Paper Reading #15 Gesture Search: A Tool for Fast Mobile Data Access

Comments

Reference information
Authors: Yang Li
When/Where: UIST’10, October 3–6, 2010, New York, New York, USA

Summary
This paper presents an Android mobile application that allows users to search through their contacts, bookmarks, applications and music by drawing, or gesturing, letters on the screen. The designer points out the need for this new modality is due to the difficulty of having to locate desired items in such large datasets. With this system, results are generated after each input character is drawn. So if the letter 'A' is entered first, then all results containing that letter are displayed. Characters that appear similar, like 'H' and 'A', generate results for both, and results selected by users (by tapping on them) are learned over time and shown at the top of the list in future searches. The system is also capable of distinguishing gestures from touch events like scrolling and selecting through the results list.

A longitudinal user study was conducted internally with selected company employees to try out the system in their day-to-day lives, not in a controlled environment. User searches were logged, which showed that two-thirds of searches were for contacts. For the top reasons to use Gesture Search, 44% of participants commented that Gesture Search saved them time to find an item. Based on the log analysis, 53% of the accesses took less than three seconds to complete. However, an even bigger advantage of using Gesture Search seems to be the low cognitive load required of users. 59% of participants commented that they liked Gesture Search because it does not require them to type or navigate in user interfaces. As users commented that they could find an item without even looking at the screen. The major criticism from our users is that the current way of invoking Gesture Search, which is done using a home screen shortcut, can be inconvenient.

Discussion
I was pretty interested in this paper because I have used this application somewhat before reading the paper. I find gesture searching to be fairly fun, but in most situations, I default to searching with the keyboard or scrolling through a large list. I believe this is because I am too impatient to invest the time needed to adjust to using gestures. If I were part of the longitudinal study, though, I may behave differently. I also question whether this really is a quicker and less stressful way of searching through data; gesturing requires sketching as well as reading through the generated results, whereas scrolling through menus only requires reading, so it seems like this could be introducing additional cognitive load, not reducing it, as the participants in the user study remarked.