CTSC: Core-Tag Oriented Spectral Clustering Algorithm on Web2.0 Tags

With the rapid development of the Web2.0 communities, many researchers have been attracted by the concept of folksonomy from the field of data mining and information retrieval. Finding out semantic correlation of tags is avid requirement for Web2.0 application. However, no proper algorithm can tackle this task very well. This paper proposes a core-tag oriented clustering method to handle the task. The main contributions include: (1) Proposing the concept of core-tag oriented space; (2) Proposing a method called Core-Tag oriented Spectral Clustering (CTSC) to cluster tags in the new space; (3) Designing experiments to evaluate the algorithm, and the results show that CTSC algorithm performs well on clustering tags.