base on Pytorch, Attention,CNN,GCN, contrasitve learning
We propose a novel user modeling approach for news recommendation, which models each user as a personalized heterogeneous graph built from user behaviors to better capture the fine grained behavior relatedness and solve the problem of data sparsity. Then we realize a self-supervised learning between the graph and an attentive model base on time sequence so as to improve the accuracy and robustness. Finally, the users’ feedback are inputted to correct and update the users’ interests.