🌐
Platforms & user behavior
How do design, recommendation, and incentives shape what we read, watch, and share? We combine behavioral logs with quasi-experimental designs to disentangle attention, selection, and treatment.
Research
What we study
Matsui Lab works at the intersection of computational social science and web information studies. We use digital trace data, statistical inference, and machine learning to ask social-scientific questions about behavior, communication, and knowledge.
📚
Wikipedia & knowledge at scale
From Wikipedia editing to academic publishing, we study how distributed communities cooperatively create knowledge — and where systematic gaps emerge (e.g., urban–rural disparities in coverage).
💬
Text, language & embeddings
Word embeddings, topic models, and large language models as instruments for measuring social and economic constructs — including monetary policy, political communication, and ideology.
⚠️
Harmful & toxic behavior
Detection and analysis of toxic, deceptive, or coordinated activity online — including the often-overlooked role of fans as a source of toxic communication.