The Research Of Web Page Recommendation Model Based On User Access Matrix

The Web mining is a kind of technique which combining traditional data mining and Web. Object of the Web usage mining is the Web log data from the process which the user use the Internet. By mining these data , it can help us understand the behavior of user, thus improve the structure of Web sites or provide the Web page recommendation to the visitors. Now In the field of the recommendation model based on Web usage mining, by the association rules to predict the user access patterns have attracted much attention, but many association rules algorithms have some drawback so that it has low efficiency, plus it has low match rate between predictable result and user actual access behavior, therefore, visitor could be dissatisfied with the run result of recommendation algorithm. The paper work is around the issue existed above, it mainly includes several aspects.Firstly, Web logs data preprocessing. The information of attribute from the rough set theory was applied to the phase of data preprocessing, the concept of quantify value of importance of property was proposed, a new data preprocessed method based on importance of property was proposed. The method could filter out the noise and redundant page so as to provide smaller-scale data sets for the late processing. It can reduce the complexity of log mining effectively.Secondly, the algorithm of RCFA path was researched. First of all, the two most representative algorithm of mining frequent access path was analyzed, and then the paper analyzed the alogrithm of mining RCFA path, the CA-Mining algorithm was proposed, the experiment indicated that the algorithm accuracy and high efficiency. And the Web page recommendation part in the fifth chapter, we used the CA-Mining algorithm to mine the frequent access path.Thirdly, the method of Matrix Cluster was researched. On the basis of the Matrix Cluster proposed by Qin-Bao Song, the improved method was proposed in the paper. The page access sequence was used into the vector similarity calculation, so the improved method of the vector similarity calculation was given, the accuracy of the Matrix Cluster was improved.Finally, the paper probed the Web page recommendation model. On the basis of analyzing the common method in the field of page recommendation, a improved Web page recommendation model was given by combining CA-Mining algorithm and matrix cluster. Browsing page repeated is a common phenomenon during the actual visit, so the CA-Mining algorithm was used when we mined the frequent patterns. The experiment indicate that the recommendation model improves the precision and coverage of recommendation effectively.