Bossam: An Extended Rule Engine for OWL Inferencing

Abstract

In this paper, we describe our effort to build an inference engine for OWL reasoning based on the rule engine paradigm. Rule engines are very practical and effective for their representational simplicity and optimized performance, but their limited expressiveness and web unfriendliness restrict their usability for OWL reasoning. We enumerate and succinctly describe extended features implemented in our rule engine, Bossam, and show that these features are necessary to promote the effectiveness of any ordinary rule engine’s OWL reasoning capability. URI referencing and URI-based procedural attachment enhance web-friendliness. OWL importing, support for classical negation and relieved range restrictedness help correctly capture the semantics of OWL. Remote binding enables collaborated reasoning among multiple Bossam engines, which enhances the engine’s usability on the distributed semantic web environment. By applying our engine to the W3C’s OWL test cases, we got a plausible 70% average success rate for the three OWL species. Our contribution with this paper is to suggest a set of extended features that can enhance the reasoning capabilities of ordinary rule engines on the semantic web.