Computational Social Science (CSS) is an interdisciplinary field that uses computational methods such as machine learning, natural language processing, and network analysis to study human behavior and social phenomena. CSS enables researchers to model, analyze, and even predict complex social dynamics. It blends the rigor of computer science with the theoretical grounding of the social sciences, offering novel insights into communication, influence, inequality, and collective behavior. CSS research in our section focuses on the following two key directions.  

The first one led by Ioannis Caragiannis studies whether and how socially desirable properties can be guaranteed by algorithms. Typical examples include the proportional representation in voting, stability in matching mechanisms, and fairness in resource allocation problems. Members of our section are particularly active in these topics with the study of the recent trend of sortition in voting, the study of new notions of stability for randomized matching mechanisms (e.g., for labor markets), and the definition of new fairness concepts for allocating indivisible items. 

The other led by Akhil Arora applies computational techniques on large-scale digital trace data (e.g., from social media, collaborative platforms, or online transactions) to understand and model human behavior. Arora’s and his group CLAN’s work in CSS spans a range of topics, from how humans search for, access, and contribute to knowledge on online platforms, to how quotations and attributions propagate in news media, to how large language models like GPT-4 impact human decision-making. They also study information visibility and inequality, exemplified by their work on “orphan” Wikipedia articles. CLAN develops tools and methods like unsupervised quote attribution systems and causal inference frameworks and applies them to rich behavioral datasets from Wikipedia, Twitter, and beyond. Their research reflects a strong methodological commitment to observational studies, experimentation, and causal reasoning, making CLAN a leading contributor to the evolving landscape of computational social science. 

Our research has significant social and practical implications by deepening our understanding of human decision making, social dynamics, and knowledge-seeking. Our work provides formal methods to reason about collective decision-making, resource distribution, representation, and social dynamics—areas where fairness, efficiency, and transparency are often in tension. By advancing our understanding of how people seek, share, and engage with knowledge, our work helps improve the design of digital knowledge systems, particularly Wikipedia that millions rely on for information daily. Tools like Quotebank enhance transparency and accountability in news media by accurately tracing who said what, helping combat misinformation and improve media literacy. Overall, our work blends theory and practice to create systems and insights that empower individuals, inform decision-making, supports ethical and socially aware use of technology in public discourse, and strengthen public knowledge infrastructures.