A Neighborhood Search Method for Link-Based Tag Clustering

Recently tagging has been a flexible and important way to share and categorize web resources. However, ambiguity and large quantities of tags restrict its value for resource sharing and navigation. Tag clustering could help alleviate these problems by gathering relevant tags. In this paper, we introduce a link-based method to measure the relevance between tags based on random walk on graphs. We also propose a new clustering method which could address several challenges in tag clustering. The experimental results based on del.icio.us show that our methods achieve good accuracy and acceptable performance on tag clustering.