Interpersonal Relationships in Group Interaction in CSCW Environments

1
Interpersonal Relationships in Group Interaction
in CSCW Environments

• Yang Cao, Golha Sharifi,
• Yamini Upadrashta, Julita Vassileva
• University of Saskatchewan,
• Canada

2
Outline

• Introduction
• Game design
• Rules
• Experiments
• Results
• Conclusions
• Future work
• Link to workshop questions

3
Introduction

• Social factors in multi-user environments
• user motivation, attitudes to others, personal
  relationships and social networks
• emerging, self-organizing social dynamics
• how environment mediates is important
• We are interested to find out
• how people develop and change their attitudes of
  liking or disliking other people
• how the motivation influences attitude change
• how the design of the environment influences
  attitude change and the emergent social fabric of
  the group
• Tool a new multi-user web-based computer game

4
The game
5
Game design (1/2)

• Goal
• To send a packet to a given other player with
  minimum loss.
• Game Description
• A player chooses a destination player and sends
  to him/her a signed packet
• It can send it only by passing it to one of the
  other players.
• The selected player can (depending on whether
  s/he dislikes or likes the originator of the
  packet).
• destroy it completely
• take away a part of the packet and pass it to
  another player
• leave it untouched and pass it to another player

6
Game design (2/2)

• This continues until the packet reaches the
  destination or is destroyed.
• After each round the player can
• see if his/her packet has arrived entirely or
  partially (proportion of 100).
• see a system generated rough representation of
  the attitudes of other players towards him/her
  (system model)
• change his/her attitudes to the other players.

7
Animation
Scenario 2I like A, so I wont destroy her
packet and I like D more than C, so I send the
packet to D
Scenario 3I like A, so I wont destroy her
packet but I like C more than D, so I will send
the packet to C
Scenario 1I dont like A, so I will destroy her
packet. (End of round)
This has to go to D
I like B more than C, so I send it through B
Send this to D
The packet reach the destination (End of round)
The packet reach the destination (End of round)
I like A, so I wont destroy her packet and I
Dont dislike D so I send the packet to D
A sends a packet to destination D
(B)
This has to go to D
Sender (A)
Destination (D)
(C)
8
Implementation

• Web-based (Apache Tomcat)
• Multi-Agent Architecture (FIPA)

9
Rules (1/3)

• A Personal Agent (PA) represents each player in
  the game
• The PA maintains a list of attitudes
• a1,, ak of the player towards the other k
  players, ai ? 1,2,3,4,5,where 1 means "dislike"
  and 5 means "like"
• PA sents the packet to the agent of the most
  liked player M aM maxi a1, a2, , ak

10
Rules (2/3)

• The PA cannot send its packet to an agent that is
  strongly disliked by the user (ai 1)
• The PA of the player who originate the packet
  cannot send its packet directly to the
  destination
• If the player dislikes strongly the originator R
  of the package (aR 1), the PA will destroy the
  packet and the packet will not be passed further.
• Otherwise, the PA takes away n parts of the
  package where n 5 aR and aR ? 2,3,4,5

11
Rules (3/3)

• The round finishes when the packet reaches the
  destination player or is destroyed.
• The player that has accummulated a highest score
  of passed packages wins the game.
• The PAs do not reveal the attitudes of their
  users to either other agents or to the system.
• Players can view their own attitudes towards the
  others at any time (player model).
• At the end of the round, each player can see the
  system model, which is computed by observing the
  passing of the package.

12
Using the game as a tool to study attitude
formation

• The initial attitude-setting in a group
• How significant is the impact of individuality in
  attitude change
• The impact of different system feedback and
  visualization
• The impact of different user motivations

13
Hypotheses

• Individuals react differently, but consistently
  to success and failure when changing their
  attitudes to the other people involved in the
  situation
• People reciprocate the attitudes of other people,
  when they become aware of them
• The feedback about other peoples attitudes is
  given plays a role in the way people reciprocate
  and in the dynamics of the attitudes.

14
Two experiments with 2 versions

• Text feedback version
• 6 participants played 50 rounds
• Questionnaire in the end

• Emoticon version
• 7 participants played 40 rounds

45 minutes, 5-6 players at any given time Players
had different gender, age, ethnic background
(ignored) Players did not know who is who
(aliases used in the game). The players were
given a general introduction about the basic
rules.
15
Results how people choose initial attitudes to
another player?
participants
Level of liking
16
Results dynamics of attitude change
17
Examples of attitude evolution
18
Another example of evolution
19
Typical reactions

• Drastically reducing level of liking as a result
  of failure / partial failure in a game-round
• Frequent for particular players
• Targeted towards one most liked player
• Targeted towards all most liked players

20
More typical reactions

• Reciprocation
• Changing ones own attitude to another player to
  match the attitude of the other player
• Comparing the mutual liking evolution curves for
  pairs of users ? pattern of delayed reciprocity
• Example
• Pronounced difference between the two versions
• An average of 43.7 (median 50) reciprocating
  changes across the players in the text feedback
  version and
• An average of 77 (median 73) of reciprocating
  changes in the emoticon version.

21
Discussion

• Individuals react differently, but consistently
  to success and failure when changing their
  attitudes to the other people involved in the
  situation
• People reciprocate the attitudes of other people,
  when they become aware of them
• The way feedback about other peoples attitudes
  is given plays a role in the way people
  reciprocate and in the dynamics of the attitudes.

22
Conclusions

• Multi-player games offer a tool for studying the
  social dynamics of a group
• Individuality plays a significant role
• It is possible to define typical reactions
• More work needs to be done to generate
  constructive results that can guide system design
  

23
How the paper addresses the WS questions
1 Taxonomy of Circumstances Requiring Affective
and Attitude User Modeling - in
multi-user virtual environments, collaborative or
not - the social experience is the
determining factor for success2 Existing
methods of Constructing Affective/Attitude User
Models - modelling relationships /
attitudes among users 3 Validation and
Evaluation - through the use of social
(multi-player) games 4 Guidelines for model
use - adapting the feedback and visualization
24
Future work

• The impact of the user motivation for
  participation (e.g. Win the game vs. Play the
  game) will be investigated
• Experiments with more participants by opening the
  game to players on the web
• To ease data analysis, synchronous rounds will be
  used
• To pinpoint the reason for changing attitude,
  user interviews and video observations, think
  aloud protocols will be used
• The role of the amount and the presentation of
  feedback information on the attitude formation of
  the user will be investigated further

25
Interpersonal Attitudes
Not necessarily reciprocal
So, each relationship is subjective,
uni-directional
26
Player Model System Model (textual feedback
version)
27
Player Model System Model (animated emoticon
version)
28
Reciprocation example
Text feedback version
Emoticon Feedback version