Mining social data

5 Feb 2024

How can social disparities be analyzed with data mining methods? Sociologist Carsten Schwemmer, Professor of Computational Social Sciences at LMU, researches this question.

Sometimes it takes a hefty dose of technology to uncover hidden discrimination. Take carpooling: Do people with migration backgrounds find it more difficult to successfully offer a ride on the corresponding platforms? Yes, ascertained Carsten Schwemmer after a detailed study: When offering a ride, people in this category had to go more than a quarter below the usual price on average to get their car full. Same trip, same offer, same driver ratings – just a first name that signals a migration background was all it took for the person to get fewer bookings.

These are the cases that have long interested the sociologist. For weeks, Schwemmer, then a doctoral student, connected computers to each other and got them to evaluate information from online carpooling platforms. “We’ve got lots of important and exciting data we’d like to analyze in the social sciences, but we often lack the tools,” says Schwemmer, who has been Professor of Computational Social Sciences at LMU’s Department of Sociology for just over a year. “To this end, we import technical know-how from computer science and apply it to social science questions.”

Professor Karsten Schwemmer stands in a relaxed pose in front of two computer monitors, one of which is in operation. In the background you can see a window with a view outside.

Professor Carsten Schwemmer

He analyses social disparities using data mining methods.

© Stephan Höck

Pioneer in a burgeoning subdiscipline

Whether it is the dubious tricks in US election campaigns, the influence of political influencers, or the recognition of inflammatory content online: “The conjunction of technology and society runs like a golden thread through the topics and methods of my research projects,” says Schwemmer. While still a sociology undergraduate at the University of Bamberg, he sought out suitable computer science courses to take. At that time, there weren’t any course offerings geared toward such a crossover. “So I pretty much pieced together my main research interest by myself.”

Today, the subdiscipline of computational social sciences has grown rapidly and Carsten Schwemmer is a much sought-after researcher and lecturer.The stations in his career to date have delimited his field of work with increasing sharpness: After completing his doctorate, also in Bamberg, Schwemmer was junior professor at the Center for Data and Methods at the University of Konstanz. Then he spent half a year as fellow at the Weizenbaum Institute in Berlin.

As a postdoc, he worked at the Center for Information Technology Policy at Princeton University. He was interim professor of political sociology in Bamberg and headed a research team in the Department of Computational Social Sciences at the GESIS Leibniz Institute for the Social Sciences in Cologne, before being appointed professor at LMU.

Tricks and subterfuge ahead of US elections

During his time at the Ivy League school Princeton, for instance, he played a major role in a project that investigated irregularities in the runup to the US elections in 2020. With sophisticated data mining methods, the researchers compiled a corpus of 300,000 emails that organizations had sent out to raise money and attract supporters for their US election campaigns. Evaluation of the data revealed that the use of manipulative tactics was the norm rather than the exception.

The researchers found various patterns of clickbait and deception – for example, in email headers – which were designed to get recipients to react to the message. For the scientists, this was a form of undue influencing of voters.Naturally, the goings-on in social media and how they can be analyzed with the methods of computational social sciences have also caught Schwemmer’s scholarly attention.

Just recently, he received funding for a new project he is undertaking with colleagues. Together, they will use AI methods to investigate how xenophobic and antisemitic content – and particularly pictures and videos – can be recognized in social media. In addition to Schwemmer himself, the project team comprises Diana Rieger, Professor of Communication Science at LMU, and Yannis Theocharis, Professor of Digital Governance at the Technical University of Munich (TUM). “Working across disciplinary boundaries is very important to me,” says sociologist Schwemmer. “It expands our horizons” – on to questions of technology and society.

What are you looking for?