The Research Of WEB Usage Mining Using Ant Colony Clustering Algorithm

With rapid popularization and application of Internet, Webinformation grows at astonishing speed. How to fast and effectivelydiscovered the useful information resource has become the pending issue.As a technology of mining the useful knowledge from magnanimous Webinformation, Web usage mining has received widespread attention as soonas it appeared.Web usage mining is one of the three main branches of Web mining.It takes the Web usage data as the mining object, expected to achieve theinterested usage pattern. Web usage mining has already applied topersonal service and Business Intelligence.This article firstly introduces the frame of Web usage mining: DataPreprocessing, Pattern Discovery, Pattern Analysis, Pattern Applications,and analyzes the exist question of methods using in pattern discovery indetail. Secondly introduces four kinds of model based on ant colonyclustering algorithm, and applies algorithm based on ant-piling to Webusage mining. Owing to Web data have the characteristics ofhigh-dimension and direction, we combine directional similarity measurewith ant colony algorithm and present a new ant colony clustering algorithm which named ant colony clustering algorithm based ondirectional similarity. The experimental results show that it can clusterdirectional data effectively. Finally we use the new algorithm to clusteringWeb user. In view of the dynamic state of Web, we use two mechanisms ofrenewal user interesting and decomposing clusters to achieve incrementalclustering. The result show the algorithm can efficiently achieveincremental clustering.