Groupware Re-Design: Analysing discourse

1
Groupware Re-Design Analysing discourse
2
Common Ground (Shared Context)

• What accumulates in a joint activity is the
  common ground of the participants. For most
  activities, the common ground at any moment
  divides into three parts
• 1. initial common ground. This is the set of
  background facts, assumptions, and beliefs the
  participants presupposed when they entered the
  joint activity.
• 2. current state of the joint activity This is
  what the participants presuppose to be the state
  of the activity at the moment
• 3. public events so far. These are the events
  the participants presuppose have occurred in
  public leading up to the current state

3
Groupware

• Support groups of people engaged in a common task
  (or goal)
• Provide an interface to shared environments
• Facilitate communication, coordination, and
  collaboration of group effort
• Same time/ Different Place
• All of interaction data available

4
GroupwareIssues Approaches

• Requires analysis of work environment and design
  of mediated interaction among users
• General-purpose methods support communication
• Email, Whiteboard, Chat room
• With practice/ Within Context
• Recurrent behaviors (convention)
• Coordinating Representation
• Add shared pre-designed and formatted
  representations to mediate recurrent problem
  areas of coordination

5
Basic Methodology

• Base system includes only general purpose
  coordination methods (whiteboard, chat)
• Sometimes this is enough
• Otherwise, analyze interaction
• Rebuild systems using coordinating
  representations (and selectively add features)

6
Analyzing the interaction

• Analyze coordination work as it reflected in
  discourse of participants
• Recurrent patterns of coordination
• Repeated errors or trouble spots in coordination
• Development of organizational structure

7
Talk is Action

• Participant design locally managed highly
  Interactionally managed
• Turn-Taking
• Next speaker can be selected by previous one
• A speaker can self-select
• Speaker can continue speaking
• Sequentially organized Adjacency Pairs
• Do you know x?
• Hello
• Third Position Repair
• Intersubjective understanding of common world

8
Talk Action

• Cooperating actors require communication to stay
  coordinated
• Difficulty at coordination and the amount of talk
  in coordinating a joint behavior are directly
  related
• Recurrent activities produce recurrent
  conversations

9
Why talk produces structure

• Organize domain activity
• Organize conversation
• Talk is action
• Action requires coordination
• Recurrence
• Convention

10
(No Transcript)
11
Talk in VesselWorld

• Maintain consistent representation of shared
  domain objects
• Produce structure to organize both
• Conversation
• Domain Activity

12
Maintain Consistent Representation of Shared
Domain Objects

• Report
• Confirm
• Review
• Repair

• Waste at 554,41 is small, dredge
• New XL! At 200 431
• I guess Ill sweep the bottom, west to east
• all remove marker at 100,425
• 8 clicks for me to hit BB
• Ill grab sX at 500 275
• The first on the barge is sX
• w8ting
• killed sX at 500,275
• I still and 2 or 3 moves till I get there
• Im waiting at SW corner
• I have a leaker on my hands

13
Review

• 1. Crane1 center is clean nothing in the
  center 4 quads
• 2. Crane2 ok, we forgot the west, I didnt get
  down that far
• 3. Tug1 600-400 square is clean, I dont have to
  worry about tit?
• 4. Crane2 correct, thats where I started
• 5. Crane2 about 100 N of sb
• 6. Tug1 okay cool, grabbing the sb

14
Repair

• 1. TUG1 mX at 400 125
• 2. CRANE1 medium at 392 127
• 3. CRANE1 thats got to be the same one
• 4. TUG1 yep
• 5. TUG1 thats an mX
• 6. CRANE2 so you guys are in S?
• 7. TUG1 yes
• 8.
• 9. CRANE1 Im going to trash the mX on the BB

15
Structures Produced to Organize Talk

• Size of waste
• xl (extra large), l (large), m(medium), s(small)
• Equipment
• X (no equipment), D (dredge), N (net)
• The location is indicated by a x and y
  coordinates 400 125.
• Examples
• mX at 400 125
• mN at 150 200
• xl D at 200 425

• Notational conventions
• Marker Check procedure
• Adjacency pairs for close coordination

