Incremental Rule-based Reasoning on Semantic Data Streams

Abstract

This thesis investigates the area of semantic stream processing, in which data streams are combined with semantic reasoning techniques. We have investigated techniques for rule-based reasoning over semantic streams in which reasoning is implemented natively over streams as data flow networks, and have developed an adaptive optimisation method to cope with the changing nature of streams. The contributions of this thesis include R4, a native rule-based reasoner for RDF streams using the Rete algorithm, and a cost-based adaptive plan optimiser designed for RDF streams. We have evaluated the performance of R4 and compared it to both a typical static reasoner and to the state-of-the-art in stream reasoners. The results show that R4 significantly outperforms these reasoners in terms of throughput. We have also evaluated the adaptive optimisation technique, with results that show the ability of the optimiser to devise and adopt better performing plans at runtime.

More information

Identifiers

ORCID for Nicholas Gibbins: orcid.org/0000-0002-6140-9956

Catalogue record

Export record