Concept learning algorithm for semantic web based on the automatically searched refinement condition

Abstract

Today, the web is the huge data repository which contains excessively growing with uncountable size of data. From the view point of data, Semantic Web is the advanced version of World Wide Web, which aims machine understandable web based on the structured data. For the advent of Semantic Web, its data has been rapidly increased with various areas. In this paper, we proposed novel decision tree algorithm, which called Semantic Decision Tree, to learning the covered knowledge beyond the Semantic Web based ontology. For this purpose, we newly defined six different refinements based on the description logic constructors. Refinements are replaced the features of traditional decision tree algorithms, and these refinements are automatically searched by our proposed decision tree algorithm based on the structure information of ontology. Additional information from the ontology is also used to enhance the quality of decision tree results. Finally, we test our algorithm by solving the famous rule induction problems, and we can get perfect answers with useful decision tree results. In addition, we expect that our proposed algorithm has strong advantage to learn decision tree algorithm on complex and huge size of ontology. © Springer International Publishing 2014.