Open Access

Patrick Harms

• User interaction data such as keystrokes, mouse movements, and touch events are increasingly used as input for machine learning (ML) and artificial intelligence (AI) approaches in areas such as user modeling, stress detection, emotion recognition, and behavioral biometrics. However, before ML algorithms can be applied, interaction data must be recorded, processed, segmented, and transformed into meaningful metrics. These preparation steps are rarely standardized and often insufficiently reported, even though they strongly influence the resulting data quality and the validity of downstream analyses. This tutorial addresses the methodological foundations of working with interaction data. It introduces participants to best practices for recording interaction events, preprocessing raw interaction logs, slicing data into meaningful segments, and calculating reliable interaction metrics. Particular attention is given to subtle effects that occur during preprocessing, such as the influence of slicing strategies on derivedUser interaction data such as keystrokes, mouse movements, and touch events are increasingly used as input for machine learning (ML) and artificial intelligence (AI) approaches in areas such as user modeling, stress detection, emotion recognition, and behavioral biometrics. However, before ML algorithms can be applied, interaction data must be recorded, processed, segmented, and transformed into meaningful metrics. These preparation steps are rarely standardized and often insufficiently reported, even though they strongly influence the resulting data quality and the validity of downstream analyses. This tutorial addresses the methodological foundations of working with interaction data. It introduces participants to best practices for recording interaction events, preprocessing raw interaction logs, slicing data into meaningful segments, and calculating reliable interaction metrics. Particular attention is given to subtle effects that occur during preprocessing, such as the influence of slicing strategies on derived metrics. Participants will learn how different data preparation choices affect interaction metrics and why transparent reporting of these decisions is essential for reproducibility and validity in ML-based research. The tutorial provides conceptual guidance, practical examples, and recommendations for reporting interaction data processing pipelines. It equips researchers and practitioners with the methodological baseline required to prepare interaction datasets that can reliably serve as input for ML and AI analysis.…