Three-Tier Clustering: An Online Citation Clustering System

In this paper, we present an online citation entry clustering system based on three-tier clustering. The objective is to further process search results returned by bibliography databases and present to the user with more accurate results. By our approach, a user first issues an author name query and it is passed to a data source chosen by the user. We then exploit the unique usage of each citation entry and cluster the returned citations according to the queried author names and present the result clusters to the user. The preliminary experimental results indicate that such an approach can greatly ease the user’s browsing by picking up clusters he/she is interested in. The architecture of such a clustering framework, feature representation of a citation entry, a brief network model for inter-object similarity calculation, a special clustering evaluation technique are discussed. Experiments on the effective of the clustering framework are also presented.