Research And Implementation Of Personalized Recommendation System Based On Web Log Mining

Along with the constant development of the Internet, Information overload fansto their own resources and people on the efficient use of the Internet has become theinformation bottleneck. It is hoped that the content of the website, to the extentpossible, to be adjusted according to the user's browsing interest, so that every userfeel like they are the only website users. The key to achieving this goal is to discoverhow Web users preferences, Dynamically customized for viewing the contents of theproposed visit or provide. This is defined as the technique of web personalizedrecommendation. It is a central issue in the research and application of webtechniques.Intending for the specific application requirement on the personalizedrecommendation for Zhejiang online web, the paper probes deeply into the clusteringalgorithm in WebLog mining. Aiming at the specialty that the WebLog has a highdimension and a huge data, CLOPE (Cluster with Slope) is Improvement.Furthermore, it is applied in mining users' browsing modes based on the WebLogfrom Zhejiang Online Educational channel. The outcome of experiments proves thatimproved CLOPE is efficient in the web mining of variable websites, especially inclustering large and sparse transactional databases. The improved clusteringalgorithm is fast and scalable.A personalized recommendation system, RSWLM (PersonalizedRecommendation Systems based on WebLog Mining), is designed and partially implemented here. The system includes three modules: preprocessing of the WebLog,mode mining and real-time recommendation module. Experiments indicate theRSWLM is capable of digging out from WebLog the browsing mode users areinterested in, offering better recommendation service to them and guiding them inbrowsing. As a result, the efficiency and accuracy of scanning are improved. At thesame time, opinions on improving the construction of website are put forward.