Research on Chinese Ontology Instance Extension Based on SVM

Abstract

Extension of ontology instance is the important part of ontology maintenance. In this paper, a novel and effective method is proposed to extending ontology instances from Chinese free text, which is achieved with classification using support vector machine (SVM). Firstly, classification features are extracted in terms of syntax and semantics from the training texts and the new texts based on the existed Chinese ontology. Then the ontology is turned into tree hierarchical structure which is used as the training and learning strategy of SVM classifier. Finally new ontology instances are extracted from the new texts according to the training results. The advantage of this method is that the semantic of ontology elements in texts is made full use of, and instances extraction and classification are completed in the identical procedure at same time. Experimental results show that the average accuracy of instances extraction and classification reached 86.6%, which is satisfactory.