Estimation of browsing states in consumer decision processes from eye movements

Eye movements can be an important cue to reveal consumer decision processes. Findings from existing studies suggest that the consumer decision process consists of a few different browsing states such as screening and evaluation. To reveal the characteristics and temporal changes of browsing states in catalog browsing situations, this study proposes a hidden semi-Markov-based gaze model, where the states in consumer decision processes are explicitly modeled as hidden states. To achieve a better understanding of consumer decision processes, eye movements are first encoded to a sequence of gaze features using semantic structure of digital catalogs. The proposed gaze model is trained and evaluated using gaze data which was collected from eight participants in an experimentally controlled catalog browsing situation. We analyze the estimated browsing states and demonstrate that the states can contribute to improve the performance of viewer interest estimation.