Web-based Mining Site Structure Optimization Studies

As the development of internet, more and more businesses are conducted on it, and what the enterprises focus more on is the law how users make use of the Web. As well, the contradiction between the explosive growth of information and the limitation of people's attention becomes more obvious. Web mining is an effective means to resolve this contradiction. It can get information ahout web users'interest and mode to access to websites by way of picking up some saved traces from the users, which could provid the website administrater with informations to get more economic benefits and optimize website structure, and then improve the website accessing efficiency as well as perfect the operation of websites.This dissertation firstly describes the background and the status quo of Web mining home and abroad, and then introduces data mining, Web data mining and Web log mining respectively. The main research is mainly about the theory and technology relating to Web usage, and emphasizes more on the association rule mining algorithms and the maximum frequent access patterns mining algorithm. The achievements of the dissertation are as follows:1. It takes the Web log data, page content and site structure information as a data source to conduct data preprocessing, which has overcome the traditional data mining problem of a single data source and improved the accuracy of recognition of users'accessing pattern;2. According to the characteristics of the page relevance and client similarity, Web pages and client groups are analyzed respectively. The relative weight of the site structure information is taken as the accordings to arrange pages. On the basis of the page analyses, an effective website index page is established to allow users to access to more effectively; according to the customer group identity, it can provide effective access recommendation and optimize network performance.3. To overcome the disadvantages of association rule mining algorithms and frequent access pattern mining algorithm, an improved FP-Tree algorithm FP-Apriori is proposed. It gets the user's frequent access patterns, and optimizes site structure in line with frequent pattern set. So users'effective access rate and site quality are improved;4. This dissertation realizes the above algorithm, develops a Web site optimization system and proposes site structure optimization program and site structural adjustment methods based on WEB mining. Furthermore, it compares the access efficiency before and after improved to obtain feasible validation.