Research on Movie Recommendation Model Based on LSTM and CNN

In order to further improve the accuracy of movie recommendation, while considering the characteristics of user data and movie data, this paper studies and proposes a combined recommendation model of LSTM and CNN. The model uses LSTM to capture the context dependency of user ratings data, and at the same time extracts the local relevant features of the movie title with CNN, and then fuse each feature to calculate the predicted ratings, through model training and optimization, the movie recommendation to the user is finally obtained according to the predicted ratings. The MovieLens data set is used to verify the effectiveness of the model, and the results show that compared with the traditional recommendation model and other recommendation models based on deep learning, the combined recommendation model of LSTM and CNN proposed in this paper have a MSE loss reduction of 4.4%~18.7% and a MAE loss reduction of 3.0%~52.2%.