Urban Grammar AI
research project
This month has been a big on in terms of academic outputs. The first one has been the conceptual paper putting forth many of the ideas that underpin much of the Urban Grammar project. Here are the full coordinates:
Arribas-Bel, D., & Fleischmann, M. (2022). “Understanding (urban) spaces through form and function”. Habitat International, 128, 102641. https://doi.org/10.1016/j.habitatint.2022.102641
https://doi.org/10.1016/j.habitatint.2022.102641
Published version (Open Access)
Code repository@Github
Github
Intro thread@Twitter
Twitter
@article{ARRIBASBEL2022102641, title = {Spatial Signatures - Understanding (urban) spaces through form and function}, journal = {Habitat International}, volume = {128}, pages = {102641}, year = {2022}, issn = {0197-3975}, doi = {https://doi.org/10.1016/j.habitatint.2022.102641}, url = {https://www.sciencedirect.com/science/article/pii/S0197397522001382}, author = {Daniel Arribas-Bel and Martin Fleischmann}, keywords = {Geographic data science, Urban form, Urban function}, abstract = {This paper presents the notion of spatial signatures as a characterisation of space based on form and function designed to understand urban environments. The spatial configuration of the dif-ferent components of cities is relevant for at least two main reasons. On the one hand, it encodes many aspects of the phenomena that created such an arrangement in the first place. On the other, once in place, this arrangement of urban form and function underpins many outcomes, from economic productivity to environmental sustainability. Our approach unfolds in three main stages. First, we propose a new spatial unit –the Enclosed Tessellation (ET) cell– to delineate space in a way that is exhaustive and matches the underlying processes at which urban form and function operate. Second, we propose to attach a large variety of form and function-based characters to ET cells to describe each of these units. Third, to build spatial signatures, information on ET cells can be clustered using unsupervised learning techniques. This process results in a theory-informed, data-driven typology of space that follows form and function. We demonstrate the flexibility of the approach to a variety of data landscapes and cultural backgrounds by providing five illustrations of spatial signatures for five cities across five continents. These showcases demonstrate the ability to successfully differentiate areas of a city that were built at different points in time and under different technological regimes, but also highlight broader comparisons about the nature of urban fabric in different regions of the world. Our contribution resides in leveraging modern data, tech-nology and methods to propose a detailed, consistent and scalable methodology that characterises urban form and function. The spatial signatures can be used across academic disciplines and by a variety of practitioners and policymakers supporting initiatives such as the Sustainable Development Goals.} }