In this paper, we utilize the mechanism of simulating human nervous system in deep learning for movie data. Feature extraction is performed automatically and unstructured data is processed to improve the model recommendation effect. Through the loss function of the output layer, iterative optimization is carried out with the help of the back propagation algorithm, which passes the prediction error from the output layer to the input layer in reverse, and updates the network parameters of each layer in turn. The analysis found that the accuracy of the movie recommendation algorithm based on deep learning increases faster, to the point that after 160 rounds the accuracy leveled off, compared with the optimal algorithm improved by 1.4%. The movie recommendation algorithm based on deep learning can better meet the user’s personalized needs and recommend the movie that best meets the user’s requirements.