Enriching Domain Ontologies with Knowledge-based Semantics

Abstract

Both association rules and ontologies are domain-based knowledge. However, the first is unexpected discovered knowledge from databases, while the second is a priori knowledge. In this paper, based on a generic meta-schema as common referential and theoretical foundation, we show how a set of computed and pruned association rules can be useful for enriching a domain ontology. To this end, the meta-schema is scanned with the itemsets appearing in each association rule. Then, according to the formal links between the concepts or the attributes involved, a semantic-based check constraint is built. As a result, the ontology and the database are continuously tuned with new semantics.