HeMGNN: Heterogeneous Network Embedding Based on a Mixed Graph Neural Network

Network embedding is an effective way to realize the quantitative analysis of large-scale networks. However, mainstream network models are limited by manually pre-set metapaths, which leads unstable performance model. At same time, information from homogeneous neighbors mostly focused in encoding target node, while ignoring role heterogeneous node embedding. This paper proposes a new model, HeMGNN, for The framework HeMGNN model divided into two modules: metapath subgraph extraction module and mixing module. In module, automatically generates filters out metapaths related domain mining tasks, so as effectively avoid excessive dependence on artificial prior knowledge. integrates when learning nodes. makes vectors generated according contain more abundant topological semantic provided Rich achieve good downstream tasks. experimental results show that, compared baseline models, average classification clustering has improved up 0.3141 0.2235, respectively.