A Clustering Approach for Tag Recommendation in Social Environments

Collaborative tagging is the process by which users classify shared content using keywords. Although its popularity keeps growing on the Web, content retrieval can be difficult since people tag differently. Moreover, there are some well-known linguistic phenomena. Recently, several attempts to avoid such limitations by recommending tags to users or by creating tag clusters have been presented. In this paper we propose an approach to cluster tags by monitoring the activity of the users in a tagging system. The created clusters can be used to recommend tags when a user uploads or searches a resource, in order to facilitate content retrieval. Experiments are performed by comparing with a classic tag clustering approach and results show the capability of the approach to cluster strongly related tags.