Improved Association Rule Mining Algorithm To Network Users Access To Log Analysis

Data Mining, which is known as Knowledge Discovery in Database, is one of the most popular study topics in Database technology field in recent years. Data Mining is a technology focused on the analysis of data and unveiling the hidden knowledge. Based on the current situation of the development of Data Mining Technology, most people think that there is still a large space to fill in this research field. Association Rules is a very important branch among it. Up to now there are many algorithms introduced. Meantime, during the huge improving of the Internet/Web technology, people are coming to realize the potential benefit of Web Data Mining, which can be used in clients'requests information analysis on the internet, increasing the efficiency of the Web Application, and so on.In this paper, we have done some research on analysis of clients visiting log on business web sites using the Association Rules for data mining. The following are the research tasks covered in this paper.First, researching the Data Mining technology, including the tasks, processes, steps of Data Ming and some important algorithms. Especially the classic Association Rules algorithms like Apriori and FP_growth are described in detail; Researching the base knowledge of Web Data Mining, and the current situation of its research and development.Second, inspecting the design and workflow of the business Web application. Analyzing the log data of clients'visiting behaviors on the site and raising the goals the Data Mining task. Based on these goals, a Data Mining Mode has been designed. First we have some previous processes to prepare the data. This will prevent meaningless association rules being generated later.Third, based on the research of FP_growth algorithm, an improved algorithm has been adopted to mine meaningful and useful association rules, which containing the information about clients'behavior modes, business web site efficiency and products marketing circumstances. This improved algorithm is more direct, more efficient, with less pointer operations, which can save more space and time. With this algorithm, the Data Mining Mode generates valuable association rules and analyzes client online behavior mode at last.