Neighborhood discovery via augmented network community structure

Abstract The geospatial characteristics of transportation networks structurally constrain their features, and as a result, analysis methods designed for social typically fail to capture useful or make informative comparisons. In the case road networks, natural constraints on edge distribution weaken ability standard community detection algorithms find clusters nodes that align with neighborhood extents. We show by adding weights based similarity localized subgraph we can apply modularity-based uncover improved shapes use local network allows feature be completed in linear time, thus making approach expandable very large networks. demonstrate this technique an application central Tokyo.