Scalable distributed ontology reasoning using DHT-based partitioning

Abstract

Ontology reasoning is an indispensable step to fully exploit the implicit semantics of Semantic Web data. The inherent distribution characteristic of the Semantic Web and huge amount of ontology instance data necessitates efficient and scalable distributed ontology reasoning. Current researches on distributed ontology reasoning mainly focus on dealing with the heterogeneity of different ontologies but pay little attention to the performance of distributed reasoning and have not presented practical approaches and systems. Our goal is to propose an efficient and scalable distributed ontology reasoning approach, making it practical in real semantic applications. We propose an approach in this paper, in which Description Logic reasoners for TBox reasoning are combined with rule engines for ABox reasoning to support both expressive ontologies and large amount of instance data. The published data from each node is distributed using a DHT-based partitioning and stored in well-designed relational databases to support convenient and efficient reasoning through cooperation of the distributed nodes. A practical distributed ontology reasoning and querying system called DORS is developed based on our proposed approach. Our experiments both in LANs and on PlanetLab using University Ontology Benchmark show high efficiency of DORS compared with the centralized OWL ontology reasoning system Minerva as well as good scalability with respect to the number of nodes and volume of data in the system. © 2008 Springer Berlin Heidelberg.