Social Relation Reasoning Based on Triangular Constraints

Social networks are essentially in a graph structure where persons act as nodes and the edges connecting denote social relations. The prediction of relations, therefore, relies on context graphs to model higher-order constraints among which has not been exploited sufficiently by previous works, however. In this paper, we formulate paradigm relations into triangular relational closed-loop structures, i.e., constraints, further introduce reasoning attention network (TRGAT). Our TRGAT employs mechanism aggregate features with graph, thereby exploiting reason iteratively. Besides, acquire better feature representations persons, node contrastive learning relation reasoning. Experimental results show that our method outperforms existing approaches significantly, higher accuracy consistency generating graphs.