A User Profile Based Pension Service Recommendation Algorithm

Recommending suitable service for elderly customers has become an important issue of modern pension service industry. In this paper, we proposed a user profile based recommendation algorithm, which consider the users' short-term behavior in pension service system such as browsing, searching, and long-term objective conditions (health status, living address, and economic condition) together to make accurate prediction. We also improved L-BFGS algorithm to enhance the convergence speed of algorithm. The extensive experiments on the real data set show that the proposed method improves the efficiency of the algorithm and the accuracy of recommendation.