Graph-based partitioning of ontology with semantic similarity

Ontology partitioning is a good solution to overcome challenges of large ontologies such as reusability, scalability, complexity and maintenance. The languages of ontology such as OWL do not allow partial reuse and one must use the whole ontology. The problem is that processing large ontologies consumes more time and space than processing only parts of ontologies. To produce high quality modules, one must consider both semantic and structure of an ontology. In this paper, we partition a given ontology according to its semantic and structure to accomplish reusability. Semantic is considered according to weights given to different relationships, and random walk distance is used to maintain a balance between semantic and structure. In our work, more relationships are taken into account than other approaches that only consider hierarchical relations. We propose an agglomerative and a heuristic algorithm to partition ontologies. Our methods produce meaningful modules and reduce required time and space to process an ontology.