A toolkit for analysis and prediction of touch targeting behaviour on mobile websites

Abstract

Touch interaction on mobile devices suffers from several problems, such as the thumb's limited reach or the occlusion of targets by the finger. This leads to offsets between the user's intended touch location and the actual location sensed by the device. Recent research has modelled such offset patterns to analyse and predict touch targeting behaviour. However, these models have only been applied in lab experiments for specific tasks (typing, pointing, targeting games). In contrast, their applications to websites are yet unexplored. To close this gap, this paper explores the potential of touch modelling for the mobile web: We present a toolkit which allows web developers to collect and analyse touch interactions with their websites. Our system can learn about users' targeting patterns to simulate expected touch interactions and help identify potential usability issues for future versions of the website prior to deployment. We train models on data collected in a field experiment with 50 participants in a shopping scenario. Our analyses show that the resulting models capture interesting behavioural patterns, reveal insights into user-specific behaviour, and enable predictions of expected error rates for individual interface elements.