Polarity-based graph neural network for sign prediction in signed bipartite graphs

Abstract As a fundamental data structure, graphs are ubiquitous in various applications. Among all types of graphs, signed bipartite contain complex structures with positive and negative links as well settings, on which conventional graph analysis algorithms no longer applicable. Previous works mainly focus unipartite or unsigned separately. Several models proposed for applications the by utilizing heuristic structural information. However, these methods have limited capability to fully capture information hidden such graphs. In this paper, we propose first neural network namely Polarity-based Graph Convolutional Network (PbGCN), sign prediction task help balance theory. We introduce novel polarity attribute based construct one-mode projection allow GNNs aggregate between same type nodes. Extensive experiments five datasets demonstrate effectiveness our techniques.