16
Marker Check

• 1. Crane1 ALL ok I will dump all the markers
  ok with every1?
• 2. Crane1 Legend SmLXL Ni no idd
  Net Dr
• 3. Crane1 from south east clockwise
• 4. Crane1 Sm-Ni 50,0Sm-Net 150,25Sm-Ni
  350,150Sm-Ni 550,50 Sm-Ni 600,100 Thats
  all south of equator. NORTH coming up.
• 5. Tug1 97441 and 72,368 already IDd
• 6. Crane2 350,150 is barge, isnt it?
• 7. Crane2 thats the problem
• 8. Crane2 stop dump I was there to thats the
  Sbarge at 350,150
• 9. Crane2 confirm with TUG
• 10. Crane1 xL-Ni 475,425Sm-Ni 450,450Sm-Ni
  525,500Sm-Ni 250,500XL-Ni 200,475 Sm-Net
  100,425Xi-net 75,375Sm-Ni 25,575
• 11. Crane1
• 12. Crane1
• 13. Crane1 - END Tug, confirm you have all
  those

17

• 14. Tug1 large barge at 400,325
• 15. Crane2 repeat I say the sbarge ar 350,150,
  not a small waste
• 16. Crane2 TUG where are the small barges?
• 17. Crane1 c2 you sure? If you saw that, ok
  ill cancel as waste
• 18. Crane1 of yeah tug can tell us
• 19. Tug1 350,150 is small barge
• 20. Crane2 ok, 12 wastes and no problems, lets
  get back to work
• 21. Crane1 okdoke. Still sweeping west

18
Adjacency pairs for close coordination

• 1. Crane1 sub lift
• 2. Crane2 LL
• 3. Crane2 k
• 4. Crane2 sub load
• 5. Tug1 the next XL needs nothing
• 6. Crane1 k
• 7. Crane2 ok, then XLD right?
• 8. Crane2 sub Lift
• 9. Tug1 yep
• 10. Crane1 k
• 11. Crane2 sub load
• 12. Crane1 k
• 13. Crane2 sub sep
• 14. Crane1 sep

19
Coordinating Representation

• Ubiquitous in everyday world
• Stop sign scene at airport
• Representation specifically designed to fix
  recurrent problem of coordination
• Functions to simplify co-referencing, timing of
  joint actions, exchange of information

20
Three Coordinating Representations

• Establishing references for, and exchanging
  information about, shared domain objects and
  their status
• Timing of closely coupled cooperative activities
• High-Level Planning

21
Object List
22
Timing of closely coupled cooperative activities
23
High-Level Planning
24
Evaluation

• Compare groups with/without Coordinating
  Representations
• Training 10 hours of problem-solving
• 3 groups without 3 groups with CRs
• 49 improvement in clock time
• 38 reduction in the number of events generated
• 57 reduction in the amount of electronic
  chatting
• 61 reduction in total errors
• The high-level planning CR was not used

25
Analyzing Referential Structure of Discourse
Iota name
Timeline for iota
Iota details
Utterance mentioning this iota (here, a proposal
for a plan involving IOTA-97

• Plans have short lifespan, very high density
• Wastes have long lifespan, short density

26
Outline

• Talk Action
• VesselWorld
• Coordinating Representation
• Developing groupware systems
• Method based on analysis of talk
• As applied to VesselWorld
• Evaluation

27
Research Questions

• Comparing coordinating representations to
  free-forms of communication.
• Methodology

28
Summary

• Attitude
• Groupware Systems
• Talk Action
• VesselWorld
• Coordinating Representation
• Evaluation

29
Why talk produces structure

• Talk is action.
• Action requires coordination.
• Recurrent coordination results in conventional
  organizations that simplify
• Conversation
• Domain Activity

30
Why?

• Previous analysis focused on coordination work
• Hypothesis Most of these problems are
  predictable from a analysis of the cognitive
  workload in sharing info
• Key referential structure of user chat can be
  used to predict cognitive workload

31
Data Summary

• Repairs have very short lifespan, very high
  density
• Plan iotas have short lifespan, high density
• Waste iotas have long lifespan, low density
• Locations have moderate lifespan, low density
• Conventions, barges, and vessel iotas have long
  lifespan, low density

32
(No Transcript)
33
Experiment Design

• Apply basic methodology
• Compare groups with/without Coordinating
  Representations
• 3 groups without 3 groups with CRs
• Training 10 hours of problem-solving

34
Groupware

• Support groups of people engaged in a common task
  (or goal)
• Provide an interface to shared environments
• Facilitate communication, coordination, and
  collaboration of group effort
• Requires analysis of work environment and design
  of mediated interaction among users
• Also analysis of practice that emerges

35
Notational Conventions

• Size of waste
• xl (extra large), l (large), m(medium), s(small)
• Equipment
• X (no equipment), D (dredge), N (net)
• The location is indicated by a x and y
  coordinates 400 125.
• Examples
• mX at 400 125
• mN at 150 200
• xl D at 200 425

36
Confirm

• 1. Crane1 done! Now we go to the small wastes
• 2. Crane2 split and go find other waste now huh
• 3. Crane1 yep
• 4. Crane2 here we go
• 5. Tug1 Cool. I left a center marker on my
  screen at 100h/380v The wastes are around that
  marker all of them re small.
• 6. Crane1 i marked both, ill head to the one i
  need to get
• 7. Tug1 And Ill hang out and you can load on
  me.

