Abstract: Aiming at the problem of low accuracy of extracting ontology information from Wikipedia information box in traditional methods, we studied the attribute structured information in Wikipedia information box. Firstly, a set of candidate features was defined to determine the relationship between information box attributes, and the association with categories, lists, articles and Wikipedia information box templates was established. Secondly, using the method of ontology matching to extract the structured information of Wikipedia information box, calculate the similarity of attribute pairs, set the boundary constraints, and construct ontology structure to explain the relationship between attributes and construct a class hierarchy with a certain accuracy. The results show that the proposed method solves the problem of low accuracy of extracting ontology inform
ation, and can extract the possible attribute structure in a given topic article effectively and correctly, and find the reasonable class relationship.

Key words:  , Wikipedia, information box, ontology, class hierarchy