05771 HCI Process and Theory

1
05-771HCI Process and Theory

• Scott Hudson
• scott.hudson_at_cs.cmu.edu
• Ken Koedinger
• koedinger_at_cmu.edu

2
The (no longer mythical) PT course

• Serves as intro to HCI research
• New course
• Experimental structure
• Broad / diverse subject matter
• Intertwined with a practical project
• A vehicle for exposure to diverse approaches
• Team taught (by much of the HCII faculty)
• Scott is instructor of record
• Ken is coordinating

3
Project orientation

• Theme project as organizing mechanism
• Real project with a lot of details reflecting the
  reality of academic research
• Finding a worthwhile problem / idea of interest
• Forming a team and a project that works for it
• Getting it funded
• Doing the work
• Structuring vehicle for
• HCI background (readings and discussion)
• Viewpoints on doing research
• Practical projects for you

4
You will propose a project

• Something
• Within the theme (broadly defined)
• Interesting and worthwhile
• Maybe leads to publication and/or follow on work
• You can sell (to faculty and other students)
• Doable this semester
• Short initial project proposal week 4

5
Course outline

• Introduction
• Intro to the theme project
• HCI research
• Finding problems / ideas
• History of problems and approaches
• Working in interdisciplinary teams
• Seeds
• Personal and disciplinary views of problems and
  directions of inquiry

6
Course outline (cont.)

• Project proposals
• Designed to be analog to a typical grant proposal
• Need to sell to fellow students
• Need to find a faculty mentor
• Projects
• Will select about 1/3 to actually do
• Work in small teams
• Lots of discussions along the way
• Final presentations

7
Administrative stuff
8
Class meetings

• Scheduling was very difficult and result is not
  optimal
• Will meet for two 90 minute lectures a week but
  time slot will vary depending on the instructors
  for the week
• No single pair of times worked for all
  instructors
• Possible times
• Monday 430-600 600-730
• Friday 1200-130 130-300

9
Class meetings

• Conflicts?
• Strategy preference (vote for one)
• (1) Split Mon/Fri
• (2) Long day
• Time Preferences (vote for two)
• Mon 430-600
• (2) Mon 600-730
• Fri 1200-130
• (4) Fri 130-300

10
Tentative grading criteria

• 15 class participation
• 25 project proposal
• 60 final project

11
Photos

• Not absolutely required, but we would really like
  to have a photo of you to create a name/face
  sheet to help various instructors
• Shoot one today
• Send me one
• Email saying you dont want a photo taken
• scott.hudson_at_cs.cmu.edu

12
Class communications

• If you are not an HCII PhD student (i.e., on the
  hcii-phd-students_at_cs mailing list) send me email
  indicating your preferred email address (today!)
• http/www.cs.cmu.edu/hudson/teaching/05-771/

13
Questions?
14
The theme project Situationally Appropriate
Interaction

• Principal investigators (PI and co-PIs)
• Scott Hudson, Jodi Forlizzi, Sara Kiesler, Chris
  Atkeson, Jie Yang, Yoky Matsuoka
• Just funded (sort of) in NSF ITR program
• Also have DARPA seed funds for 1yr
• Proposal that describes this project assigned as
  reading for today
• http/www.cs.cmu.edu/hudson/teaching/05-771/hu
  dson_ITR_no_sal.pdf

15
Motivation

• Exponential growth of technology offers wonderful
  promise, but
• In an information rich world the scarce resource
  is human attention Simon 67
• We are going to loose a bunch of that benefit to
  the human costs
• Attention is a big cost, but also others
• This project is about finding solutions for that

16
Motivation

• Currently interfaces are mostly blind to the
  human situation they sit in
• They cant tell the difference between
• working alone at home at 2am,
• at work in a big meeting,
• giving a talk, or
• attending a funeral or a movie
• Blunder blindly through the human world
• Try to create systems that maneuver through the
  human (social) world
• One view rudimentary manners

