TSQUAIR: Table Space for Queries

1
T-SQUAIRTable Space for Queries
Architectural Image Retrieval

• Ame Elliott Marti Hearst

2
Work in Progress

• Study is still going on
• Talk is a little rough
• Im interested in feedback

3
Overview

• Nature of architectural design
• UIs for Info Access in Architecture
• Why we think the desk can help
• User study design
• Preliminary results
• Video highlights

4
Why Study Architects?

• Multi-disciplinary, collaborative domain
• Culturally learned practices of design lend
  themselves to ethnographic research
• Novices are different than professionals
• Hierarchical
• Characteristics of Design
• Visual thinking
• Ill-defined tasks

5
Architects Info Access Tasks

• Monitoring - (ODay Jefferies)
• What has Tadao Ando done lately?
• Find image of something seen before
• For client, student, presentation, etc.
• Get background for a new project
• Site information (context of Kansas City)
• Similar projects (other hospitals)
• Get ideas, inspiration

6
Info Access in Architecture

• Related to organization of precedent knowledge
• Description of a world of architectural
  knowledge, how to access it
• Most information access tasks are done better
  without a computer
• Paper is important for sketching, juxtaposing,
  collage making
• Current UIs to Web-based architectural
  collections arent very good.
• Dont follow general considerations to support
  search
• Minimal support for browsing

7
Whats Good About the Desk?

• Size
• Drawings are large (24 x 36)
• Juxtaposing many images easier
• Luminous
• Easy to trace, functions as a light-table
• Flat horizontal
• Good for drawing and spreading out papers
• Lightweight UI
• Pen input intuitive to designers

8
Motivations for User Study

• Evaluating interfaces for IR is complex
• Isolate effects of the physical affordances of
  the desk
• Display size, flat, luminous, like a drafting
  table

9
Experiment Design

• Two tasks
• Sketching groups of figures
• Sorting images into groups
• Three environments
• Digital Desk/pen 1
• Sketch 14 LCD tablet/ pen 2
• Monitor/Mouse

10
Display Specifications

• Digital Desk (31x26) Projection
• Resolution 1280x 1024
• 1626 pixels/inch
• .04 PowerPoint slides/inch
• Sketch 14 (11x 8.5) LCD
• Resolution 1024 x 768
• 8411 pixels/inch
• .37 PowerPoint slides/inch
• Monitor (17x12) Screen
• Resolution 1280x 1024
• 6425 pixels/inch
• .17 PowerPoint slides per inch

11
Quantitative Data Gathered

• Background Info
• Described art/architecture training, computer
  use, and work environments
• 2 Post-task questionnaires
• Used Likert Scale
• Asked about references, display quality, input
  device
• Saved copies of sketches and sort results
• Times of sorting tasks

12
Qualitative Data

• Open-ended verbal questions at the end of the
  questionnaires
• Video and still images of behavior during the
  tasks

13
Hypothesis Development

• Architects (as opposed to users in general) would
  prefer the Digital Desk and work faster on it
• Pre-test showed that even though I was faster,
  new users werent (learning curve for the pen?)

14
Revised Hypotheses

• Some affordances of the desk would hinder image
  sorting
• People would be slower and make mistakes because
  of large area, and large movements required
• The poor display quality and trouble selecting
  and dragging with the pen would be problems
• Those with more architectural background would
  like it more anyway

15
Who were the respondents?

• 9 men and 3 women from 5 continents
• 1 left-handed
• Computer experience

16
Years of Experience
17
Procedure- Sketching

• Informed consent
• Background questionnaire
• Comfort adjust angle, height, light
• Sketching same group of figures in 3 different
  patterns on each of 3 devices (order randomly
  determined)
• Post-sketching task questionnaire
• Open-ended verbal questions

18
Sketching-Results Not Discussed

• Confidential for Now
• May enlist help evaluating sketches for
  differences between devices
• Dont want to bias observers

19
Procedure - Sorting

• Comfort Adjust height, angle, light
• Tutorial
• 3 sorting tasks in PowerPoint on each of 3
  devices
• One task each sorting icons, one set of photos,
  one set of buildings
• Which of 3 variations on which device in which
  order randomly determined
• Post-task questionnaire
• Open-ended verbal questions

20
(No Transcript)
21
(No Transcript)
22
Preliminary Results

• Sorting task only
• Disclaimer work in progress
• Rough sense of trends, not serious number
  crunching

23
Were Not Sure What Predicts Preferring the Desk

• Background (neither artiness or computer
  experience) didnt predict who liked desk
• Work environment didnt predict it either
  (standing, drafting tables) although
  qualitatively it came out
• Speed didnt predict who liked the desk
• 1 who was fastest on it liked it most
• 2 who were slowest on it liked it most

24
Display Comparison - Sorting
25
Sorting Times

• On average, people are slowest on the desk
• However, 5 people did their fastest version of
  the building sorting task on the desk
• 3 more people were faster than on the tablet
  with the buildings (8/12)

26
(No Transcript)
27
(No Transcript)
28
(No Transcript)
29
Observations

• Why the desk is hard
• PowerPoint is designed for the mouse
• Selecting and dragging with pen hard
• Body language
• Body size matters
• smaller people make bigger movements and have a
  harder time reaching
• Architects are a physical group of people
• They like to hang on to the equipment
• They like to move around

30
Video 6 minutes

• People struggling with the pen
• Body language
• Testimony
• Responses to What else would you like to tell me
  about sorting on the desk?

31
Future Work

• Analyze sketches (volunteers?)
• Real quantitative analysis
• Next year ethnography of practicing architects

32
Implications

• Is bigger better?
• The drafting-table size is comfortable for some
  designers
• Trade-offs between taking it all in at once and
  physic
• You tell me?
• Disucssion

33
Thanks!

• Jonathan Henke
• Roberta Stern, Bryan Lewis, and Kevin Head(SIMS
  IT staff)
• Jason Hong, Jimmy Lin, Mark Newman