Internal structure and semantic web link structure based ontology ranking

Abstract

The semantic Web is an extension of the World Wide Web with new technologies and standards that enable interpretation and processing of data and useful information for extraction by a computer. The World Wide Web Consortium (W3C) recommends XML, XML schema, RDF, RDF schema and Web Ontology Language (OWL) as standards and tools for the implementation of the semantic Web. Ontologies work as the main component in knowledge representation for the semantic Web. It is a data model that represents a set of concepts and the relationships between those concepts within a domain. Building an ontology starting from scratch is not an easy task since it makes heavy demands on time in addition to expert knowledge related to the domain. However, we can use the existing ontologies to develop semantic Web applications. But, there are a large number of ontologies available and the ontology search engine will generate a bulk of results with different ontologies for search queries. Therefore, ranking of ontologies is needed to find the most appropriate and relevant ontologies. We consider the ranking techniques and algorithms attached to the semantic Web: (i) Swoogle Ranking (ii) Ontokhoj Ranking (iii) OntoQA Ranking (iv) AKTiveRank (v) OntoSearch Ranking (vi) content-based ontology ranking (vii) SemSearch Ranking (viii) ReConRank. Our effort considers most popularly used ranking techniques and algorithms attached to the semantic Web. We analyze the above ontology ranking techniques with algorithms and then mainly categorize into two groups. One group is based on the semantic Web link structure and the other one is based on internal structure of the ontology. We identify that some features are not addressed in ranking of ontologies selected by the above ranking techniques and algorithms. Therefore, we propose a ranking method that considers both internal structure and semantic Web link structure of ontologies to improve the ranking of ontologies. We finally evaluate the proposed ranking method. According to the results with evaluation, we allocate more weighting for internal structure and low weighting for semantic Web link structure to get the best ranking results.