Presentation Information
[1Yin-B-58]Community Detection in Corporate Network Hypergraph
〇Yuri Murayama1, Kiyoshi Izumi1 (1. Univ. of Tokyo)
Keywords:
Community detection,Hypergraph,Corporate network analysis
In corporate research, transaction data has traditionally been utilized to understand the realities of inter-firm relationships and community structures. However, extracting relationships that cannot be fully captured by such structured data remains a challenge. In this study, we constructed a hypergraph by extracting relationships between companies and industry associations from InfluenceMap, a database that tracks the extent to which these entities support or obstruct climate change policies. We performed community detection using the hyperedge percolation method proposed by Kovács et al. (2025). The results showed that when using the approach based on large hyperedges, the hypergraph was returned as a single community due to the presence of large hyperedges (i.e., industry associations) containing up to 48 nodes. In contrast, the approach based on small hyperedges successfully partitioned the hypergraph into two distinct communities.