Discovering Trends in Collaborative TaggingSystems

摘要

Collaborative tagging systems (CTS) offer an interesting social computing application context for topic detection and tracking research. In this paper, we apply a statistical approach for discovering topic-specific bursts from a popular CTS - del.icio.us. This approach allows trend discovery from different components of the system such as users, tags, and resources. Based on the detected topic bursts, we perform a preliminary analysis of related burst formation patterns. Our findings indicate that users and resources contributing to the bursts can be classified into two categories: old and new, based on their past usage histories. This classification scheme leads to interesting empirical findings.