What is it about?
Polarization is a prevailing concern in algorithmic recommendation, but difficult to conceptualize and measure. In this article we put social network users in opinion spaces, we simulate network evolution via friend recommendation, and mesure polarization as property of the distribution of the opinion of users in time.
Featured Image
Photo by Todd Diemer on Unsplash
Why is it important?
Traditionally, polarization has been examined by looking at how algorithms connect or disconnect communities through recommendations (data driven approaches using social network data), or how they tend to cluster users around different opinions (simulation driven approaches, using synthetic data due to the unobservability of opinions). In this article we bridge both approaches, starting simulation from real social network in which users are embedded in opinions spaces in which dimensions are indicators of attitudes (from most opposed to most favorable) towards different issues of public debate.
Perspectives
By bridging social network analysis and opinion dynamics with our ideological spaces, we open up a fruitful lane of research in how algorithms affect social phenomena online. This is also relevant in other related phenomena, such as news media consumption, or in the study of social psychology mechanisms and political competition online.
Pedro Ramaciotti Morales
Sciences Po
Read the Original
This page is a summary of: Auditing the Effect of Social Network Recommendations on Polarization in Geometrical Ideological Spaces, September 2021, ACM (Association for Computing Machinery),
DOI: 10.1145/3460231.3478851.
You can read the full text:
Contributors
The following have contributed to this page
