Research On Ontology Modularization In The Domain Of Radio

The emergence of the Semantic Web brings ontology into a vigorousdevelopment stage. Due to the special advantage in expressing information, ontologyhas been used in many disciplines about computer in the recent years. At the sametime, there have been an increasing number of industries that use ontology torepresent expertise. However, with the explosive growth of information, theever-expanding ontology scale has led to a series of new problems, such as theincrease in the complexity of ontology management and the difficulty of ontologydevelopment. The modularization of ontology is widely regarded as an effective wayto solve the management and maintenance of a large ontology by researchers. Thus,continuous research efforts lead to a number of modular ontology descriptionlanguages and methods, or improved schemes.In this paper, we propose a new method of ontology modularization, based on thestructural characteristics of a radio management ontology knowledge base. Thismethod is a graph-based modular approach, divided into three levels: namelyextracting the terms of input ontology to generate undirected graphs, calculating thedependence of edges to generate dependency graphs, and partitioning undirectedgraphs and dependency graphs based on three partitioning strategy. The relationshipsin an ontology are divided into two categories, object properties of concepts which areconsidered in the partition stage, and inheritances between concepts and otherrelationships been considered in the optimization stage. The impact on the ontologystructure that caused by individuals has been taken full account of, because of the factthat a concept contains a number of individuals, and the total number of individuals isfar more than the total number of concepts in the radio management knowledge base. Frequent contacts between individuals will inevitably affect the degree of associationbetween the concepts.In this paper, OWL API provides an interface to achieve the three partitioningstrategy, and the radio management ontology knowledge base been used as the sourcefile to test the three partition strategy. We get the better strategy by analyzing andcomparing the experimental results. At the same time, we provide a feasible methodfor partitioning ontology which could be rational references to other modularizationapproaches.