Datatype evolution on RDF ontologies

Abstract

Semantic web ontologies are not static but evolve as the understanding of the domain or the domain itself grows or evolves. This evolution is usually independent of the ontological instance descriptions which are stored in knowledge bases KBs. Therefore, the effects of an ontology update on instance descriptions which are migrated to a new ontology version should be studied. In this paper, we are interested in the effects of the migration of literal-valued instance triples of an RDF KB to a new ontology version. Note that the datatype of the property of such an instance description may have evolved in the new ontology version, e.g., from integer to decimal or from float to double. We manage the quality of these migrated instance triples and design flexible interactive methods through which the curator of the KB can intervene to correct the 'inaccuracies' of the literal values that are incurred by the migration process. In addition, specialised algorithms are presented, which are highly efficient, since their time and space complexity is linear with respect to the size of the literal-valued instance triples of the KB. Our work is particularly important to e-science applications that are commonly curated and have many datatype properties.