Quantitative User Interface Profiling
Metrics and Visualization
Quantitative profiling has proven an effective tool for system
optimization and tuning in many areas such as operating systems,
networking, database, and software development; but is essentially
absent from user interface (UI) development. Profiling is effective
because it produces concrete, comparable, and reproducible performance
information. In addition, profiling software is generally inexpensive
and off-the-shelf. Most often, UI testing refers to observational
research done in usability labs in which a usability engineer manually
records and analyzes the user's experience with the UI.
The goal of this work is:
- To produce a set of metrics that will quantitatively
describe aspects of a UI from user trace logs. These metrics will
measure UI elements such as learnability and flow--the balance of
a UI's efficiency, intuitiveness, and logic.
- To develop an interactive, graphic representation that
correlates these metrics to actual usage traces. This representation
is meant to help the developer get a sense of the metrics and associate
them with trends in usage.
It is our hope that this level of UI profiling will enable the UI
developer to better compare and understand the nature and severity
of UI defects and provide UI development team managers with
more concrete information on which to base project planning and
resource allocation decisions.
Status: 9/24/00
We are currently preparing a QUIP distribution for download and writing a paper for UIST '01.
Contact Information
|
