An empirical evaluation for defining a mid-air gesture dictionary for web-based interaction

Abstract: This paper presents an empirical evaluation of mid-air gestures in a web setting. Fifty-six (56) subjects, all of them HCI students, were divided into 16 groups and involved as designers. Each group worked separately with the same requirements. Firstly, designers identified the main actions required for a web-based interaction with a university classroom search service. Secondly, they proposed a set of mid-air gestures to carry out the identified actions: 99 different mid-air gestures for 16 different web actions were produced in total. Then, designers validated their proposals involving external subjects, namely 248 users in total. Finally, we analyzed their results and identified the most recurring or intuitive gestures as well as the potential criticalities associated with their proposals. Hence, we defined a mid-air gesture dictionary that contains, according to our analysis, the most suitable gestures for each identified web action. Our results suggest that most people tend to replicate gestures used in touch-based and mouse-based interfaces also in touchless interactions, ignoring the fact that they can be problematic due to the different distance between the user and the device in each interaction context.