Incorporating auxiliary information in betweenness measure for input-output networks

The betweenness centrality plays an important role in input–output analysis. Existing betweenness measures are mostly defined based on the intermediate flow network, without accounting for node-level information which can be of great value in some applications. Here we propose a novel betweenness centrality measure that incorporates available node-specific auxiliary information for weighted, directed networks with a specific focus on input–output network analysis. The proposed measure is defined upon strongest paths reflecting the pull effects of sectors in an economy, which distinguishes itself from many other classical versions. The search of strongest paths is done by the Dijkstra’s algorithm. The proposed betweenness integrates the structural information of a network and auxiliary information which may come from sources beyond the network. Through two simulation experiments and applications to the 2018 national input–output network of China, we demonstrate the importance and effectiveness of incorporating auxiliary information relevant to the research objectives for betweenness computation and node centrality assessment. The implementation is publicly available in an R package ionet.

  title = {Incoporating Auxiliary Information in Betweenness Measure for Input-Output Networks},
  author = {Xiao, Shiying and Yan, Jun and Zhang, Panpan},
  journal = {Physica A: Statistical Mechanics and its Applications},
  volume = 607,
  pages = 128200,
  year = 2022,
  publisher = {Elsevier}