37
Other Work in Lab

• Tools to Analyze Interaction
• Component-Based Toolkit
• Adaptive techniques
• Web Design Tool
• Collocated pair programming
• In Class projects
• Cognitive Model/Software Agent

38
Research Questions

• Comparing coordinating representations to
  free-forms of communication.
• Methodology

39
Analyzing the interaction

• Analyze coordination work as it reflected in
  discourse of participants
• Recurrent patterns of coordination
• Repeated errors in coordination
• Development of secondary of structures
• Analyzing Cognitive Workload (wrt sharing info)
• Model reference structure of discourse
• Compute cardinality, density, and duration of
  topics

40
Together these two lines of argument produce the
argument

• Recurrent user activity depend on the recurrent
  conversations that occur to organize those
  activities.
• Organizing the conversation about recurrent
  activities not only organizes the conversation
  but it organizes the activity

41
Analyzing the interaction

• Analyzing Coordination Work
• Recurrent patterns of coordination
• Repeating errors
• Development of secondary structures
• Analyzing Cognitive Workload (wrt sharing info)
• Model reference structure of discourse
• Compute cardinality, density, and duration of
  topics

42
Coordinating Representations

• CRs are external representations that were
  created in a prior activity and not on the fly in
  the current one.
• CRs are created by non-participants of the
  current activity (the intergenerational bit).
• Not all external representations function to
  organize a cooperative activity (but many do).
• CRs are explicitly designed to fix a recurrent
  problem of coordination.

43
Close Coordination

• 1. Co-ordination of talk
• a. Adjacency pairs to propose and confirm next
  action
• b. Expectations that adjacency pairs will occur
  for each of the actions in an extended sequence
  of tightly coupled cooperative behaviors
• 2. Coordination of action

• 1. Crane1 sub lift
• 2. Crane2 LL
• 3. Crane2 k
• 4. Crane2 sub load
• 5. Tug1 the next XL needs nothing
• 6. Crane1 k
• 7. Crane2 ok, then XLD right?
• 8. Crane2 sub Lift
• 9. Tug1 yep
• 10. Crane1 k
• 11. Crane2 sub load
• 12. Crane1 k
• 13. Crane2 sub sep
• 14. Crane1 sep

44
Experiment

• Compare groups with/without Coordinating
  Representations
• Training 10 hours of problem-solving
• 49 improvement in clock time
• 38 reduction in the number of events generated
• 57 reduction in the amount of electronic
  chatting
• 61 reduction in total errors
• One of CRs wasnt used some features of other
  CRs were not used
• Hypothesis Function of cognitive workload (wrt
  sharing info)

45
Shared Domain Objects

• 1. Harbor
• a. Areas that have been searched
• 2. Wastes
• a. The locations of the wastes
• b. References to the wastes
• c. Equipment
• d. Size
• e. Whether or not it is leaking
• 4. Large Barge
• a. Location
• b. What is on it
• 5. Equipment
• a. Who has which equipment (this is permanent)
• b. Whether the equipment is currently deployed
• 6. Other Actors
• a. Their location
• b. Their intended future action for (or recently
  completed action with) some shared domain object

46
Rounds and Events

• Segments of activity are divided into rounds.
• During a round of activity, any number of events
  may occur
• Once a participant has submitted to the system
  her next action, she can no longer change it.
• When all three participants have submitted
  actions, the round ends, the system updates the
  state of world, and the next round begins.

47
Secondary Structures

• Created by participants to simplify talk at
  critical junctures of cooperative activity
• Indicate problem areas where the introduction of
  a coordinating representation could be productive

48
Predictions from analysis

• Information that is accessed frequently
• plans, repairs, wastes
• Information that is relevant for a long time
• wastes and harbor not plans
• Information that is too complex to be easily
  retained in working memory
• waste details (not waste existence) plan details

49
Predictions from analysis

• High scores in any or all dimensions may indicate
  difficulty in sharing information
• accessed frequently
• plans, repairs, wastes
• relevant for a long time
• wastes and harbor not plans
• complex representation
• waste details (not waste existence) plan details

50
Results

• Shared Planning
• Plans have a short lifespan of relevance, but
  dominate conversation for that time. They
  generally are not reviewed.
• A non-persistent, easy to author, easy to access
  CR is indicated chat may be sufficient for
  non-synchronized actions
• Object List
• Waste iotas have long lifespan of relevance, with
  segments of frequent reference interspersed with
  long periods of irrelevance
• A persistent, easy to access, and easy to modify
  CR is indicated Object List meets these
  criteria, and was accepted
• High-level Planning
• Very few plans have a long lifespan of relevance
• Cost to create external representation of plans
  was high, offsetting possible benefits from
  making a persistent representation for later
  reference.

