Authors: Yongxiang Dou; Jiyuan He; Dongsu Jiu
Addresses: Department of Information Management, Xidian University, 2 South Taibai Road, Xi'an, 710071, China. ' Department of Information Management, Xidian University, 2 South Taibai Road, Xi'an, 710071, China. ' Department of Information Management, Xidian University, 2 South Taibai Road, Xi'an, 710071, China
Abstract: In this paper, we propose a comprehensive approach for deriving ontology from folksonomy. In our method, we mainly utilise spectral clustering algorithm (SCA) based on the concept of similarity between points instead of the distance. The details are as follows. Firstly, we filter tags that are wrong or non-English words. Secondly, overlap similarity is used to measure similarity between co-occurrence tags. And SCA is applied to cluster tags with semantic relations. Moreover, another algorithm is adopted to obtain hierarchical structure of tags by running SCA repeatedly. Finally, online ontology resource WordNet is applied to add definitions and other semantic relations between tags to finish the work. We also conduct an experiment in which we use the tags from Delicious website as the external knowledge source for the domains of the tags. And the result shows that this method is appropriate for ontology construction of emerging areas due to that folksonomy covers the latest concepts.
Keywords: folksonomy; ontology construction; spectral clustering algorithm; SCA; overlap similarity; hierarchical structure; tag filtering; semantic relations.
International Journal of Services Technology and Management, 2012 Vol.18 No.1/2, pp.88 - 101
Available online: 09 Sep 2012 *Full-text access for editors Access for subscribers Purchase this article Comment on this article