Abstract: Collaborative filtering is the most successful technology for building recommendation systems. Unfortunately,the efficiency of these methods decline linearly with the number of users and items .To address these limitations, a high efficient personalization recommendation algorithm is presented, which includes two phases: dimensionality reduction and item-based recommendation methods. This algorithm reduces the computation consumption based on enhancing the accuracy, etc. It may solve questions well such as sparsity, scalability. It can create accurate personalization recommendation quickly.