[We’re pleased to welcome authors, Jesse Hoey of the University of Waterloo, Tobias Schröder of Potsdam University of Applied Sciences, Jonathan Morgan of Potsdam University of Applied Sciences, Kimberly B. Rogers of Dartmouth College, Deepak Rishi of the University of Waterloo, and Meiyappan Nagappan of the University of Waterloo. They recently published an article in “Small Group Research“ entitled “Spotlight on Methods: Artificial Intelligence and Social Simulation: Studying Group Dynamics on a Massive Scale,” which is currently free to read for a limited time. Below, They discusses some of the findings of this research:]
Technological and social innovations are increasingly generated through informal, distributed processes of collaboration, rather than in formal, hierarchical organizations. In this article, we present a novel combination of data-driven and model-based approaches to explore the social and psychological mechanisms motivating these modern self-organized collaborations. We focus on the example of open, collaborative software development in online collaborative networks like GitHub (github.com). The synthesized approach is based in affect control theory (ACT), and a recent framing in Artificial Intelligence known as Bayesian affect control theory (BayesACT). The general assumption of ACT is that humans are motivated in their social interactions by affective alignment: They strive for their social experiences to be coherent at a deep, emotional level with their sense of identity and general worldviews as constructed through culturally shared symbols. This alignment is used in BayesACT as a control mechanism to generate artificially intelligent agents that can learn to be functioning members of a social order (see bayesact.ca for further information).
We show in this article how such a model solves two basic problems in the social scientific study of groups and teams. First, because empirical research on groups relies on manual coding, it is hard to study groups in large numbers (the scaling problem). Second, conventional statistical methods in behavioral science often fail to capture the nonlinear interaction dynamics occurring in small groups (the dynamics problem). The ACT-based models we present allow for sophisticated machine learning techniques to be combined in a parsimonious way with validated social-psychological models of group behaviour such that both of these problems are solved in a single computational model.
The purpose of the present article is to discuss the promises of this cross-disciplinary, computational approach to the study of small group dynamics. We review computational methods for using large amounts of social media data, and connect these methods to theoretically informed models of human behaviour in groups. To use a metaphor, we are digging into digital group dynamics data with a sophisticated, artificially intelligent shovel, and showing how computational social science can be taken to a new level with this unique and novel combination of data-driven and model-based approaches. The work is an international collaboration called THEMIS.COG (themis-cog.ca) between researchers in Canada (University of Waterloo), the USA (Dartmouth College), and Germany (Potsdam University of Applied Sciences).
Stay up-to-date with the latest research from Small Group Research and sign up for email alerts today through the homepage!