Abstract
As one of the important aspects of semantic web, building ontological models became a driving demand for developing a variety of semantic web applications. Through the years, much research was conducted to investigate the process of generating ontologies automatically from semi-structured knowledge sources such as Wikipedia. Different ontology building techniques were investigated, e.g., NLP tools and pattern matching, infoboxes and structured knowledge sources (Cyc and WordNet). Looking at the results of previous approaches we can see that the vast majority of employed techniques did not consider the linguistic aspect of Wikipedia. In this article, we present our solution to extract ontologies from Wikipedia using a linguistic approach based on the semantic field theory introduced by Jost Trier. Linguistic ontologies are significant in many applications for both linguists and Web researchers. We applied the proposed approach on the Arabic version of Wikipedia. The semantic relations were extracted from infoboxes, hyperlinks within infoboxes and list of categories that articles belong to. Our system successfully extracted approximately (760,000) triples from the Arabic Wikipedia. We conducted three experiments to evaluate the system output, namely: Validation Test, Crowd Evaluation and Domain Experts' evaluation. The system output achieved an average precision of 65 %.