| The concept of Semantic Web was first proposed by Tim Berners-Lee in 1998.Recently, with the great public attention on internet, Semantic Web has achieved rapidprogress: The size of LOD (i.e. Linking Open Data) was grown 500 times, from 5million triples in 2007 to 25 billion triples in 2009. Furthermore, the technology ofSemantic Web is widely applied in areas like government, military, medical treatmentand internet search. Despite the full-grown situation of Semantic Web worldwide,there are some difficulties that pull down its development in China. They include: 1)The lack of Chinese LOD, i.e. Chinese ontologies and matchings between them. 2)Existing matching algorithms based on similarity calculation between English ontolo-gies can not be applied on the matching task of Chinese ontologies. 3) No solutionsexist for evaluating matching results in a large scale.To solve the above difficulties, this paper presents a matching framework specificfor Chinese ontologies matching. In this framework, we propose matching-finding so-lutions, matching-reasoning solutions and matching-filtering solutions for discoveringseed matchings, reasoning seed matchings and filtering error matchings, respectively.This framework is designed to maximally find high-quality matchings between Chi-nese ontologies. This is accomplished by updating seed matchings with positive andnegative matchings from filtering results, and repeating the reasoning and filtering pro-cess once again. Additionally, we implement an online system for manually evaluatingmatching results in a large scale. Extensive experiments have been carried out to firstextract Chinese ontologies from online Chinese encyclopedias, and then match eachtwo of them in our matching framework. The experimental results show that we fi-nally find 1.35 million matchings among these Chinese ontologies, 99% of them havean accuracy higher than 80%. |
PDF Full Text Request