Analyzing Motion of Touching Screen for Inferring User Characteristics

Due to advances in technology, today's users can enjoy a variety of services more comfortably on a larger screen than in the past. However, due to the large size, it is difficult for the users to touch all of the display with one hand. Moreover, most smartphone applications have inflexible forms that fill the entire screen, and the placements of buttons are usually fixed. To overcome such limitations, we focus on diverse sensor data obtained while touching the screen and propose the analysis method to infer user characteristics. The proposed method utilizes different sensor data from not only a touchscreen but also a gyroscope and an accelerometer in combination. Moreover, the proposed method utilizes the unsupervised clustering algorithm to quantify how much of the screen the user could reach. We conducted experiments involving users to evaluate our analysis method. The results show that, by using the proposed method, it is possible to know whether a user is right-handed or left-handed. Furthermore, the results verify that our analysis method is able to distinguish the natural, reach, and unreachable zones on the screen well.