Characterizing Large-Scale, Long-Term Use of Applications

Опубликовано 17 августа 2016, 21:07

Identifying patterns and trends in large-scale, long-term application use is critical to ensuring an application meets the needs of its user base. However, there are few published analyses of long-term use of direct manipulation applications. In this talk, I present an analysis of 2 years of usage data from over 200 users of an instrumented version of GIMP. In line with previous work, I show that individuals tend to have relatively small, idiosyncratic command vocabularies. I also demonstrate a new means of inferring higher-level tasks, from logs of command use alone, via use of a recent clustering technique. Finally, I discuss implications for interface design and demo a new version of GIMP we're building to demonstrate these design implications. This new version of GIMP, Adaptable GIMP, demonstrates new possibilities for social customization and shared community learning.