Concepts extraction in ontology learning using language patterns for better accuracy

Abstract

The identification of concepts and relations via automatic or semiautomatic are tasks in Ontology Learning. The Ontology Learning is important in minimizing effort of ontology development. It has been used in many disciplines including development of Quran ontology. In the Quran ontology development, there have been efforts to identify concepts and relations for ontology development using various methods. Among the methods employed to discover concepts is a regex pattern. The pattern is based on NLP which use tagging in their rules. This paper proposed a method that used patterns to extract concepts for Hajj Ontology development. It also has been compared against a prominence Ontology Learning system i.e. Text2Onto. The patterns also have been compared with Qterm pattern which is specifically designed for Solah domain in the Quran. Results indicate that the proposed patterns improve the precision with 82.4% and recall with 85.7% as compared to the both approaches.