Using Contextual Information
in Personal Informatics Systems
to Reveal Factors that Affect Behavior
Mar 31, 2010 / Wed
10:00 AM
GHC 4405
Committee
Anind Dey, Carnegie Mellon University (co-chair)
Jodi Forlizzi, Carnegie Mellon University (co-chair)
Niki Kittur, Carnegie Mellon University
John Stasko, Georgia Institute of Technology
Abstract
Today, there is a personal informatics system for almost any behavior
(see a list at
http://personalinformatics.org/tools).
These systems help people collect behavioral information
to explore and reflect on.
Because most systems only show behavioral information,
finding factors that affect one's behavior is difficult.
Incorporating contextual information, such as location, may help.
To explore this, I developed prototypes of IMPACT,
a system for physical activity awareness with support
for contextual information.
Previous deployments showed that context can increase people's awareness
of opportunities for physical activity
and automation facilitates long-term use but reduces immediate awareness.
I will develop a third prototype that supports better selection
of contextual information, maintenance of immediate awareness
during automated collection, and improved visualizations.
I will compare the prototype in a field study to a steps-only system
and identify features critical to its effectiveness.
I will take the lessons learned and describe how they may apply
to supporting contextual information in personal informatics systems
for other types of behaviors.
