Towards an Effective Means of Collecting Validated Behavioural Data in VR

This paper introduces a novel methodology for collecting and validating non-verbal behavioural data in virtual reality (VR) environments. Focused on addressing the complexities of behavioural prediction in VR, our approach integrates a machine learning (ML) model and a human subject-based validation process. The ML model is tested across diverse tasks to evaluate its accuracy in capturing subtle behavioural nuances, while the human validation aspect provides a unique cross-reference, ensuring the reliability of the re-enacted avatar behaviours. Central to our methodology is the emphasis on ethical considerations in data handling, aligning with current legislative standards and best practices. This innovative approach enhances the precision of behavioural data collection in VR and offers valuable insights for designing more interactive and user-centric VR systems, particularly in education and marketing contexts.