An Approach For Measuring And Comparing Structural Semantics Of Ontologies Based On Graph Derivation

With the rapid development of the society, information technology has reached a new level. Information-sharing and efficient access to effective information is being paid more and more attention. It depends on effective information representation and management. Facing amounts of information content, it is necessary to exploit a technology of information integration and sharing. Ontology becomes a preferred technique because powerful ability of information representation and reasoning to achieve the goal. People have built a lot of ontologies in different fields for a variety of different application purposes. But some of them express similar content but with differences in naming and structure. That is the problem of ontology heterogeneity. It is the primary obstacle for interoperation of ontologies. Building so many ontologies of similar content not only wastes a lot of time and effort but also brings many problems such as ontology heterogeneity and information interoperability. So, we need the technology of ontology reuse to solve the problems mentioned above. Measuring and comparing ontologies provide a feasible solution to ontology reuse. However, the state of art approaches for measuring ontologies neglect the problems of both the polymorphism of ontology representation and the additional implicit semantic knowledge. One way to tackle these problems is to devise mechanism for ontology measurement that is a stable, the basic criteria for automatic measurement.In this paper, we present a graph derivation representation based approach (GDR) for stable semantic measurement. It captures structural semantics of ontologies and addresses those problems that cause the unstable measurement. We first introduce two concepts:semantic measurement and stable measurement and give their definitions respectively. A GDR based approach is proposed to transform an ontology to its GDR by a three-phase process. In the first phase, each axiom and assertion is transformed into the corresponding GDR named Ga by some rules. In the second phase, all the GDRs will be merged into one GDR called Go by integration operation. Then we obtain a GDR of given ontology for further processing. Last but not least, we treat Go by eliminating cycles of class inheritance and non-direct relations with transitive property. After three phases, graph Go can be used in stable ontology measurement. We formally analyze the important properties of GDRs and give the corresponding theoretical proofs. GDRs keep the semantics of ontologies and the measurement between them is stable. The effectiveness of our GDR based approach is empirically validated on a dozen of real world data with the comparison of existing graph based methods.According to the methods proposed in the paper, we develop a prototype system that can be used in ontology measurement based on common graph model and GDR by different ontology measures. With the prototype system, we carry out a series of ontology measurement and comparison experiments and the results prove the theory and method mentioned in this paper are valid.