The existing methods using social information can alleviate the data sparsity issue in collaborative filtering recommendation,but they do not fully tap the complex and diverse user relationships,so it is difficult to ...The existing methods using social information can alleviate the data sparsity issue in collaborative filtering recommendation,but they do not fully tap the complex and diverse user relationships,so it is difficult to obtain an accurate modeling representation of the user.To solve this,we propose a multirelationship aware personalized recommendation(MrAPR)model,which aggregates the various relationships between social users from two aspects of the user’s personal information and interaction sequence.Based on the comprehensive and accurate relationship graphs established,the graph neural network and attention network are used to adaptively distinguish the importance of different relationships and improve the aggregation reliability of multiple relationships.The MrAPR model better describes the characteristics of user interest and can be compatible with the existing sequence recommendation methods.The experimental results on two real-world datasets clearly show the effectiveness of the MrAPR model.展开更多
基金supported by the National Key R&D Program of China under Grant No.2020YFB1710200the National Natural Science Foundation of China under Grant No.61872105 and No.62072136.
文摘The existing methods using social information can alleviate the data sparsity issue in collaborative filtering recommendation,but they do not fully tap the complex and diverse user relationships,so it is difficult to obtain an accurate modeling representation of the user.To solve this,we propose a multirelationship aware personalized recommendation(MrAPR)model,which aggregates the various relationships between social users from two aspects of the user’s personal information and interaction sequence.Based on the comprehensive and accurate relationship graphs established,the graph neural network and attention network are used to adaptively distinguish the importance of different relationships and improve the aggregation reliability of multiple relationships.The MrAPR model better describes the characteristics of user interest and can be compatible with the existing sequence recommendation methods.The experimental results on two real-world datasets clearly show the effectiveness of the MrAPR model.
