Search

Social recommender systems (SRS) aim to study how social relations influence users' choices and how to use them for better learning users embeddings. However, the diversity of social relationships, which is instructive to the propagation of social influence, has been rarely explored. In this paper, we propose a graph convolutional network based representation learning method, namely multi-perspective social recommendation (MPSR), to construct hierarchical user preferences and assign friends' influences with different levels of trust from varying perspectives. We further utilize the attributes of items to partition and excavate users' explicit preferences and employ complementary perspective modeling to learn implicit preferences of users. To measure the trust degree of friends from different perspectives, the statistical information of users' historical behavior is utilized to construct multi-perspective social networks. Experimental results on two public datasets of Yelp and Ciao demonstrate that the MPSR significantly outperforms the state-of-the-art methods. Further detailed analysis verifies the importance of mining explicit characteristics of users and the necessity for diverse social relationships, which show the rationality and effectiveness of the proposed model. The source Python code will be available upon request. [ABSTRACT FROM AUTHOR]

As an important task of Graph Neural Networks (GNN), graph classification has received increasing attention, as it can be widely used in numerous fields, such as protein prediction and community prediction. The GNN models for graph classification usually require aggregating the structure and feature information of an input graph into a hidden representation vector. Such a technique can be generally referred as graph pooling, which intends to maximize the removal of noise while minimize the loss of important features. Therefore, there exists a trade-off between the noise reduction and information maximization that needs to be balanced for better performance. However, the existing pooling-based GNNs for graph classification depends on the manual setting of the graph-pooling degree, making it difficult to balance the trade-off. To this end, we propose a Feature Pyramid-based Graph Convolutional Neural network for Graph Classification (FPGCN-GC), which constructs multi-scale hierarchical information fusion to reduce information loss, and achieves an adaptive feature fusion through a learnable weighted residual connection and self-attention mechanism. The superior performance of the proposed method is verified on multiple graph classification datasets, thus illustrating the effectiveness and superiority of FPGCN-GC. In addition, we visualize FPGCN-GC by T-SNE to further analyze the underlying reason for its effectiveness. [ABSTRACT FROM AUTHOR]