Segregation and polarisation are two pressing societal problems. Residential segregation exacerbates social inequality and hampers inter-group contact and trust. Political polarisation both breeds and is fuelled by echo-chambers in social media, which provide fertile grounds for the spreading of falsehoods and further extremization. Division lines in society running across geographic space (i.e. segregation) or opinion space (i.e. polarisation) can lead to group conflict and threaten the functioning of democracy. Currently, we simply do not know how these phenomena are interrelated in the societies we live in and when and where a vicious circle may arise in which segregation and polarization between groups may mutually reinforce each other. This project, therefore, aims to answer the following research question:
How, when and where is residential segregation related to political polarization in the Netherlands?
This project will use geographic data at a fine granular level mapping the whole Netherlands, for measuring different dimension of segregation. Multiple rounds of geo-coded national election results are used to measure different dimensions of polarization. Our geo-coded and time-stamped data is big (> 1.500.000 rows/examples) and thick (>1.000 columns/features). Our data has already been preprocessed and is ready to be analyzed. Together with the RSEs of the eScience Center, we aim to develop scalable machine-learning algorithms to predict where polarization is likely to occur and to detect geographical outliers with respect to segregation and polarisation.