Mining Web Structures Using Multi-Dimensional Data Mining Model

Mining interesting structures of web organizations is a hot topic today. This paper proposes a method to mine non-trivial structures of an intranet-style web organiza,tion under multi-dimensional restrictions. To do so, we use our new datacube system named "ztemset cube". The itemset cube system does frequent itemset data-mining tasks in a datacube-style under multidimensional restrictions. Using this itemset cube, this paper analyzes structures of a web organization under the multi-dimensional restrictions about "what cores are important to which groups" and "wha,t cores are so when looking from which groups". As a new a,pproach, we model relationship between cores by a new core-community graph, and ca,lculate ranks of the cores there. Effects of our method are demonstrated by using all web sites under a university's domain.