Dynamic hypergraph neural networks based on key hyperedges

The technique of graph/network embedding can help computer to efficiently analyze and process the complex graph data via vector operations. Graph Neural Network, which aggre-gates the topological information of the neighbourhoods of each node in a graph to imple-ment graph/network embedding, has attracted wide attention. With the explosive growth of information, large amounts of data need to be expressed in the form of hypergraphs. As a result, the hypergraph neural networks arise at the historic moment. However, most cur-rent work is based on static hypergraph structure, making it hard to effectively transmit information. To address this problem, Dynamic Hypergraph Neural Networks based on Key Hyperedges (DHKH) model is proposed in this paper. Considering that the graph structure data in the real world is not uniformly distributed both semantically and struc-turally, we define the key hyperedge as the subgraph composed of a small number of key nodes and related edges in a graph. The key hyperedge can capture the key high-order structure information, which is able to enhance global topology expression. With the sup-porting of hyperedge and key hyperedge, DHKH can aggregate the high-order information and key information. In our experiments, DHKH shows good performance on multiple datasets, especially on the SZ dataset and LOS dataset which have inherently some key structures.(c) 2022 Elsevier Inc. All rights reserved.

Automated summary

The technique of graph/network embedding is used to analyze and process complex graph data efficiently, and the Dynamic Hypergraph Neural Networks based on Key Hyperedges (DHKH) model is proposed to address the information transmission issue in static hypergraph structure.