The Role of Design Rationale in the Ontology Matching Step during the Triplification of Relational Databases

A common task when publishing relational databases as RDF datasets is to automatically define a vocabulary using the schema data, called Database Schema Ontology (DSO). To automatically reuse terms from existing vocabularies, Ontology Matching (OM) algorithms are used to generate recommendations for the DSO vocabulary. However, when a relational database is partially published due to privacy reasons, the DSO vocabulary loses contextual information and OM results in quite general recommendations, such as FOAF or Dublin Core vocabularies. To reduce the loss of context and to achieve recommendations better related to the DSO context, this paper proposes to use the design rationale captured during the publication of relational databases to RDF, represented as annotations. The case studies show that this annotation strategy helps find recommendations better related to the DSO context.