Personalized Recommendation of User Interfaces Based on Foot Interaction

Foot interaction is a type of input method gaining traction in recent years. It is often combined with other interaction methods to complement other interaction methods (e.g., hand interaction). However, recent advances in literature have not paid much attention to user interface design for foot interaction. In this paper, we propose a user interface recommendation method based on foot interaction that may customize the interactive interface by learning from the users’ behaviors and features, resulting in higher accuracy of interactions. The proposed method utilizes personalized features of users to classify them into different classes and filter out user interfaces with varying element attributes for different user classes. Further, the filtered interface set is recommended to users using the UCB1 method, which helps prioritize the most suitable interfaces for each user. The experimental results show that our method may recommend interfaces with lower misstep conditions to the users, reducing the likelihood of user error operations during the movement process. This verifies the effectiveness of the personalized recommendation approach in optimizing interface layout for foot interaction. To contribute to the research community working on foot interaction, the paper introduces a user interface dataset specifically designed for foot interaction. This dataset integrates individual user characteristics and current design principles, possibly providing valuable guidance for designing foot interaction interfaces.