Solutions for problems of existing E-Commerce recommendation system

With the development of E-Commerce, it's harder and harder for consumers to find the product they want, and the recommendation systems are applied more and more widely. A recommendation system includes user model, the recommended model and recommendation algorithm. The improvement considered in this paper mainly refer to the recommendation algorithm. Limited resource, data valid time and cold start problems are not well considered in existing E-Commerce recommendation system. According to the problems descripted above, a algorithm based on limited resource and a solution to cold start problem are proposed.