Parallel Reasoning With OWL Ontologies

Ontology reasoning is the core of the application of knowledge graph.In recent years,with the appearance of Semantic Web and the further development of semantic technology,ontology reasoning has gained more and more attention of researchers and the industry.Conjunctive query answering is the most fundamental reasoning task.In the existing mainstream reasoning algorithms for solving query answering,there is a drawback that the reasoning efficiency is not high and the correctness and completeness of the results can not be guaranteed.Therefore,it's a challenging task to solve effectively the problem of query answering on large-scale OWL ontology.In this paper,we propose a distributed query answering method for OWL ontology.It can efficiently solve large-scale OWL ontology reasoning problem by materializing the universal model off-line.In this method,first,its universal model is obtained named U_k.That is the expansion of ABox by adding additional instance name tag in semantics.Second,we optimized the model,in order to parallel reasoning we analyze the inherent correlation between the rules and allocatethe concurrent threads reasonably.Then we proposed a limited model method,the universal model is often infinite.Therefore,the technical problem is obtaining the step size of the query,and we according to the size of the query to limit the size of the model.Then,the conjunctive query in the standard reasoning task is transformed into a SPARQL query,and each vertex in the RDF graph is treated as a unit of executable computing.The whole graph is mapped into a vertex set that can convey the message to each other,making full use of the graph characteristics,the use of message delivery method to improve the query graphand reduce the number of variables gradually,the use of gStoreD,RDF-3X,TriAD query engine to get the query results.The experiment designed in this paper is validated and evaluated from both the query performance and the correctness and completness of the results.To sum up,based on off-line universal model materialization method,which improves the reasoning efficiency and the correctness and completeness of the result effectively.Providing a new solution for ontology reasoning.It also shows that this article has certain theoretical and practical significance to further improve the knowledge graph and solve reasoning problems such as query answering.