The Cognition Lab at NASA Ames Research Center, specifically the developers
of the Apex Cognitive Modeling tool asked us to help them learn how to make
human-performance modeling easier and apply this knowledge to improve their
Apex modeling tool.
This project began in the
Spring semester 2001 with a team of eleven Masters in HCI students. During this
semester we engaged in Contextual Design to understand users and their work
and identify breakdowns in the users' work process. We applied these insights
to develop design solutions that resolve breakdowns and initial designs of tools
to meet users' needs.
During the Summer semester,
we split into two groups and each group pursued creation and refinement of a
prototype to the selected designs. My group of five worked on a Trace Visualization
Tool, which we dubbed "Vista." Throughout the summer we applied knowledge of
interface design principles to develop working prototypes, which we tested with
users. We had a weekly rapid prototyping schedule during which we refined our
design based on the results and tested again.
We recognized that analyzing
trace output generated from running an Apex model is difficult because: a huge
amount of raw data is produce with every run, it's hard to anticipate what filter
is needed beforehand, and information about a single task may be spread over
the entire document. These difficulties also make it hard to answer specific
questions, and even harder to explore the results.
Vista facilitated answering
factual questions and performing exploratory analysis by:
- Making the model elements - Tasks, Resources and Simworld Objects, their relationships, and changes
over time visually apparent
- Linking related information across views
|