Application of SVD technology in video recommendation system

Abstract

The most direct access to evaluate what kinds of topics are valuable for video producers, and bring them inspiration is to seek subjects which specific groups concern currently. We can obtain massive user information from social networking platforms, large video sites and search engines, and then exploit the data to produce more practical works with the combination of business requirements. In views of the existing disadvantages of inferior scalability, sparsity problem and huge volume test data, the application of Singular Value Decomposition Method(SVD) actualize the unknown prediction score function of set of tests. The simulation results show that scalability, sparsity and omputational efficiency improved effectively.