Visualization and retrieval of data by fuzzy clustering

摘要

A method which makes use of a fuzzy clustering is presented for visualization and retrieval of data. The first step in our technique is to represent data by graphs. We then use the spectral graph method to extract fuzzy clusters sequentially from major one to minor ones. We devise an iteration method for obtaining memberships of data in extracted clusters for undirected, undirected bipartite, directed, and their mixed graphs. Obtained membership values can be used for hunting retrieval and visualization of cluster structures by the quantification theory 3. This visualization method is applied to browsing of web networks and suggestion of new links for each web page. Clustering of web pages based on both of the keywords and web links is used for browsing of web pages and their bunting retrieval queried by keywords.