Integration User Registration Information Is Collaborative Filtering Algorithm

With the rapid development of the Internet and e-commerce, how to recommendthe right products and services to users, has become one of the most challengingissues of the information age. Personalized recommendation system in this eraof information explosion came into being, it can take the initiative to the user’spreferences and predict the user may prefer products recommended to them, thetechnology has been widely used in the Internet market, and obtained goodresults.In recent years, the industry made a number of advantages recommendationalgorithms, collaborative filtering can handle complex because of itsunstructured objects, and a higher degree of automation and personalization,has become one of the most widely recommended and most effective algorithm. Whilecollaborative filtering algorithms have been recognized by the market, but therestill exist such as cold start, the sparsity of other problems need to be resolved.Aiming at these problems, expand the following exploration and research:First, this paper presents the user registration information fusioncollaborative filtering algorithm to solve the problems of traditionalcollaborative filtering algorithms. The traditional collaborative filteringalgorithms only for the interaction between the user and system information (suchas evaluation information), recommended severely constrained by the accuracyof the size of both mutual information, and can not meet the new needs of users.Improved algorithm combines the user first visit the website and sign up to becomea site member left the information, and use the registration information to findsimilar neighbors, new users may predict interests and recommend appropriateinformation to solve the cold start and sparsity.Second, through the user registration information (such as gender, age) toanalyze user interests, the result is more practical to use for new users. Thetraditional collaborative filtering algorithm for data processing is a unifiedprocess, this paper presents an asynchronous differentiation recommendationalgorithm, by setting the number of threshold assessment were for new users andold users of different personalized recommendation to improve the recommendationalgorithm accuracy.