Web Services Analysis For Broadband Network Users

Mass data of user logs has been generated with the rapid development of Web services. It is miscellaneous or even redundant but contains huge value. It records users’information reliably and can be used as the first-hand source to acquire users’behavioral habits. The conclusions of analysis are capable of presenting accurate information to support the fine-type operation, for example, targeted marketing and personalized customized for users. The data mining technology is at the cross road of several research fields, such as machine learning, nature language processing, information retrieval and so on. It can be used to extract useful information from mass data effectively, with the result that the analysis of users’behavior based Web mining technology is significant to them who own the data.Clustering analysis is one of the main methods of data mining research, which is helpful to find the natural model in data. The technique has a broad prospect of application that has been applied to kinds of industries like finance, internet and telecom. Some access models can be summarized by applying clustering algorithm to Web access records, especially when the data set is huge, the aggregate effect will be more obvious. In this paper, we studied and classified some usual clustering algorithms and a proper algorithm was chose and researched intensively. Then we developed Closed Frequent Item-sets Hierarchical Clustering based on Quantities algorithm (CFIHCQ), which can not only compress the data dimension properly but also discovery the similarity of the data set more reasonably.Through the network traffic monitoring equipment we can obtain the users’Web access records. This paper aimed at analyzing the data comprehensively to conclude the statistical regulations and preference characteristic of every type of Web services, and further analysis was conducted to describe the data features and variable trends within a certain time. Finally, we assigned each user into different groups according their stability of preference. We can comprehend users’ behavioral patterns through the description of clustering result. Furthermore, the general characteristics of all groups were analyzed.