Activity based computing

1
Activity based computing

• Rahul Nair

Advisor Dr. Elizabeth Mynatt Collaborator
Steve Voida
22 April 2004
2
History of Kimura

• Multitasking support
• Peripheral displays
• Background Awareness
• Montage visualizations

3
Motivation

• People seldom perform simple atomic actions
• Look at the big picture
• Lots of tools to do individual stuff
• No good way to tie in all the work related to an
  activity
• But we dont know what an activity is...

4
What is an activity?

• Different for each individual
• Granularity varies
• Getting a masters
• Printing a paper
• How do people think about activity?
• Micro level
• Monolithic
• Hierarchical

5
Goals

• Learn about users mental model
• Collect the artifacts of an activity
• Look for links between actions on a computer and
  activities
• Can we use actions to predict/classify activities?

6
Study Design Alternatives

• Ethnographic
• Really difficult to follow people around
• Limited manpower available
• Diary study
• Users often forget to note down tasks
• Experience Sampling
• Good compromise between the two
• Can be made adaptive and context aware

7
Study Design

• Pre-study questionnaire
• Record user activities over a 2 week period
• Documents
• Email information
• Web pages
• Window activity
• Users name each activity
• Long semi structured post study interview

8
Software

• Visual basic application for window, web page and
  PDF tracking
• Detect task changes using window rhythms
• Pop ups to remind users to name tasks
• Adapt to user feedback
• VBA code for Outlook, Word and Excel
• Instrumented version of Pine

9
Pilot study

• 3 users over a one week period
• Web pages and email contacts provide good
  identification
• Accuracy is dependant on users task granularity
• Varies from 40-80
• Need more data to draw final conclusions

10
Current status

• Started last week
• Currently have 8 users
• Grad students, professors, non academic users
• Try to get the widest possible range of users
• Interviews at the end of the two weeks

11
Future work

• Data analysis
• Find a common task model for users
• Integrate with Kimura
• Indexing activities to aid recall
• Possible release as a time management tool

12

• Questions?
• Would you like to participate? ?

13
Related work

• Virtual workspaces
• Rooms, virtual desktop managers, etc...
• Multitasking studies
• Gonzalez 04, Czerwinski 04
• Email studies
• Bellotti, Gwizdka 02