道路网络表示学习：基于双图的方法（英文版）

英文标题：Road Network Representation Learning: A Dual Graph based Approach中文摘要：本文提出了一个基于超图的图神经网络模型 HyperRoad 用于道路网络表示学习。此模型通过构建超图捕捉道路网络中的高阶联系，并利用图重建、超图重建和超边分类任务去学习结构信息和长距离联系。英文摘要：Road network is a critical infrastructure powering many applicationsincluding transportation, mobility and logistics in real life. To leverage theinput of a road network across these different applications, it is necessary tolearn the representations of the roads in the form of vectors, which is named\emph{road network repr