Expanding Social Polarization Models by Incorporating Belief Credibility
Modern technology makes information and global communication accessible in an instant, but can lead to echo chambers and other forms of polarization. By modeling polarization, techniques to identify it can be developed and used to promote discussion of polarized topics. This project expands the polarization model developed by Mário S. Alvim, Sophia Knight, and Frank Valencia by developing a simulation that can more accurately represent the transference of beliefs. In real-life interactions, multiple viewpoints are present within a discussion. Improvements to the model are mainly the addition of two major components. The first component is a credibility table that modifies the influence, or “credibility” of each belief based on its relation to the agent’s other beliefs. The second is through refining the influence function to incorporate the agent’s confirmation bias and resistance to opposing views based on their current belief. Through the addition of these components, the simulation can model more nuanced social interactions where multiple views and beliefs are involved. It is also capable of representing uninfluenceable agents, such as a broadcaster or politician. These nuances allow for a gradual change in a related view to eventually overturn the existing status quo. The simulation has many parameters that affect the update, most notably the influence table, initial agent states, and the value used to scale the change in belief each update. Possible future work could include defining more intricate or larger scenarios, or using different influence tables to describe other systems.