Comparison of Conventional and Rough K-Means Clustering

This paper compares the results of clustering obtained using a modified K-means algorithm with the conventional clustering process. The modifications to the K-means algorithm are based on the properties of rough sets. The resulting clusters are represented as interval sets. The paper describes results of experiments used to create conventional and interval set representations of clusters of web users on three educational web sites. The experiments use secondary data consisting of access logs from the World Wide Web. This type of analysis is called web usage mining, which involves applications of data mining techniques to discover usage patterns from the web data. Analysis shows the advantages of the interval set representation of clusters over conventional crisp clusters.