Research On Click-stream Technology Based On Data Warehouse

With the rapid development of Internet, online transactions become an important form of people's daily dealings behavior. E-commerce therefore gets a vigorous development, and is becoming a trend. What followed by this is a large number of web data are generated. The web data are widely distributed in the network servers, around the world. As long as you have a web server with an e-commerce system in, you can get the web data with huge potential commercial value. The development of information technology allows people to study the web data becomes a reality, but still there are two issues: how to effectively organize and store such a large magnitude data? How to analyze the massive data with effective data analysis methods, and find valuable information? These two questions are also the content to be studied in the paper.Data warehouse was firstly proposed as a solution for massive data storage problems based on relational database. Data mining was also proposed as a method to analyze the large amount of data stored in relational database. With the continuous development of web technology, data warehousing, data mining technology were gradually integrated with the web, allowing web data warehousing and web data mining techniques emerged.The paper first give an overview of the basic concepts and relevant technologies of data warehouse and data mining. With the knowledge of web data warehouse, the process of building click-stream data warehouse in sales automation system is detailed described from data warehouse overall architecture, model design, metadata design and so on. Then, we come to an available click-stream data warehouse solution. On this basis, the click-stream data warehouse is implemented by using Microsoft data warehouse tools. After that,the paper discusses the click-stream data collection methods, and divides click-stream into two types: static and dynamic. It analyzes the problems of static click-stream from web server log files and difficulties in data pre-processing. A thought of dynamic click-steam data collection strategy is proposed. The method makes click-stream data collection efficient and easy, and avoids the problem of data pre-processing. According to web data mining technology, a user frequent preferred path mining algorithm is proposed. Users' co-interests of browsing can be found and users' behavior can be deeply studied from multidimensional. The structure of sales automation system can be improved by using the results of the algorithm. The mining algorithm is implemented by using web development technology and the results are displayed.