Research Of Web Data Mining For Electronic Commerce

Web Mining is by means of data mining techniques, to discover useful knowledge from Web documents and users' behaviors. In e-Commerce, the users' browsing patterns and preference can be acquired from Web server logs and business data, then the successful e-Commerce mode is achieved. According to the mined results, the group users' behavior patterns can be clearly understood; the service quality of the e-Commerce Web site based on different user groups can be dynamically improved, helping the business marketing strategies to be effectively carried out.Although researchers have done much work on the Web mining theory, their main works focus on the design and improvement of the efficiency of the mining algorithms, little done for the design of Web mining systems.This paper analyzes the characteristic of existing prototype systems, and proposes a new prototype of Web mining system for e-Commerce (eWSM). This prototype has been implemented as a practical Web mining tool eWSMiner. The mining functions are integrated from a series of mining functions embedded in arithmetic library. The call interfaces of these mining functions are provided, which enables multiple mining tasks to be carried out in the same mining tool.we present the structure of the eWSMiner, and some data mining algorithms such as association rule, clustering, sequential pattern, which are implemented and tested in the tool, we also give a detailed analysis for the data preprocessing. eWSMiner is helpful to find users' frequent traversal path, and analyze the uses of Web sites, and improve the structure of Web sites for e-Commerce tradeoff.As a Web data mining tool, eWSMiner is able to process different kinds of input data, support common mining algorithms and the visual expression of mining results. This tool also has a good integration and extensibility.