Combination Of Time Windows User Access Interest In Cluster Analysis

In recent years, the sites have accumulated lots of business data and to discover the groups with similar access interest is of great significance.In this context,two new approaches,which conduct research on access interest clustering through considering both of page preferences and visiting time,are proposed to solve the problem of insufficient accuracy caused by only considering page preferences.This paper proposes a model whic captures page and time preferences of users'visits.This model contains three sub-models:Pagepreference Table, Timeframe Table and Pageprefrenence Timeframe Table, which respectively depict the users'interest in each page,the times for users'visits to the site within each timeframe and the interest in each page within each timeframe.Based on the sub-models,two approaches which group assess interest by associating and integrating page and visiting time preferences are put forward.The associating approach includes the page preferences associating timeframe algorithm and the timeframe associating page preferences algorithm.The integrating approach is the integrating page preferences and timeframe algorithm.The experiment results have indicated that the associating approach results in more accuracy clusters than those which only focus on page preferences and the integrating approach gains a better independence of the initial centroid compared to K-Means.