Automatically clustering social tags into semantic communities would greatly boost the ability of Web services search engines to retrieve the most relevant ones at the same time improve the accuracy of tag-based service recommendation. In this paper, we first investigate the different collaborative intention between co-occurring tags in Seekda as well as their dynamical aspects. Inspired by the relationships between co-occurring tags, we designed the social tag network. By analyzing the networks constructed, we show that the social tag network have scale free properties. In order to identify densely connected semantic communities, we then introduce a novel graph-based clustering algorithm for weighted networks based on the concept of edge betweenness with high enough intensity. Finally, experimental results on real world datasets show that our algorithm can effectively discovers the semantic communities and the resulting tag communities correspond to meaningful topic domains.
Pan, Weisen; Chen, Shizhan; and Feng, Zhiyong
"Automatic Clustering of Social Tag using Community Detection,"
Applied Mathematics & Information Sciences: Vol. 07
, Article 34.
Available at: https://dc.naturalspublishing.com/amis/vol07/iss2/34