Research On Algorithm Of Browsing Pattern Mining In Web Log

The development and popularity of Internet techniques makes the way of information acquiring and publishing take place great leaps and essential changes. The main applications are E-commerce, E-library, distance learning and so on, which spurs web to develop at a high speed. Web brings people rich information and great convenience, meanwhile the high requirement is desired on the design and function of website. Web is required to be intelligence and can find out the information rapidly and accurately that users need. It can also provide different services for different users, allow users to design pages according to their own requirement, provide product-marketing strategic information for users and so on. Among many direct and indirect solutions, employing traditional data mining techniques on web log is a promising approach. That is to say, basing on the principles and ideas of data mining, in accordance with the new characteristics of web log, traditional way of mining is expanded and improved, which is applied to web log and explored useful mode. We design web services that integrate users' browsing pattern so that the website is personalized and becomes intelligent. Web log mining has become a new and important research field in the world and its research is of great realistic significance.The thesis systematically introduce the entire process of web data mining and web log data mining, and research emphatically into the mining algorithm of users' browsing pattern from web log data. In the process of getting the browsing mode of the customer, the traditional proposed for data mining of association rules require several passes over the analyzed database, the I/O overhead in scanning the large database can been extremely high. In addition, the current little algorithms in browsing mode mining do take time-validity into account. In order to solve the problem, the thesis put forward to an algorithm of browsing mode mining with temporal constraint. The algorithm consideration is how to facilitate the tedious candidate generating operations in the mining procedure, simply scanning the access sequence database once. By means of assemblage operation to acquire the support of item sets. In the course of computing support, gradually modify session time to get the valid time of frequent mode. Compared with like-Apriori, it possesses the characteristics of consuming less time and expanding better. On the foundation of the new Web log data mining, the thesis supplies an improved mining algorithm of incremental updating that can be used in temporal data mining, so that it can apply the changes of Web log data quite well.At last, the thesis design and realize a web log mining prototype system and demonstrate the rapidity and efficiency of the algorithm.