Interactive Document Clustering System Based on Coordinated Multiple Views

This paper proposes an interactive document clustering system based on coordinated multiple views (CMV). Recently, the amount of documents has become extremely large, which makes it difficult for a user to under-stand relationship between documents and topics mentioned in a set of documents. In order to support a user, interactive clustering is a promising approach. When designing an interactive clustering system, it should be considered that a user has to examine clustering result from various levels such as clusters, documents, and words. The proposed system divides documents into four levels and displays each level in parallel with different views. By designing coordination among those views, a user can understand document information from various levels and perform interactive clustering efficiently.