Study Of Personalized Recommendation Technology Based On Social Tags

With the rapid development of Internet and e-commerce, it brings information overload problem. The customers need to spend more time to browse excess products information , and difficult to find products they interested in from huge products information, so some users will be gradually loss in this process. Therefore, whether e-commerce sites can obtain more profits and the needs of users can get satisfaction are closely related. Under this background, the personalized recommendation system is emerging. Recommendation system can help users to find products they are interested in, enhance user’s trust to the e-commerce sites, and create more profits for the sites.This paper first introduces e-commerce, personalized recommendation system, Web mining technology and social tag research, etc. Then it describes the most widely used two recommendation technologies now in detail which are known as cooperative filter recommendation technology and based on content recommendation technology. It does feasibility analysis and discuss in-depth on social tags application in e-commerce sites. Then we find that social tag technology opens up a new way for research of personalized recommendation system.In traditional e-commerce websites, social tags are used in commodity classification only, and not applied in the domain of personalized recommendation technology. This paper presents a personalized recommendation model based on social tags. It reflects user’s interest and products characteristics directly by social tags, and it can construct model of user’s interest to products. This paper optimizes the interest model by social tags clustering. In order to find out the high user’s interest degree products, I design a personalized recommendation algorithm based on this model, and then realize personalized recommendation to user.Further studies showed that existing tag sorting in tag cloud is according to the number of tags marked times in descending order. This arrangement does not consider the users personalized demand and tags timeliness. Therefore this paper put forward a kind of tag sorting algorithm in personalized tag cloud. It can accord to user’s interests and mined user’s potential interests to construct a personalized tag cloud, and then update on real-time by user’s interests changing.We make experiments to test the proposed personalized recommendation models and algorithms and tags sorting algorithm. The experiment result shows the proposed models and algorithms have well recommendation effect, can improve the quality and accuracy of recommendation, and have certain effectiveness and feasibility.