Heterogeneous graph neural network node embedding model based on self-attention mechanism

Abstract

This study presents MultiHeCo, a heterogeneous graph comparison model that incorporates a multi-head self-attention mechanism to improve the ability of graph neural networks to process complex interactions. For the complex structure of heterogeneous information networks, the MultiHeCo self-attention mechanism is able to process and capture complex information in parallel. As this mechanism is suitable for capturing the characteristics of long-distance dependencies, we apply it to meta-path view generation to identify key features and relationships. In terms of specific work, we introduce the multi-head self-attention mechanism in heterogeneous graph neural networks for the first time, propose a new meta-path view node embedding structure, and validate the model's excellent performance in several downstream tasks and outperforms some semi-supervised learning models through extensive experiments on two public datasets.