PRSPR: An Adaptive Framework for Massive RDF Stream Reasoning

Abstract

In this paper, we propose a plugin-based framework for massive RDF stream reasoning to support complicated tasks on RDF stream in an adaptive and flexible way. Within this framework, the problem of RDF stream reasoning can be equivalently reduced to the combination problem of SPARQL querying and rule-based reasoning. Take advantage of the plug-in method, we can apply various off-the-shelf SPARQL query engines and rule-based reasoners in a simple way. Moreover, to efficiently support real-time reasoning on massive RDF stream, we develop a multi-threaded batch processing approach to manage resources and an adaptive reasoning plan for dynamically managing inference rules in the stream reasoning. Finally, our experiments evaluate on dataset built on the benchmark LUBM and DBpedia. The experimental results show that our framework is effective and efficient.