Transformation of Local Communities from Neighborhoods to Urban Commons in the Production of Social Representations of Space
Published in Computational Science and Its Applications – ICCSA 2024, 2024
Aleksandr Antonov, Galina Gornova, Georgii Kontsevik, Leonid Turkov, Vladimir Vorona, Sergey Mityagin (2024). Transformation of Local Communities from Neighborhoods to Urban Commons in the Production of Social Representations of Space. Computational Science and Its Applications – ICCSA 2024, 436-447.
Overview
The study explores how digital communication shapes hyperlocal and translocal urban identities. It combines topic modelling, place-name and keyword extraction, and spatial knowledge graphs to identify connections between territories and hotspots of community activity. The semantic content of these connections supports the study of how residents collectively represent urban space.
Method and implementation
SOIKA is the accompanying Python library for enriching digital city models with information extracted from residents’ textual digital footprints. Its workflow collects social-media text, extracts and geocodes mentions of urban places, and links the text to a spatial model built with OpenStreetMap data. Keywords and semantic relationships are then used to construct a spatial-semantic graph. The library also provides text and emotion classification, topic modelling and regional activity analysis.
Processing scheme from the SOIKA repository, with original Russian labels. Click to view full size.
Graph example
Example spatial-semantic hypergraph from the repository README. Place names and extracted terms are shown in their original language.
Paper and code
Address extraction model: flair-ner-addresses-extractor on Hugging Face extracts addresses from Russian-language texts.