The common data sparsity in recommendation systems makes the nearest neighbor search is not accurate and lets the search results of the nearest neighbor is too small. This will affect the recommended quality and accuracy of the recommendation system, moreover it is difficult to solve in the traditional collaborative filtering recommendation. To overcome the difficulty of data sparsity in recommendation systems, a novel collaborative filtering algorithm is presented which is based on the combination of trust relationship between users and the similarity of scores of the projects. This algorithm constructs the trust relationship among users by using a directed network graph, which can make up the defect that the user's relationship cannot be accurately measured by the user's similarity. The experimental results show that the proposed algorithm can improve the quality and accuracy of the recommendation system.