Research On Movie Recommendation Algorithm Based On Convolutional Neural Network And Recurrent Neural Network

Today’s era is the era of big data information,and people’s living standards have also increased.They want to spend less time getting the information that is most useful to them.And how to mine people’s potential behaviors from massive data and make accurate recommendations for them is the problem at this stage.The recommendation system came into being.Many of the products we use use a recommendation system that will predict and recommend products you may be interested in based on your previous series of behaviors.Collaborative filtering is the earliest recommendation algorithm.Usually,users who have similar hobbies with the designated user are first found,and the users with similar hobbies are evaluated on a certain product,and then the degree of preference of the designated user on the product is predicted.The disadvantage of traditional collaborative filtering is that it cannot combine the changes of time and user’s interests.This thesis combines Convolutional Neural Networks(CNN)and Recurrent Neural Network(RNN)to build a more flexible model of some of the user’s previous behaviors and the attributes of the product itself,so as to provide users with more accurate recommendations.The feature extraction of static attributes such as users and movies is processed using a convolutional neural network,and the two features obtained after the processing can be fitted to the score in any way.The information about the user’s rating is regarded as a sequence of data,which is processed by the recurrent neural network to obtain the user’s predicted rating value for the movie.The prediction results obtained after the two models are processed are combined,and the average value of the two prediction results is used as the final prediction score result.Select movies with higher final prediction ratings to recommend to users.Finally,the model verification is performed on the public movie dataset Movie Lens.Through a series of comparative experimental results,we can see that the recommendation algorithm based on convolutional neural network and recurrent neural network proposed in this thesis has good results on the corresponding evaluation indicators.In addition,the recommended result is to recommend a product list to the user.When recommending to users,the users will be clustered first.Through model training and prediction recommendation on the clustered data,it is found that the results obtained after clustering on the accuracy rate are about 25% higher than those without clustering.