Session-based Recommender System

论文笔记-A Survey on Session-based Recommender Systems

https://blog.csdn.net/qq_20965753/article/details/90234568

一些方法

https://blog.csdn.net/like_red/article/details/83416323

用GNN做推荐系统的开山之作 Session-based recommendation with graph neural networks.

https://blog.csdn.net/qq_40210472/article/details/89839803

https://www.kesci.com/home/project/5d18d91a1951a9002c864fc4

Graph Contextualized Self-Attention Network for Session-based Recommendation

https://www.jianshu.com/p/a73972a8fe39

优点：

缺点：

The logical relationship between these three parts（feature-rich and feedback-rich GNN recomandations, GNN word embedding model, anti-refelection image enhancement） is weak. It is more appropriate for each part to corresponds to a separate project.

The graphical convolution operation in section”GNN word embedding model” is too plain so that the representation capacity may be weak. Many fancy and advanceced graphical operations can take place of it .

A key reference “Shu Wu et al(2019), Session-based recommendation with graph neural networks” is missed.

提问：

是否将其他的特征，如时间、季节、地点、流行趋势等也囊括在特征丰富上

你的计划中没有考虑用户的因素，能否考虑进去

你的计划假设每个项目是相关的