Sequence recommendation algorithm based on Graph Neural Network and long short term prefer-ence
To solve the problems that sequence recommendation has obvious shortcomings in capturing us-ers'dynamic preferences,it is difficult to capture users'complex long-term dependencies.Therefore,a sequential recommendation algorithm integrating graph neural network and long-term and short-term prefer-ence is proposed.The algorithm mainly includes short-term preference learning and long-term preference learning.Firstly,the short-term preference learning is carried out based on the graph neural network.The graph neural network has a strong ability of fitting graph data.The graph neural network is used to capture the connection of the user's interest points and accurately generate the short-term preference representa-tion.The historical long-term preference is global and has less fluctuation.Bidirectional LSTM is used for long-term preference interest learning to obtain the user's long-term preference representation.The experi-ment results show that the sequence recommendation algorithm integrating graph neural network and long-term and short-term preference is significantly better than other advanced sequence recommendation methods.
万方数据
维普
