Metapath Pattern Search Algorithm Based on Path Ranking
To address the issues of poor flexibility and completeness in existing heterogeneous graph neural network models based on metapath pattern modeling,and to further improve the efficiency of heter-ogeneous graph neural network models,a Forward-Reverse Path Ranking algorithm is proposed based on a detailed analysis of existing model principles.The algorithm leverages the superior performance of graph neural networks as a deep learning-based graph representation technique in extracting graph data features,and can automatically identify and rank metapath patterns in heterogeneous graphs.Through experiments,comparative analyses are conducted with existing models capable of extracting metapath patterns from heterogeneous graphs on commonly used open-source heterogeneous graph datasets,verifying the effective-ness of the algorithm.
万方数据
维普
