Quantification of echo Chambers: a Methodological Framework considering Multi-Party Systems

The possibility of distributing user-generated content through online social networks (OSNs) has had liberating effects on society, with prominent examples such as the Arab Spring. Yet, since then, many dark sides of OSNs have been brought up. An example is the echo chambers phenomenon. Theory suggests that cognitive dissonance causes individuals to associate themselves with groups of like-minded individuals that are only exposed to content that confirms their previously held beliefs. In turn, deliberation amongst segregated groups increases social extremism and causes polarization, rather than moderation. Previous research endeavors to identify echo chambers in OSNs have scarcely investigated the community structures of a network on a fine granular level, specifically in the context of multi-party systems. To contribute to the scientific body of knowledge, we propose a framework that summarizes existing work and outlines a way for future research to fill this void. We further propose a new way to measure homophily in multi-party systems based on the cosine similarity between users. We evaluate our framework through real world data and find that members of the political right experience the least amount of crosscutting communication and the highest degrees of homophily.