The Application Research Of The Improved Association Rules Algorithm In Web Mining

In the increasingly drastic competition of network economy,the one who wins the customers gets the final competition of superiority.The web site is the platform of the enterprise issues information , is the image and the window for foreign .Along with the web sites scale and complex increase,the site design and the maintenance work become more and more difficultly.For the enterprise operating side, in order to attract and detain more users, need a better method or tool grasps the visitor's interest,the visit frequency and so on, thus the dynamic adjustment on page structure and the improvement service meet visitor's need. In order to solve this aspect demand, more and more people pay attention the Web data mining. The Web data mining means that uses the thought and methods of the data mining to excavate the useful information on Web.This article based on the domestic and foreign research results, first introduces the basic concept of the traditional electronic commerce , and data mining technology, expatiate on the data mining classification and the process.When data mining specifically applies to the Web , this paper emphatically introduces the Web visit information mining, from its integrant part, the characteristic, the mining process and so on. When comprehensively analyzing the Web visit information mining in the data preparation process,it mentions the Web log mining,and proposes a reducing data pretreatment work load method - uses the web site filter for the difficult problem which met in view of the data pretreatment process .Simultaneously this article through the analysis on the insufficiency of the traditional association rule algorithm Apriori , and produce improvement algorithm DedApriori which inserted the data mining tool Weka.Then paper compares the different scale data mining results.Finally this article uses the Web mining thought, and many kinds of mining tools to mine commercial web site logs and the partial databases transaction record.At last paper proposes a plan improved the web site structure on the data mining results which are carefully analysed.