Design And Implementation Of Movie Recommendation System Based On Graph Neural Network

With the construction of the modern information society,the density of information has reached a point where it is difficult to calculate,with different information flooding our eyes.This is especially true for all types of video.Today,thousands of movies are produced each year,and because of the epidemic that has been rampant in recent years,many people are used to having movies on demand from the internet,but with a lot of complicated and useless information mixed in,it is difficult for users to find the right movie for them to watch.As a result,it has become a priority for media companies to improve the user experience.Recommendation systems have been a hot topic of research in both academia and industry,but their rapid development comes with a number of challenges.For example,capturing too little information between users and items leads to poor recommendations,and the sparse matrix formed by a large amount of data leads to difficulties in cold starts.The advantages of graph neural networks for recommender systems have been identified through research and analysis of graph neural network algorithms.After some improvements,the improved graph neural network algorithm is incorporated into the most widely used collaborative filtering in recommendation systems.The algorithm captures the higher-order connectivity between users and items and embeds it into user items,making the embedded vector more informative.Finally,the algorithm is compared with several mainstream recommendation algorithms on the Movie Lens dataset and is found to be more effective.A real-time recommendation algorithm incorporating movie chronological weights is also designed,which incorporates movie chronological weights into the movie similarity calculation and adds a discount factor based on user ratings to make the recommended results more in line with user expectations.Based on the above algorithm,a movie recommendation system was built using the mainstream Spring application development framework,as well as Spark big data processing tools,Mong DB and other mainstream databases.The system reads user behaviour data and calculates user ratings in conjunction with local datasets to obtain a series of recommendations and present them to the user.The system can make realtime recommendations based on the user's real-time needs,and can also make more accurate recommendations to the user based on more precise offline recommendation algorithms.Finally,the system is implemented and tested to verify that the system does have good recommendation results.