User activity-based CF algorithm in value-added services

In order to overcome problems of scoring subjectively, this paper applies CF to recommendation systems of Mobile value-added service, presents the scoring way in value-added service, transfer the users' behavior to the extent of agreement. Meanwhile, using user contact degree to calculate the product rate which users don't give rate in the real world, so as to solve the sparsity problem of CF. At last, experiment is adopted to prove that the method of using service activity with contact degree is better than traditional method in the field of recall frequency and precision.