Web-based Data Mining Technology And Its Application

With the rapid development of Internet and information technology, information is changing every minute and increasing at an amazing rate on the Internet, thus, Web-based data mining technology has an unprecedented development. Now it is developing a hot research topic that how to use the wealth information resources effectively on the Internet, find the access patterns of users and the useful information timely, and provide personalized information service for users. Web usage mining is proposed in time for us. Web usage mining is also known as Web log mining, which is an important Web-based data mining technology. It can find user access patterns of the web by analyzing the Web log. If we apply the mining results to optimize the website design and provide users with personalized service, service quality of the website will be improved and users' access interest of the website will increase.First of all, the research background and research status of this field, data mining, Web-based data mining technology and personalized recommendation technology were introduced in this paper. Next, Web usage mining and collaborative filtering algorithm were explored and Project-based collaborative filtering algorithm was highlighted in this paper. On the basis of the traditional algorithm's analysis and research, a novel and effective algorithm is presented. The new proposed algorithm can solve the data sparseness problem, solve the new project-user problems and improve the accuracy of the recommendation; its performance is superior compared with the traditional algorithm. In order to verify whether the proposed algorithm has better recommendation effect than the traditional algorithm, a comparison experiment was designed in chapter 3 and MovieLens data set was used in the experiment. The experiment results show that project-based collaborative filtering recommendation method has better recommendation effect and quality than the traditional algorithm in the case of extremely sparse data. At the same time, a movie recommendation system was developed in chapter 4, which has used the proposed algorithm and reflect recommendation effect of the proposed algorithm intuitively.The summary of the main research work were concluded and suggestions for the future work was proposed in the last chapter.