Lightweight Rule Extended Ontology Languages

Abstract

OWL is the current W3C recommendation for representing ontologies on the Semantic Web. While ontology languages (such as OWL-DL) provide a rich set of class constructors, the language’s expressive power is lacking when describing properties. For this reason, rule languages for the Web have been proposed. For ontologies in general to be useful on the Web, scalable reasoning services are required. For example, the MusicBrainz.org community driven music metadata service includes around 380,000 artists, 585,000 releases and 6,800,000 tracks. Even this subset of the MusicBrainz.org repository implies that reasoning support for millions individuals (and possibly even billions of statements) is required. It has been shown that OWL-DL is not scalable for large ontologies. The DL-Lite ontology language has been proposed for scalable reasoning with millions of individuals; providing a restricted set of constructors. A similar approach could be taken for rule extended ontology languages. However, users may not find that the limited expressive power of such ontology languages convenient for practical use. Semantic Approximation has been identified as a method to reduce an ontology defined in an expressive language, to its equivalent in a less expressive language. This method allows users to build ontologies in an expressive language, while enjoying the scalable reasoning services provided by the less expressive language. Therefore, an alternative, second, approximation-based approach could be adopted for rule extended ontology languages.