| With the rapid progress of Web technology,more and more Web-based applications have emerged, such as news service, mail service, online game, online shopping, online paying. More and more people get the service they want from Web. And Web has become an essential element of an individual’s daily life. The frequent usage of Web by the numerous Web users lead to a lot of Web users’ visiting records which contains implicit and useful knowledge. Web service providers urgently need for the technology to find out Web users’ particular interests so as to provide them with personalized service to help Web service providers get the upper hand in the fierce competition.As a pervasive computing theory, Genetic Algorithm has the advantages of global optimization, implicit parallelism, etc, and has been applied to many fields such as artificial intelligence, manufacturing, social science and so on. Firstly this paper discusses the main content of Web behavior character research which presently focuses on Web user clustering, analysis of associate Web pages, analysis of preferred accessing sequence and so on, and also discusses the related technology of data mining. Secondly this paper discusses the general process of the research on Web behavior character, and emphatically discusses the process of data pre-processing in detail. Thirdly, this paper discusses the basic principles of genetic algorithms and its operating procedures, basic principles of clustering and its common algorithms, and analyzes the problems to be solved when using genetic algorithms to achieve clustering. Then we use the theory of clustering and genetic algorithm, combined with topology of web site, to implement user clustering on Web log data and study Web behavior character emphatically. In the end, the basic thinking and implementing details of the system are given, and the experimental data and experimental analysis are also given. Experimental results show that the proposed algorithm has some practical value. |
PDF Full Text Request