Design And Implementation Of Movie Recommendation Algorithm Based On Deep Neural Network

With the development of the emerging film industry,the global film market has maintained a basic trend of steady growth.The Internet also has extremely large film resources.The recommendation system has become a way for people to quickly find interesting movies.In recent years,deep learning has been widely used in image recognition,natural language processing,etc.At the same time,some recent studies have also shown that deep learning is reliable in recommendation systems.In this context,this paper studies the basic theory of personalized recommendation system,and uses the related techniques of deep learning in the recommended field to propose a personalized recommendation algorithm based on deep neural network,called Deep Neural Matrix Factorization.First,Th is model use a new way to extract potential vectors for users and projects.In this paper,the memorization and the generalization are defined to extract the features of the input vector of the user and the project,and then sneak into a new dense space.The memorization is essentially a shallow linear unit,and the generalization is actually a deep nonlinear neural network.It has been proved in experiments that both the memorization and the generalization have significant effects on the performance of the model.Then,Train different potential vectors in multiple ways.For the user or project vector generated by the memorization,this paper is called the memory vector.The generalized matrix decomposition model is used to multiply the user and the project memory vector by element to describe the potential interaction characteristics between the user and the project.Linear transformation,so the effective memory information of the memory vector is saved.For the generalization to generate the user or project vector,this article is called the inductive vector,which connects the user and the project's respective features,and then uses the standard multi-layer perceptron model to learn the interaction between the user and the project's underlying features.Finally,the two models are combined and the joint training method is used to model the implicit feedback of the user-project and complete the recommendation.Finally,This paper has carried out extensive experiments on the MovieLens dataset.The experimental results show that the proposed method outperforms the most advanced recommendation algorithms.The paper also evaluates the performance of the modified model in different experimental environments.