Research on a Session Recommendation Algorithm Based on Self Supervised Learning
Session recommendation aims to provide reasonable recommendations based on a continuous sequence of interactions.The existing session recommendation algorithms are mainly based on recurrent neural networks and have achieved good results compared to traditional methods.However,these methods only consider the sequence information of the session and ignore the transfer information of items in the session.To solve this problem,this paper proposes a session recommendation algorithm S-SRGNN based on graph neural networks,which models the transfer information between items through graph neural networks and captures the complex relationships between items.In addition,combining self supervised learning on local and global graphs enhances session modeling by maximizing the interaction information between different view session representations.Experiments on three publicly available datasets have shown that S-SRGNN achieves better performance in session recommendation tasks.
computer system architectureneural networksconversation recommendationself supervised learning
万方数据
维普
