Preliminary Exploration of Eye-Tracking Technology to Classify Human Personality Traits During Conversations with Virtual Humans

The recognition of personality traits has attracted considerable attention across various research fields. Identifying these traits can enhance the quality of life and improve interpersonal relationships, significantly impacting social behavior. However, conventional assessment methods like self-report questionnaires and face-to-face interviews can be expensive and resource-intensive, leading to inefficient recognition processes. Based on these considerations, this work proposes a preliminary exploration of Eye-Tracking technology as an implicit measure for personality trait classification. An experimental study involving 64 healthy participants engaging in conversations with Virtual Humans (VHs) was conducted to establish a controlled context. The collected data underwent preprocessing and classification using Machine Learning (ML) algorithms, with their performance assessed through appropriate validation techniques. The experimental results present encouraging findings, suggesting that eye-tracking holds the potential to facilitate a deeper comprehension of human personality traits during causal conversations with virtual humans.