Generating co-evolutionary polarized opinion networks

Abstract: In this thesis we present a co-evolutionary opinion network model, which aim to efficiently represent an online social network, with each person represented as a node attributed with an opinion, and the relations between people by edges. For this, we needed to classify the existing models through the prism of co-evolution, i.e. how the topology and the state of the network interact between each other, so one can find a model of classification for network models. This model will be compared under certain aspects to empirical data, as well as previous works. We show the emergence of a polarization of the opinions in the network, which appears only under certain conditions : a strong homophily between nodes, as well as a co-evolutionary behavior, i.e. a strong interaction between the topology and the state, of the model.