The Research On Personalized Recommendation Technology Based On XML And Association Rule

With the rapid development of Internet, the demand of information service via Internet has increased sharply. However, the huge amount of information has added to the difficulties in the information selection and digestion; the wide distribution of information has made it hard for individual user to find the information he or she is interested in. Thus, a new technique is badly in need, which could automatically dig out desirable information from the huge pool of resources, withdrawing it and at the same time filtering out other information unwanted. Fortunately, the emergence of personalized recommendation technology relieves us of infinite data and the commercialized world by saving us plentiful time and energy on searching information. In addition, this new technology also successfully transforms the service of website from webpage-centered mode to user-oriented one. It supplies users with personalized services and forges ahead toward the realization of supreme level of the whole Internet services.Because the existing personalized recommendation technology has not solved the problems about anonymous user information collecting and real-time recommendation and accurate recommendation. This thesis focus on how to make use of the Web mining and XML analysis log to get the customer's access to the website pattern, the adoption of personalized recommendation technique based on association rules promotes the websites'visiting efficiency.The main works of this thesis are listed as follows:â‘ To analyze and elaborate the research background, current situation, application of personalized recommendation technique as well as the theoretical foundation of web mining, and to discuss the basic principle of association rules.â‘¡This thesis discusses the realization of the storage and representation of Web log through the XGMML and LOGML deriving from XML technique, which accomplishes the preprocessing of web log by following such procedures as data cleaning, user identification, session identification, path completion and transaction identification.â‘¢After the analysis of the Apriori algorithm and FP-growth algorithm, to put forward the new FP-growth algorithm based on MFIT. The new algorithm has better efficiency in the implementation.â‘£To design and realize a personalized recommendation system prototype which improves the recommendation quality by calculating the distance between the page.The research work is a practical improvement to the FP-growth algorithm. It has valuable reference to the research of association rule mining algorithm. The research on user visiting mode improves information service quality and promotes the development of the intelligence information processing, which is significant in both theory and practice.