In the current traditional collaborative filtering recommendation algorithm, there is a lack of mining potential features in the data, and the impact of the distance between the scoring time and the recommendation time on the recommendation is not considered. To solve this problem, this paper proposes a movie recommendation model based on deep learning and time factor. The experimental results in the MovieLens 1M data set show that after the introduction of deep learning and time factors, the MAE value is reduced by about 0.02-0.04, and the accuracy of the recommendation is improved.