An Efficiency Revaluation of Graph Labeling for subClassOf Inference based on OWLJessKB Inference Rules
Jae‐Hoon Kim, Seog Park · 2011
It is an important research problem to speedily inference large OWL data. As an effort for this problem, some graph labeling techniques for OWL data have been studied like XML labeling techniques. However, the former studies for graph labeling gave their idea in the level of DAG (Directed Acyclic Graph) into which OWL data are simplified. They could not consider OWL inference rules based on OWL Semantics like the entailment rules of RDF Semantics. OWLJessKB developed in Drexel University is representing many OWL inference rules in Hom clause based Jess language. However, OWLJessKB is incomplete. In this paper, based on the actually used OWL inference rules, we reevaluate the efficiency of graph labeling for subClassOf inference. OWL inference is explained by using Rete algorithm adopted by Jess which is an rule-based inference engine. Experimental results show that graph labeling speeds up inference and enhances space cost and execution performance by omitting storing the new RDF triples generated by subClassOf inference.