Research On Bionics Algorithm Based Web Log Mining

The rapid development of Internet, especially the global popularization, makes the Web information plentiful. By mining Web data, people can obtain the universal information of group user's access behavior and manner, and then to improve the Web service design, Furthermore, the understanding and analysis of user's character is helpful to develop the aimed electric business activities. In order to resolve the problems which puzzle both users and website administrators, adaptive web sites emerged. By learning the access mode of users from web server logs, it can present different interfaces for different users and perfect its own topology structure.As its positive feedback mechanism, Ant Colony Algorithm is paid more and more attention, and its application fields become wider and wider. By analyzing the relationship between ant and web user, the paper proposes an ant based self-adaptive algorithm. With this algorithm, the user can reach objective page by passing the optimal path. In addition, the paper presents the Web user mode-clustering algorithm, which is inspired by the behavior that an ant can distinguish between friends and enemies when it meets another ant .It can implement the clustering of web user by using the similar degree matrix of user's transaction.On the other hand, based on the analysis of genetic algorithm's basic theory, the paper, combining standard genetic algorithm with K-mean partition clustering method organically, presents a method that can analyze Web users the behaviors with mixed genetic clustering. The method is a clustering method that has the most optimal solution in the overall situation. Its result is obviously better than standard genetic clustering method. This algorithm can effectively eliminate the noise and obtain better user and page clustering, thus provide the effective decision for the website's administrator while the administrator design individuation business website.At last, a kind of synergetic Web mining algorithm is proposed which comprehensive utilizes the rapid global search capability of genetic algorithm and positive feed back convergence of ant colony algorithm. Compared with ant colony algorithm and genetic algorithm, this algorithm has the certain advantage on optimizing performance and time performance.