The extended abstract: Finding Neighborhoods in Santiago, Chile: A Graph Neural Network Approach to Urban Cluster Detection has been accepted in the Urban Complex Systems Workshop that is part of the Complex Systems 2021 Conference.
This paper makes two contributions to the literature on urban cluster detection. First, it applies a novel graph neural network approach to uncover urban structure. Second, whereas most of the previous literature’s clustering is based on ethnic and socioeconomic indicators, we include two unexplored sources of data: namely, the last names of residents and how neighborhoods are perceived from an aesthetic point of view. We apply the proposed method to detect urban clusters in Santiago, Chile.
DOI of the document: coming soon.