| Along with the network information technology rapidly development; while enjoyed on-line shopping convenient also face the difficult which the information overloaded, How to seek interested and valuable information in the massive information, the personalized recommendation technology arises at the moment, the personalized recommendation according to user's interest characteristic and the purchase behavior for user recommendation information and the commodity, make the user enjoys"one to one"information service, it will be the future information service development tendency.This paper describes the main algorithms of personalized recommendation, collaborative filtering technology is one of the most successful recommendation technique, but with the site structure, content, complexity and number of users increases, collaborative filtering technology also exposed some shortcomings, Evaluation of individual users for goods account for onlyt 1% to 2%, it is resulting evaluation matrix data very sparse, it is difficult to find similar set of users, resulting in recommendations effect are greatly reduced, for this situation, the collaborative filtering algorithm should to improved, put forward collaborative filtering recommendation algorithm based on web data mining, mainly use the association rules algorithm and clustering algorithm recommended, focus on the clustering algorithm as a pre-association rules, a cluster analysis of the data as a guide is divided into several categories, Used in the class association rules algorithm to find similar users, it can improve the recommendation accuracy was verified by experiment.Personalized recommendation system based on web data mining is divided into two parts, offline and online. Offline part of the data preprocessing module and the model for web data mining module pattern analysis, online using offline section provides some of the major mode of use of online users recommended. And according to experiment simulate online bookstore recommendation process, and through experimental compare results of this improved algorithm which has a better recommendation quality, and to apply this technology to the online bookstore site. |
