Michael Belfrage

I am a Computational Social Scientist dedicated to interdisciplinary work which my background reflects, with a BA in Political Science and an MSc in Computational Social Science. Currently, I am pursuing a WASP-HS PhD in Computer Science at the Faculty of Technology and Society here at Malmö University where I am also a part of the Internet of Things and People (IoTaP) Research Center. I am an outgoing individual who enjoy reading, exploring new ideas, and engaging in stimulating conversations. My research primarily focuses on applying computational methods to political science applications. I have a genuine passion for complex systems and Agent-based Models, as well as network science and other graph-theory-related work. I have previously worked on evaluating the statistical accuracy of peer influence diagnostics in longitudinal networks using two different models, the Stochastic Actor Oriented Model and proportional odds logistic regression. This research involved analyzing the dynamics of social influence within social networks across time, examining how individuals are influenced by their peers and their behavior, and assessing the accuracy of peer influence diagnostics of the two aformentioned models. Additionally, I have worked in a project involving the collection of public information to construct and analyze bipartite networks. The goal of the project was to shed light on the portfolio accumulation of Swedish politicians and power concentration in municipal institutions. By examining the connections between politicians and the portfolios they hold in various institutions, using community detection algortihms, we sought to uncover impotant clusters and power concentrations within political networks.

Realizing the Potential of Agent-Based Social Simulation for Public Policy

My research project is a part of the larger overarching WASP-HS project: Realizing the Potential of Agent-Based Social Simulation funded by the Wallenberg Foundations. Together with my supervisors, Paul Davidsson and Fabian Lorig, and my PhD colleague Emil Johansson, we aim to make computer simulations a more accessible and reliable tool for policy analysis. Our ambition is to understand how Agent-based Social Simulation (ABSS) – a powerful simulative paradigm capable of imitating human behaviour – and artificial intelligence (AI) could be leveraged to facilitate evidence-based policy making in public systems. This, we hope to achieve, by identifying the mechanisms of social systems and policy instruments which could be used to regulate them. By using ABSS to simulate the social system of interest, different 'policy-experiments' can be performed by manipulating these mechanisms in-silico. Empirically evaluating policy options is often infeasible, if not impossible, as it requires a lot of time, funds, and personnel. Without accessible analytical tools to provide insights into the system, decision-makers may have little choice but to engage directly with the target system without any information. Our aim is to allow decision-makers to explore different policy options before implementing them in the real world. By performing these policy-experiments in safe, simulated environments, unintended consequences associated with different policy prescriptions can be identified and diverted.