51
Results, cont.

• Harbor status
• Many groups came up with secondary structures to
  ensure that the harbor was fully cleared
• Harbor and location iotas referred to
  occasionally throughout entire session
• A persistent, easy-to-access, easy-to-update CR
  is indicated

52
With Coordinating Representations

• Close coordination
• Reduce conversation Catch each others error
  Easier to synchronize
• Object list
• Reduced work of maintaining consistent
  representation Not all columns used equally
  secondary structures
• High-Level planning

53
(No Transcript)
54
Analysts Tool Overview

Coordinating Representation Wizard
Semi-automatic Structure Extractor
Query Construction Tool
Segment Comparison
Collect Related Collaborative Activities
Tagging Chat
Statistical Analysis Mechanisms
Segmentation
Quantitative Analysis Tools
Qualitative Analysis Tools
SAGE (Playback)
THYME
55
SAGE (Replay)
56
Tagging and Segmenting
57
Segment Comparison View
FRAME INFORMATION QUERY -------------------------
---- SLOT 1 NAME Request for information COMPLI
CATIONS no response (busy/ignoring)
(A3-A4) self-correction (B2-B3) SECONDARY
STRUCTURES none recorded yet ------------------
----------- SLOT 2 NAME Information
returned COMPLICATIONS mismatch of common
ground (B6-B9) wrong information returned
(A3, A6) SECONDARY STRUCTURES jargon
(B2-B3 B4)
Segment A what size is it segment
Segment B are you sure thats net

1. . . .
2. tug1 waste at 300/125 needs net
3. crane1 what size?
4. crane1 tug1 What size is the waste at 300,
   125?
5. tug1 waste at 320/98 needs nothing
6. tug1 net
7. crane1 tug1 not equipment, what size is it?
8. tug1 small waste at 300/125

1. . . .
2. crane1 whats the waste at 300 190?
3. crane1 oops 190 300
4. crane2 ln_at_300,190!l
5. crane1 thanks
6. tug1 Are you sure thats net?
7. crane2 I thought so
8. tug1 I had no equipment. Let me check before
   you get it.
9. crane1 ok.

58
Summary

1. Groupware Systems
2. Cognitive Foundation
3. In Practice

59
Our Approach

• Leverage users participation to help make system
  tailor-made
• Participants design structures to organize their
  activity
• Secondary structures emerge that organize
  recurrent group activity
• Realize design of activity in a coordinating
  representation (e.g., stop sign)
• Methodology for doing this and tools to rapidly
  build and rebuild groupware systems

60
Analyze Discourse(part one)

• Examine free forms of communication during
  activity
• Look for
• Recurrent problem areas of coordination
• Recurrent errors
• Secondary structures produced by participants

61
Methodological Comments

• Object list (indicator 1)
• In more extensive data one of the groups created
  a secondary structure
• Shared Planning Window (indicators 2 3)
• High-Level Planning (indicator 1)
• No secondary structures in more extensive study
• Too much functionality (?)
• Short period of relevance of information (?)
• AI techniques

62
System View

Analysts Tool
VesselWorld
More Groupware Systems
Web Design Tool
Adaptive Components
SAGE (replay tool)
Basic Groupware Tools (chat room, shared
whiteboard, etc.)
THYME
Key Coordinating Representations
63
GROUP(Suite of tools, methods, and applications)

• Coordinating Representation
• Experimental Platform (VesselWorld)
• Tools to Analyze Interaction (Feinman, Hao)
• Component-Based Toolkit (Landsman)
• Adaptive techniques (Introne)
• Performance Benchmarks (Storer)
• Web Design Tool (Hickey, Langton, Svetlena
  Taneva)
• Collocated pair programming (Head, Granville)
• In Class projects (Fall)

64
Schedule

• Overview (Rick)
• Demos
• VesselWorld
• Second Version of Analysts Tool
• Talks ( working lunch)
• Analyzing cognitive workload for maintaining
  common ground (Feinman)
• Toolkit for building groupware applications
  (Landsman)
• Using CRs to reduce workload in managing
  information (Introne)
• Performance Benchmarks (Storer)
• Another Example WebTool (Hickey)
• Collocated pairs programming (Head)
• Wrap Up

65
Object List
Timing of Joint Actions