Victor Kristof

Victor Kristof is a researcher at École Polytechnique Fédérale de Lausanne (EPFL), affiliated with the INDY2 laboratory. His work spans interdisciplinary domains, combining Natural Language Processing (NLP), Machine Learning, and Social Process Modeling to analyze legislative dynamics, vote prediction, and environmental perception.

Research Focus: Kristof develops interpretable models for democratic transparency, including aligning interest group positions with parliamentary speeches. He pioneered methods for predicting legislative edit acceptance using matrix factorization and NLP. His work on Swiss referendum prediction integrates historical data with real-time analysis via the Predikon platform.

Methodological Contributions: He applies Bayesian statistics, time-dynamic pairwise comparison models, and active learning algorithms to diverse problems, from carbon footprint perception to sports analytics. His War of Words framework reveals ideological patterns in EU law-making, while his Player Kernel model improves football match prediction.

Labs & Collaborations: Based at EPFL's Laboratory of Dynamic Information and Networks (INDY2), he collaborates with researchers like Matthias Grossglauser and Patrick Thiran. His datasets on legislative edits and carbon perception have advanced transparency studies.