17
Scenario

• Lee (AKA Scott) moves between situations
• Public vs. private
• Focused individual work vs. less focused work
  vs. group work
• System acts appropriately
• Attention demand for displays
• Communications filtering
• Privacy issues

18
Illustrates framework of Sense, Model, Act
Appropriately

• Sense
• Gather basic useful information about what is
  happening in the situation
• Model
• Analyze and structure the basic information into
  something usable
• Act appropriately
• Do the right thing in the interface
• Display and interaction

19
Goals for results

• Systems which can deal with some of the basics of
  the human (social) world, e.g.,
• know when not to interrupt,
• demand the right amount of attention,
• limit private information in public settings, etc.

20
Discussion

• Is the scenario possible/practical/doable?
• Other scenarios?
• Other situations
• Other task domains

21
Sensing

• Concentrate on non-invasive sensors
• Vision based sensing (also spatial audio)
• Expand later to other modalities
• Provide basic information such as
• Who is in a space
• How many
• Where are they looking
• What objects are they touching
• How much are they moving

22
Discussion

• Most predicative cues?
• Other sensor technologies?
• Cheap and easy to exotic

23
Modeling

• This is a key piece
• Probably the hard part
• Advantages (levers) we have
• Understanding social phenomena involved
• Cognitive modeling expertise
• Integration and tight feedback with other parts

24
Modeling

• Models of social situation
• Social engagement
• Conversation detection
• Meeting detection and classification
• Also individual work models
• Multi-tasking levels
• Task switching and stacking
• And task models
• Probably task specific

25
Discussion

• What models?
• Whats possible/promising to create?
• Whats most useful?

26
Acting appropriately

• Given information from modelsdo the right
  thing
• At least two parts
• Display
• Interaction

27
Acting appropriately Display

• Attention demand is again a key
• Know how to create displays that demand a lot of
  attention
• Whole literature on alarm design
• Know a lot less about how to create displays that
  demand less attention
• Much harder because they have to be able to
  deliver information
• Not even a lot in the way of design examples
• Ambient information displays

28
Acting appropriately Display

• Goal for results
• Library of display techniques graded/sorted by
  salience (prominence or attention demand)
• Mechanisms for composing displays at a given
  level of salience
• Methods
• Two bodies of knowledge to draw from
• Sensory and perceptual psychology
• Design
• Build and experiment

29
Discussion

• What experimental setup should be used to do the
  grading/sorting of salience?

30
Acting appropriately Interaction

• Displays will be much more useful if they are not
  static, but can be interacted with
• Similar issues
• We know how to build interfaces we are fully
  engaged with
• We have few examples of systems which are useful,
  but dont require a lot of engagement

31
Sample interactions

• Acknowledgement
• Deferral
• Delegation
• Drill-down

32
Discussion

• Other interactions?

33
Observation and evaluation

• Background formative evaluation
• Wizard of OZ study
• Salience evaluations
• Technology trials
• Usability studies

34
Wizard of OZ study (ongoing)

• Which sensors/models are most useful for
  predicting interruptability?
• Simulate sensors
• A/V recording
• Human plays sensor
• Self report interruptability
• Experience sampling
• Randomly poll the user to rate interruptability
• Find correlations

35
Discussion

• What other observation and studies should be done?

36
Project overview wrap-up

• For this class we want to take a wide
  interpretation of this project
• Many good problems in there in many areas
• Next, some discussion of where did this research
  direction come from

37
Readings

• The Coming Age of Calm Technology, Mark Weiser
  and John Seeley Brownhttp//nano.xerox.com/hypert
  ext/weiser/acmfuture2endnote.htm
• Techniques for addressing fundamental privacy
  and disruption tradeoffs in awareness support
  systems, Scott Hudson and Ian Smith, Proceedings
  of CSCW 96.
• http//www.acm.org/pubs/citations/proceedings/cscw
  /240080/p248-hudson/

38
(No Transcript)