Application Of HFP-growth Algorithm In WAP Log Mining

With the popularization of the Internet, and more Internet users, there are mass logging data in server side and client side. If those data can not be used, it will become the dead sea. So the data mining rise in response to the conditions. Web log mining is a branch of data mining in Web production. For analysis and processing of the web log, we can find out the user’s behavior patterns, such as hobbies. A large amount of data has been applied, the dead sea alive. Mining Web log can be used for electronic commerce system, exploring the potential customers, improving site performance etc..The development of the network continues to the present era of mobile phone. For mobile phone Internet customers is not a minority, the WEB log mining is necessary to extend to the direction of WAP log mining. The current research in this area is still relatively less, but considering a group of users can be found, the research in this aspect will get more and more attentions. WAP log is slightly different with WEB log. WAP log mining can ignore the user identification problem of pretreatment, simply put the same mobile phone number as the same user. But the mining process is almost the same. They are both implemented on the URL data analysis, so the method of WEB mining can also be used for WAP log mining.This paper introduces the creation and development of data mining, and discusses the method of data mining and the general process flow. Especially on the Web log mining. First of all, we introduced every steps and methods in data pretreatment, and meantime we deal with the real WAP logs for preparing to the mining. Secondly, two algorithms of association rules are introduced, APRIORI algorithm and FP-GROWTH algorithm. Aiming at the shortcomings of FP-growth algorithm, HFP-growth algorithm came out. These three algorithms were compared, and exampled. In the end we choose the HFP-GROWTH algorithm to complete the system of mining. Finally, we implemented the mining system in VS2010. We have tested the system for using standard data in algorithm and finally used in the real WAP log data, and got the satisfactory answer.