Service Recommendation Method Based on Collaborative Filtering and Random Forest

With the development and popularization of Ecommerce, more and more information services have appeared on the web. In order to meet users requirements more accurately, several service recommendation systems had been set up. Many methods have been proposed to discover users' interests for service recommendation, such as collaborative filtering and content based service recommendation. In this paper, a new service recommendation method is proposed based on user's interest, which combines collaborative filtering based on multiply users and random forest based on single user, and this fusion method uses cross validation model. This method can improve cold start and pick up speed. Experiment results show that the method can discover users' interest efficiently and is more accurate. This method can combine two basic methods so that the result is more accurate.