Abstract
This research addresses the semantic and knowledge gap problem in information retrieval by proposing an ontology-based semantic feature-matching approach, which uses natural language processing, named entity recognition and user-oriented ontologies. The approach comprises four steps: (i) user-oriented ontology building; (ii) semantic feature extraction for identifying information objects; (iii) semantic feature selection using user-oriented ontologies to enhance the semantic representation of the information objects, and (iv) measuring the similarity between the information objects using their enhanced semantic representations. The experiment conducted explores the retrieval performance of the proposed approach and shows that it consistently outperforms its corresponding term-based approach by demonstrating improved precision, recall and F-score.