Research On Movie Recommendation Model Based On Convolutional Neural Network

With the improvement of material living standards,spiritual entertainment has become more and more important.As a relatively popular spiritual entertainment project,film and television entertainment is gradually receiving people’s attention.However,in recent years,the film industry has developed rapidly,and the production of movies has also increased year by year.How to quickly and accurately find the movies that users like in a large amount of movie data has become an urgent problem.Calculating the movie that the user may be interested in based on the viewing information,browsing information,and rating information of the registered user is the key to solving this problem.After reviewing and summarizing a large number of documents and materials,the paper proposes a movie recommendation model,which includes the preprocessing of movie data sets,the user’s prediction of the movie scores,and the prediction scoring processing.In the study of movie scoring prediction,according to the relevant theories of deep learning,the principle of convolutional neural network(CNN)is applied to the movie scoring prediction module.The convolutional neural network is trained based on registration information of registered users,movie related information,and registered users’ scores on movies.In the experiment,the performance of the network on the training set is improved by adjusting hyperparameters;in order to ease the convolutional neural network In the course of calculation and network training,the"overfitting" phenomenon occurs,and the appropriate Dropout value is found to reduce the overfitting problem.In the process of optimizing the movie recommendation model,the data set is first preprocessed and then processed.The used datasets use K-Means clustering algorithm for further clustering calculations.The users are clustered by the movie scores.To a certain extent,the similarity calculation in the collaborative filtering algorithm is referenced,and the accuracy of the score prediction is greatly improved..The paper applies Movielens-1M datasets which containing movie ratings to the movie recommendation model.The experimental results show that the movie recommendation model established in the thesis can accurately recommend the user’s favorite movies,which has certain reference significance in the field of movie recommendation.