CRank: A Novel Framework for Ranking Semantic Web Ontologies

To support the reuse of ontologies on the Semantic Web (SW), various approaches have been proposed to rank these ontologies to help the users or ontology engineers to choose the appropriate ones that suit their needs. However, although some of the existing approaches for ranking ontologies are very effective, they all have been designed with a complete disregard of the degree or level of complexity of the ontologies on the SW. In fact, it is argued that the study of the complexity of ontologies in a domain provide useful information for the selection of the appropriate ones for reuse. This study proposes the CRank framework for ranking ontologies on the SW based on their degree or level of complexity. The CRank framework consists of two phases, namely, the pre-processing and ranking. In the pre-processing phase, the graph of each ontology in the dataset is processed to compute seven complexity metrics that measure the design complexity of ontologies. Thereafter, a decision making algorithm is applied in the ranking phase to rank the ontologies in the dataset by aggregation of their complexity metrics. The CRank framework was applied on a set of 100 ontologies of the biomedical domain and displayed promising results.