Multi-turn opinion dynamics simulation

Nitish Talekar, Arnav Jhala

Keywords: opinion dynamics, conversational simulation

TL;DR: Investigating opinion dynamics in a multi-turn multi-agent conversational setting

Abstract: This paper describes a simulation framework for multi-turn multi-agent conversational agents with rich personality models. Consistency and opinion dynamics are tracked across different personality combinations for long conversations. These conversations are on discussions related to facts, debatable assertions, and controversial topics to capture the range of opinion change possibilities. Our experimental setup and results seek to provide insight into design and setup of agents that can be part of populations with a wide range of personalities and evolving narrative interactions based on opinion dynamics.

Paper Track: Technical paper

Submission Number: 49