The Research On Movie Recommendation Algorithm Based On Graph Convolutional Neural Network

In the era of big data,"information overload" is a major problem that plagues users to obtain valuable information from the Internet quickly and efficiently.As an effective method to solve this problem,recommender system has become the focus of attention in academia and industry and has been widely used.However,due to the increase in data size and data diversity,traditional recommendation algorithms have reached the performance bottleneck.At the same time,with the widespread application of deep learning technologies in the fields of image,video,speech,etc.,deep learning-based recommendation algorithms have become research hot spot.Therefore,this paper proposes a recommendation algorithm model based on graph convolutional neural networks,also,this paper utilizes local sampling and clustering algorithms to improve and optimize graph convolutional networks.This paper mainly studies the recommendation algorithm based on graph convolutional neural network.Graph convolutional network is a deep learning model for graph structured data.It has been widely used due to its powerful representation capabilities.The recommendation task can be regarded as a link prediction problem in graph structured data.Compared with traditional recommendation algorithms,graph convolutional network-based recommendation algorithms can better learn the deep features that characterize data objects,thereby improving the quality of recommendation results.However,with the rapid growth of data scale,existing graph convolutional networks lack the processing capacity for large-scale data,and the generalization ability for unknown nodes is also relatively weak.In view of this,this paper improves and innovates on the basis of the existing technology,and proposes a local sampling algorithm based on random walks to sample user nodes,and trains the model by generating local computing graphs,thereby improving the performance of graph convolutional networks.Processing efficiency of large-scale graph data.And this paper also introduces the node clustering algorithm to cluster user nodes in the model prediction stage,and uses the clustering results to increase the relationship between the newly added nodes and known nodes,thereby improving the generalization ability of the network model for unknown nodes.This paper uses Movie Lens standard dataset to conduct comparative experiments on the proposed recommendation algorithm and the corresponding improved algorithms.It uses the Root Mean Square Error,Precision and Recall to evaluate the recommendation algorithm’ results.The experimental results fully verify the effectiveness of the proposed graph-based convolutional network-based recommendation algorithm and the corresponding improvement methods to improve the performance of the original algorithm model.