My research contributes statistical methods to understand and augment fundamental social processes in systems of our information society. The ambition of my research is to design fair and representative social computing systems, such as rating systems in social media, predictive models of human decision-making, and recommender systems. Important questions concerning the design of these systems include: how to train fair and explainable machine learning models and how to avoid biases in social systems? To address these questions in practical settings, I collaborate with Donald Tomaskovic-Devey, the founding Director of UMass Center for Employment Equity.
In addition, we are running computational social science projects "Current Affairs 2.0: Agenda Setting in the European Union" and "Agendas and frames in a global pandemic: Evolution of cross-country media coverage", funded by the Volkswagen Foundation in 2017-2022. Principal investigators include Scott Hale (Oxford), Fabian Flöck (Leibnitz School of Social Sciences), David Jurgens (University of Michigan), and me. More information is available at: http://www.euagendas.org. Among others, we will publicly release (for research purposes only) the largest semi-labeled dataset of multilingual news articles, including several thousands labeled articles in addition to millions of unlabelled articles.
