Professional Documents
Culture Documents
{ben.daniel,gordon.mccalla,richard.schwier}@usask.ca
Outline of Presentation
• Background
• Social Capital (SC) in Virtual Learning Communities (VLCs
-Purpose
-Problems
•Research Goals
•Fundamental Variables of SC in VLCs
•Modelling SC using Bayesian Belief Network (BBN)
•Overview of BBN
•Constructing and Updating a BBN
•BBN Model of SC in VLCs
•Empirical Analysis to Update the Model
•Examples of Scenarios and Results
•Methodological Challenges
•Summary
Background
• A VLC is a group of people who gather in cyberspace with
intention of pursuing learning goals
• SC is a connections among individuals—social networks and
norms of reciprocity and trust worthiness that arise from them
(Putnam, 2000)
• SC is a residual side effect of social interaction and the enabler
of future interactions (Resnick, 2002)
• SC is a stock of active connections among people(Cohen &
Prusak 2001):
-It involves trust, mutual understanding, respect and
shared values and behaviours within a community
What is SC in VLCs?
• SC is a web of positive or negative relationships which can
facilitate information exchange, knowledge sharing, and
knowledge construction (Daniel, Schwier & McCalla,2003)
• SC is enhanced through continuous interaction, built on trust
and maintained through shared understanding
• SC exhibits multidimensional, multilayered, and multivariate
properties and variables
Why Study SC in VLCs?
Sample Transcript
Web of + or –
relationships
Message
Results Scapital Model
Sentence
Ways to study SC Tools to support Codes
in VLCs SCapital Cluster 1
Themes Cluster 2
Cluster n
BBN SC Model
Evaluate
No
Message
Yes
Determine Graphical
Structure
Unit of Analysis Paragraph Semantics
No
Evaluate
Sentence
Yes
Code(s)
Elicit & Compute Initial Cluster 1
Probabilities
No
Evaluate Themes Cluster 2
Yes
Build Scenarios to
Update Model Cluster n
No
Evaluate
Yes
Conduct Sensitivity
Analysis
l
de
o
M
No
Evaluate
e
at
d
p
Yes
u
Validate Model
to
s
io
ar
en
c
S
No
Evaluate
Yes
Computational Model
BBN Model of SC
Analysis of SC in a VLC: Study 1
• Purpose
-Preliminary examination of how SC relates to knowledge
sharing in actual VLC and identify additional variables
• Goals
-Understand the process involved in knowledge creation in
VLCs
-The role of different variables that can influence the process of
knowledge creation in VLCs
• Procedure
-A content analysis method to examine a body of interaction
transcripts taken from a graduate course
-A grounded theory approach to analyze and interpret data
Data Sources & Research Context
• Transcripts of online discussions
• Transcripts of email
• 12 students enrolled in a 6 credits graduate course in the
foundation of educational technology
• Online discussions were scheduled for each topic in the
course
• Synchronous and asynchronous tools were used for over a
two-week period
Data Analysis Procedures
• Coding done in Atlas ti Qualitative Data Analysis Software
-Purposeful coding for anticipated variables
-Grounded theory approach for novel variables not earlier
identify in the model
-Transcripts were coded by one researcher
-Inter-coder reliability estimates were not calculated
Total Frequncies of
Observations
0
100
200
300
400
500
600
700
Interaction
Professional
Awareness
Capability
Knowledge
Awareness
Demographic
Awareness
Technology
Community
Variables
Hospitality
Shared
Key Findings of Study 1 Cont..
Understanding
Information
exchange
Social Capital Discourse Variables in a Formal Virtual Learning
Social
Protocols
Analysis Interactions in a Video-Mediated
Virtual Community: Study 2
• Purpose
-To understand interaction patterns in video-mediated VCs
• Goals
-Identify different themes that emerge from interaction in a
video-mediated virtual community
-Identify the indicators of SC in this community
• Procedure
-Employed a content analysis to examine a body of interactions
-A grounded theory approach to analyze and interpret data
Data Sources and Research Context
• Data obtained from transcripts of interactions
• The data sets were collected daily, in one half hour segments
at different hours over a period of one month during the
summer of 2002
• Results reported here are only from one data set drawn a
particular VC (“Café Americano”)
Total Number of
Utterances Observed
0
100
200
300
400
500
Demographic
Awareness
Economic
Food
Information
Exchange
Social
Technology
Problems
Community
Community
Language
Hospitality
Utterance Categories
Similes
Informal
Study 2 Key Findings…1
Language
Visual
Awareness
Interaction
Study 2 Key Findings..2
• Results revealed sustained level of interactions (SC
characteristic)
• Discourse themes were variant
• Discourse of social issues seems to be dominant
• Emergent community language
• There is evidence of high level of awareness (SC
characteristics)
Examples of Scenarios
Scenario 1
Imagine a virtual learning community of graduate students with diverse professional
training and cultural backgrounds. Members of this community are not exposed to each
other in face-to-face settings. The community has a strict formalised structure and goals of
interactions. Members of the community are all motivated to learn more about the subject
domain and determined to successfully complete the course. As interaction progress
individuals begin to disagree more on issues under discussion either because they disagreed
with their colleagues’ opinions, or their understanding of the issues are grounded on their
prior experiences. What is the SC of such a community?
Scenario 2
A distributed community of practice of software engineers with goals of sharing information,
and providing peer-support to each other. Members in this community shared common
concerns and are drawn from all over the world, and are composed of highly experienced
software developers and novices. Evidence of positive interactions was observed, and
individuals eventually got exposed to each other and began developing social networking by
privately exchanging personal information among themselves. There was reasonable level of
shared understanding since individuals are either professional software engineers or students
but formal social norms for interaction were not explicitly stated and most of the activities and
discussions were informally organized. What is the level of SC in this community based on
this scenario?
Results of Scenario 1 after Propagation
Results of Scenario 2 after Propagation
Methodological Challenges
• Research into SC in VLCs, is still new with lots of challenges
• The method employed in studying and analyzing natural
conversations in VLC faces reliability issues
• Though internal validity can be proven, claiming external validity
might be challenging to achieve
• The fundamental limitations of BBN relate to:
-Learning the structure and potentials for each node in the net
-It is always necessary to provide training data to learn the
structure and parameters of the BN and or to learn data from a
network topology
-A significant effort is required to identify accurately the
conditional probability of each node in a network
-The power of the model depends on the accurate identification of
prior probabilities and the sensitivity analysis of each variable in
the network
Summary of Results
• SC is a multivariate construct
• In VLCs, trust remains a fundamental variable of SC but it is not
the only proxy for observing the presences of SC
• Different forms of awareness are critical to understanding SC in
VLCs
• SC is both a collective property of a community as well as the
individuals in the community
• Shared understanding in a community is an important ingredient
for maintaining SC in VLCs
Thanks!
Questions and Discussions
Laboratory for Advanced Research in Intelligent Educational Systems (ARIES) Research Group1,
Virtual Learning Community Lab (VLC), Educational Communications &
Technology2
University of Saskatchewan, Saskatoon, Canada
{ben.daniel,gordon.mccalla,richard.schwier}@usask.ca