Modeling User Experience of Large Display-Based Interaction with Physiological Indicators and Machine Learning Techniques

While large display-based interaction has gained increasing popularity, its user experience is usually assessed by self-reporting measure and subject to bias. This study proposed an alternative approach by modeling user experience of large display-based interaction with physiological indicators and machine learning techniques. Twenty-four participants attended an experiment where they were asked to interact with a large display under varied body postures and interaction distances. Both self-reporting user experience (i.e., perceived usability and workload) and electromyography measures during task performance were collected and trained by three different machine learning models, out of which the best model could predict over 70% of the variance of user experience. Several electromyography measures were identified as effective indicators for user experience. The study demonstrates the feasibility of modeling user experience with physiological indicators and machine learning techniques in large display-based interaction.