Conference presentation
Clustering tagged web based on the tripartite network of folksonomy
Drexel University. College of Information Science and Technology. Research Day Posters.
01 Nov 2010
Abstract
Social tagging, also called social annotation and collaborative tagging, is a recent phenomenon in the collaborative web space. The social tagging activities by users form a tripartite network, which is composed of three types of nodes: webpages, tags, and users. The underlying structure of such a tripartite network can be analyzed and utilized for many application purposes, such as webpage classification and cluster, social interest discovery, automatic tag suggestion, and personalized web search. The research outlined in this poster aims to identify the hidden structures from such a tripartite network formed in social tagging systems. For this purpose, we adopt a model proposed in previous research for analyzing the structure of K-partite graphs. Based on this model, a relation summary network is constructed to approximate the original tripartite social tagging network through a range of distortion measures. The constructed relation summary network not only provides information about the local cluster structures of the three type of nodes (webpage, tag, user), but also the global connection structure of the whole social tagging network. Experiments are carried out on a real-world social tagging dataset sampled from del.icio.us from January 2009 to February 2009. Meaningful results are obtained and discussed.
Metrics
8 File views/ downloads
9 Record Views
Details
- Title
- Clustering tagged web based on the tripartite network of folksonomy
- Creators
- Caimei Lu (Author) - Drexel University (1970-)Tony Hu (Author) - Drexel University (1970-)Jung-ran Park (Author) - Drexel University (1970-)
- Publication Details
- Drexel University. College of Information Science and Technology. Research Day Posters.
- Resource Type
- Conference presentation
- Language
- English
- Academic Unit
- DU; College of Information Science and Technology (1995-2013)
- Identifiers
- 991014632433504721