Research On The Technology Of Personalized Search Engine Based On Social Labeling System

In the 21st centry, with the rapid development of information and network technology, social network walk into our daily lives, and also bring endless information to us. Face these overload information and proliferation of information, we often felt helpless because we are unable to filter out the really need part from these information. Therefore, the subject on personalized recommendation came out.In this paper, I take aspects of algorithm and architecture of personalized recommendation for a detailed analysis, the main research directions are the following. First of all, introduce the history of personalized recommendation engine, analyze their respective implementation through some examples of the engines. Then, summarize the advantages and disadvantages of them, with which, I propose plans for future improvements.Next, I give the implementations of personalized recommendation algorithm, take collaborative filtering as main algorithm for implementation of personalized recommendation engine. Then define two ways of collaborative filtering algorithm, which are Item-based and User-based.Then, combined with the features of social network, which mainly is relationship between users, build a recommendation algorithm based on social labeling system. Describe the formula for calculating the similarity between resources, and the formula for related prediction.Finally, build a personalized recommendation system prototyped by previously mentioned algorithm and architecture.