Paper Number
1071
Paper Type
CRP
Abstract
Due to advances in extended reality technology, an increasing number of head-mounted displays are equipped with eye trackers. These sensors allow to predict customers’ preferences on-the-fly. Such information can serve as features for recommender systems. We propose to treat eye tracking data as time series and utilize a deep time series classifier for inference. Our evaluation investigates possibly early predictions about customer preferences for healthy products in a virtual reality environment. The results, that are based on data from a large-scale laboratory experiment, demonstrate superior performance of the time series classifier, compared to a shallow gradient boosting baseline. They indicate a trade-off between prediction quality and how early this prediction is made. Overall, our study suggests that eye tracking and time series classification are valuable avenues for research and practice. Adaptive (shopping) assistants and recommendations based on artificial intelligence and bio sensors seem to be in close vicinity.
Recommended Citation
Weiss, Tobias; Pfeiffer, Jella; and Pfeiffer, Thies, "Early Bird - Predict healthy product choices in virtual commerce" (2024). ECIS 2024 Proceedings. 3.
https://aisel.aisnet.org/ecis2024/track09_coghbis/track09_coghbis/3
Early Bird - Predict healthy product choices in virtual commerce
Due to advances in extended reality technology, an increasing number of head-mounted displays are equipped with eye trackers. These sensors allow to predict customers’ preferences on-the-fly. Such information can serve as features for recommender systems. We propose to treat eye tracking data as time series and utilize a deep time series classifier for inference. Our evaluation investigates possibly early predictions about customer preferences for healthy products in a virtual reality environment. The results, that are based on data from a large-scale laboratory experiment, demonstrate superior performance of the time series classifier, compared to a shallow gradient boosting baseline. They indicate a trade-off between prediction quality and how early this prediction is made. Overall, our study suggests that eye tracking and time series classification are valuable avenues for research and practice. Adaptive (shopping) assistants and recommendations based on artificial intelligence and bio sensors seem to be in close vicinity.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.