Prediction Of Interests Of Web Users Based On Association Rule Classification

Web mining is an important research direction in data mining. It breaks through the restriction of traditional data behaving format, and with the internet development, it becomes more and more important. It is an effective way for web users to get information.Web users classification is one of the most commonly used tasks in web mining. It can help web site provide personalized services for web users and so the users can get information conveniently .How to find all those classification patterns of web users is a research issue in web mining and has great theoretic significance and practical value.The paper makes some researches of the application of data mining to web data. It introduces the background, the significance and the structure of the paper in the first chapter. In the second chapter, the paper introduces the concept, the structure and the process of data mining techniques. The third chapter describes web mining, including the state and the difficulties in the web mining. In this chapter, the" paper summarizes web usage mining, including process and pattern finding arithmetic and puts emphases on the preprocessing of web log.In the fifth chapter, we analyze the traditional association rules arithmetic and point out some deficiencies of the arithmetic. Then we analyze the FP-growth arithmetic. Based on the analysis, we put forward a classification association rule finding arithmetic for web mining.In the sixth chapter, we attempt to apply some arithmetic, including ID3, traditional classification association rule finding arithmetic and the association rule finding arithmetic for web mining. The result shows that the association rule finding arithmetic for web mining is effective.In the seventh chapter, we summarize the whole paper and make aprospect of our researches. Main works of the paper:1. FP-growth arithmetic dose not produce the candidate items and the efficiency of the arithmetic is better than Apriori arithmetic. Based on the analysis, the paper puts forward a classification association rule finding arithmetic for web raining.2. We attempt to apply some arithmetic, including ID3, traditional classification association rule finding arithmetic and the classfication association rule finding arithmetic for web mining to predict the web users' interests. We compare the results and conclude that the classification association rule finding arithmetic for web mining is effective.3. We put forward a framework of a personalized recommended system for web sites and the framework integrates web usage mining, web content mining and web usage mining.