Geographia Technica, Vol 14, Issue no.1/2019, pp. 82-91
ROAD NETWORK BASED COMMUNITY DETECTION. CASE STUDY FOR AN EASTERN REGION OF AUSTRO-HUNGARIAN MONARCHY
Zsolt MAGYARI-SÁSKA (Gheorgheni, Romania)
ABSTRACT: The aim of the study was to find ways of measuring the relationship between clusters resulted on road networks topology and regional communities, before modern transport vehicles had widely spread. We have chosen as investigation area an eastern region of Austro-Hungarian Monarchy because at the beginning at the XX. century it had a modern transportation infrastructure for its time and a multicultural environment where different local or regional communities were present. Using data of that period we constructed a network model, attaching the populations’ mother tongue as attribute data for settlements. After applying community detection algorithms considering only road network topology and later also the attribute data, we tried to compare the resulted clusters. We observed that in some cases mother tongue-based communities form sub-clusters for solely road network base clusters, while in other cases mother tongue-based communities restructure the road network based clusters. To quantify the differences, we identified three similarity/dissimilarity comparing methods, one based on clusters’ spatial extends and two on statistical approaches.
Keywords: Heterogeneous network, Modeling reality with graphs, Clustering, R, Spatial relation, Comparing communities
Full article here