Analyzing a Multilayer Comprehensive Passenger Transport Network Through Overlapping Community Detection

Abstract

To analyze comprehensive passenger transport systems, we abstracted five modes of passenger transportation to construct a multilayer comprehensive passenger transport network (MCPTN). The MCPTN is a heterogeneous network that combines high-speed railways, ordinary railways, expressways, national highways, and air passenger transport subnetworks. Travel efficiency is defined as the intra-layer edge weight, whereas transfer efficiency is defined as the inter-layer edge weight in the MCPTN. Owing to the large-scale and complex structure of a MCPTN, nodes often play a role in multiple communities within the network. Considering intra- and inter-layer relationships, we propose an overlapping community detection method and introduce the cross-layer cost into the calculation of link pair similarities. By comparing the density of the network community before and after introducing the cross-layer cost, we found that our method could identify overlapping communities more accurately. The results showed that the network was divided into fifty-three groups of overlapping communities, forming 344 overlapping node cities. Forty-eight influential node cities were found to maximize the MCPTN connectivity. Meanwhile, some influential node cities have imbalanced comprehensive transport systems, and the connectivity between the regions is weak. Finally, the weaknesses of the MCPTN are identified, and optimization suggestions are presented. We proposed an innovative method for detecting overlapping communities in multilayer heterogeneous networks. This study introduces new ideas to address interconnection issues in MCPTNs, and provides a theoretical basis for optimizing, designing, and maintaining comprehensive passenger transport systems.