Visual similarity of pen gestures

Author(s): A. Chris Long, Jr., James A. Landay, Lawrence A. Rowe, and Joseph Michiels
Venue: Proceedings of the SIGCHI conference on Human factors in computing systems
Date: unknown

Type of Experiement: Controlled Experiment


Location: The Hague, The Netherlands
Pages: 360 - 367
DOI Bookmark:

This paper examines pen gestures in pen-based user interfaces. In an attempt to improve pen gesture design, the authors perform a pair of experiments to determine why users find pen gestures similar. The authors suggest that spacial similar actions, such as scroll up and scroll down, should be similar to facilitate learning. Conversely, more abstract actions, such as cut, copy, and paste, should be different to so that the user does not confuse them. The authors state, "we have derived a computational model for predicting perceived gesture similarity that correlates 0.56 with observation."

The authors ran a pair of experiments to measure the similarity of a variety of pen gestures. The goal of the experiments was to determine a set of metrics to measure human-perceived similarity in gestures. The authors ran the experiment twice to on separate sets of data and with separate users to verify the results. The first experiment was ran with 21 student volunteers. Users were asked to select the most dissimilar gesture of a choice of three gestures. Many gestures were tested and were randomized between each user. The authors were then able to come up with a model from their experimental data. The authors validated this data on twenty new users and a new set of gestures in their second experiment.

The authors were able to develop well defined computable features of pen gestures such as curviness. These computable features can be used to determine the similarity or dissimilarity of gestures which can be desirable when designing pen-based user interfaces.