Complex network science has been increasingly leveraged as a tool for studying public transport networks (PTNs), yet key challenges that hinder the further development of this research field still remain. This study makes a twofold contribution to address key challenges related to network representation. First, we propose a multiscale representation of PTNs which is consequential for constructing the topological representations needed for complex network analysis. This can help to improve the consistency and reproducibility of the research on PTN analysis and modeling. Second, we incorporate domain knowledge by enriching several topological representations of PTNs with service properties. Time-dependent weights that are derived from the data on scheduled services are added to the topological networks. In addition, we also demonstrate the implementation of the proposed methodology based on the General Transit Feed Specification (GTFS) data. The Amsterdam tram network is used in the case study to demonstrate the proposed methodology. The multiscale representation of this network is established first, followed by the construction of plain (unweighted) and enriched (weighted) topological networks. Our analysis further shows that the centrality measures computed in the enriched topological representations are capable of quantitatively characterizing the dynamic service properties of public transport systems from different aspects. This study paves the way to a range of applications that can make complex network science more relevant for public transport research and planning.

CICTP