are formal specifications of conceptualizations. Their designs require to understand the concepts involved in the domain to be mapped. One well-known method to produce ontologies is to extract their concepts from relational databases. We conducted a practical study over a real-world scenario on applying existing rules and we identified open issues to be addressed, such as the utilization of logical metadata as a proper vocabulary, the implementation targeted to specific domains and mappings of hierarchical and self-hierarchical structures. In this paper, we present a novel approach that overcomes these issues. Our solution uses physical and logical models to enrich the terminology produced in the target ontology. It also contains a more comprehensive set of rules, taking into account instances and (self)hierarchies. We validate our approach with 2 experiments from the healthcare domain as input.

Ontology learning from object-relational mapping metadata and relational database

“…Fabro et al [12] emphasized the use of logical database metadata and schema mapping rules to enhance ontology readability and naming conventions. Lakzaei and Shmasfard [2] expanded schema mapping rules by incorporating extraction from stored procedures, user-defined functions, and views in their ontology generation process.…”