Our overall goal is to study polarization in a framework that jointly considers definitions of polarization based on opinion and the social structure. The way to do this is to link two networks, one derived from each approach.
Our first goal is therefore to develop an opinion network representation that can be used to study polarization. We then propose a new method for extracting opinion-based networks from word embeddings trained on media and social media data.
Our second goal is to link social networks and opinion networks into multilayered representations that we call People-Opinion Networks (PONs). The social networks will be developed from interactions between users on social media.
Our third goal is to use our multi-layered PONs model in a comparative survey of polarization in Switzerland and South Africa. These investigations will use the full potential of our multilayered PONs representation to investigate how polarization arises from the potential of social and opinion structures.
Our survey focuses on the migration debate in Switzerland and South Africa, studying opinion and interaction around two specific events in each country between 2010 and 2020. This research collaboration between Switzerland and South Africa aims to make an original contribution to the social sciences by developing a new method to study polarization and the dynamics of social change more generally.
This multidisciplinary research collaboration will bring insight and methodological development on how opinions get polarized in the media and social media in Switzerland and Southern Africa. We propose a novel method, using Natural Language Processing and grounded in social psychological theory, for studying how social networks develop around opinions and become polarized. Our overall objective is to study polarization in a framework that jointly considers opinion-based and social structure definitions of polarization. The way to do this is to link two networks, one derived from each approach. While the social-structure network is straightforward to construct, the opinion-based network is not clear at all, and we develop an opinion-based network extracted from word embeddings.Our first aim and deliverable is to develop a network representation of social opinion that can be used to study polarization. We propose a novel method for extracting opinion-based networks from word embeddings trained on media and social media data.Our second aim and deliverable is to link social networks and opinion networks in multilayer representations that we call People-Opinion Networks (PONs). The social networks will be developed from interactions between agents on social media. The opinion networks that we extract from word embeddings will also include agent names, and we will use established methods to link the two structures, creating multilayer networks.Our third aim is to use our multilayer PONs model in a comparative investigation of polarization in Switzerland and South Africa. We will perform cross-sectional and longitudinal analysis of polarization in both media and social media. These investigations will be informed by the view (from social psychology) that polarization arises from intergroup comparisons, and that the emergence of polarization depends on the spread of opinions between groups. These investigations will exploit the full potential of our multilayer PONs representation for investigating how polarization arises from potentialities in both social and opinion structuresOur investigation will focus on debate around migration in Switzerland and South Africa. We will study opinion and interaction around two polarizing events in each country between 2010 and 2020. For each of the four instances of polarization we will sample data from three time periods, three months immediately before, during, and immediately after the polarizing event. In each of these twelve samples periods we will collect two layers of data: (1) media data from a broad spectrum of newspapers and news sources; and (2) social media data from Key Opinion Makers (journalists, politicians, etc.), their followers, and a sample of other users extracted by keywords. ?The research collaboration thus proposes to make an original contribution to science, developing a novel method to study social polarization and social change dynamics more generally. In the process, we will also be training the next generation of South African social science researchers with links to leading labs groups in Europe and skills in cutting edge data science methods.
Last updated:20.06.2022
