Rating Methods for Proactive Recommendation on Smartwatches

This paper analyzes possible interaction methods for using a recommender system on a smartwatch. As prerequisite, we describe interaction patterns currently used by Android Wear applications. Based on a prototype implementation the interaction methods action buttons, two button card and swipes are compared against each other. In a user study, 31 participating students were asked to rate restaurant recommendations offered in the setting of a context-aware, proactive recommender system. For each of the interaction methods, they saw the same three recommendations. Afterwards, the participants judged the user experience of the methods. The reaction time was measured for each recommendation. The study showed, that the two button card and the swipes methods outperform the action buttons method with regard to reaction time and user experience. The two button card was ranked highest in terms of use quality, the swipes method in terms of design quality. About half of the participants liked each of these two methods best. These findings are especially interesting, as the action buttons method is the interaction pattern included in most Android Wear applications, because it is easier to implement as the other methods.     «

This paper analyzes possible interaction methods for using a recommender system on a smartwatch. As prerequisite, we describe interaction patterns currently used by Android Wear applications. Based on a prototype implementation the interaction methods action buttons, two button card and swipes are compared against each other. In a user study, 31 participating students were asked to rate restaurant recommendations offered in the setting of a context-aware, proactive recommender system. For each o...     »