Research Of The Clickstream Data Warehouse And Data Mining

E-commerce website produces a large amount of clickstream data everyday. There is a lot of valuable information for enterprises in the clickstream data. People can see a lot of front-end systems; however, they pay less attention to the structure of the site, user's access time, residence time, the relation between the pages and so on facing with many websites. Log analysis tool also can make a good evaluation to the site's basic statistical data. But log analysis tool doesn't refer to an important content of the website—the analysis of the users' activities who access the website. The purpose of establishing clickstream data warehouse is to promote the commerce development of website by the analysis of the user's activities.Clickstream data warehouse (Web log data warehouse) is one important kind of the data warehouse. Compared to the traditional data warehouse, the main data sources of clickstream data warehouse are the web log files which are left behind in various network servers when people are in a variety of network activities and other data related databases. It can not only solve the problem of mass data storage but also ensure the usability, high efficiency of e-commerce system and data security by establishing a clickstream data warehouse with a rational structure and making an effective analysis combining with Data Mining (DM) technologies for its massive data.In this thesis, the establishment of clickstream data warehouse is facing the analysis of user's information interest. In the process of the application, this thesis introduces a concept of Operational Data Store (ODS) for the sensitivity of the response time for e-commerce environment. At the same time, this thesis mainly introduces and analyses the scheme of a 3-tier data warehouse based on ODS for the shortcomings in the traditional 2-tier (DB-DW) data warehouse. In the process of data preparation, this thesis focuses on the researches of using cookies to handle of the local cache and proxy server. It can obtain the potential access information of user by making an effective analysis of the web log data of an e-commerce website of Dalian Maritime University using clickstream data warehouse and data analysis, data mining technologies based on SQL Server 2005 combining with Data Mining System based on Java so as to provide better quality services for web users. It will be a tendency of future development that transfers the well-rounded data mining arithmetic to the domain of web log and makes a deep analysis of the characteristics of user activities based on data warehouse.