Analysis Of A Learning Community As A Social Network

Cypriot Journal of Educational
Sciences
4 (2009) 46-57
www.world-education-center.org/index.php/cjes
Analysis of a Learning Community as a Social Network
Nina Bijedica
*, Indira Hamulicb
a
E-learning, M. Tita bb, Mostar 88104, Bosnia and Herzegovina
b
Mathematical modeling, M. Tita bb, Mostar 88104, Bosnia and Herzegovina
Received November 13, 2008; revised May, 2008; accepted June 02, 2009
Abstract
Online knowledge-sharing communities are usually small. In this paper, we present the challenge of analyzing a
large online community in order to determine if it is feasible for knowledge-sharing. We deploy social network
techniques to analyze patterns of interactions critical to information- and knowledge-sharing among learners in a
virtual community. Based on this, we determine the characteristics of the network and the roles of the actors. The
research was carried out at the Faculty of Information Technologies in Mostar (FIT), after the end of the 2007/8
academic year, during which 293 freshmen were enrolled. We collected data from the FIT Community Server
(FITCS), for the period from 1.10.2007 until 30.9.2008, and modeled them into three networks (N, N1 and N2).
Network N (overall communication) has 273 vertices, N1 has 143 (fall semester course in Programming), and N2 has
99 (summer semester course in Programming). With regard to the number of enrolled freshmen, we estimate that
85 to 90% communicated via FITCS. The characteristics of the analyzed networks are as follows: density varied from
0.17 to 0.25, average distance from 1.80 to 1.93, cohesion from 0.53 to 0.63, betweenness varied from 7.83 for N to
56.23 for N1, and closeness varied from 52.66 to 56.78. The results for the characteristics of selected actors are
degree (ranging from 10 to 119), betweenness (ranging from 0 to 88.17), and closeness (ranging from 16.93 to
60.62). The results of this research show that some educators do not have a proper roles in the online knowledge-
sharing community, but that can be due to the fact that the actors are freshmen. On the other hand, some of the
most successful students, with extrovert personalities, were the stars of the three analyzed networks. Therefore, we
can conclude that the analyzed part of FITCS is a knowledge-sharing community. Recommendations are that we
should motivate educators to support online knowledge-sharing, that we should educate educators about their
proper role in such a community, and that we should motivate successful students with introvert personalities to
be more active in knowledge-sharing.
Keywords: Online Community; Social Network Analysis; Knowledge-sharing
© 2009 Academic World Education & Research Center. All rights reserved.
* Corresponding author. Tel.: +38761203214; fax: +38736570730.
E-mail address: nina@fit.ba (N.Biyedic), irea@irea.uvt.ro (I.Hamulic)

Nina Bijedic & Indira Hamulic / Cypriot Journal of Educational Sciences 4 (2009) 46-57
47
1. INTRODUCTION
1.1. Related research
Network analysis in the social sciences developed from conjunctures about anthropologist's
observations about relations in face-to-face groups and mathematical graph theory. Consequently,
a very large part of social network methodology deals with relatively small networks. This follows
from the fact that we have to have confidence in the reliability of our observations about the
relations among the actors in the network. Most of the tools of social network analysis involve the
use of mathematical functions to describe networks and their sub-structures.
Recently, however, some of the focus of social network research has moved away from these
roots. Increasingly, the social networks that are being studied may contain many nodes; and,
sometimes our observations about these very large networks are based, not on censuses, but on
samples of nodes. Network researchers have come to recognize that the relations that they study
may be constantly evolving, and that the relations observed at one point in time may not the
entirely typical because the pattern of relations is not "in equilibrium." They have also recognized
that sometimes our observations are fallible -- we fail to record a relation that actually exists, or
miss-measure the strength of a tie.
On the other hand, when it comes to learning, some of the recent research reveals that, in
community settings, people learn as they solve problems together (Koschmann, Kelson, Feltovich
& Barrows, 1996) or design representations of their understanding (Suthers, 1999). It also turns
out that much learning between and within communities occurs with boundaries that are rich in
interactions, whether formal, informal, or through a computer-based system (Wenger, 1998).
Consequently, the processes of learning in virtual learning communities can also be improved if
the different actors are known. In particular, in virtual learning communities where learners are
often isolated from each other and their instructor, knowledge-sharing is fundamental to effective
learning. Therefore promoting knowledge-sharing requires an understanding of social
relationships and connections that are critical to information and knowledge-sharing.
In order to understand information flows and knowledge-sharing, we have to deploy social
network analysis (SNA). SNA is the study of mathematical models for interactions among people,
organizations and groups. According to SNA theory, social relationships are viewed in terms of
nodes and ties (edges). Nodes are individual actors within the network, and ties represent the flow
of relationships between the actors. These relationships, defined by linkages among units/nodes,
are a fundamental component of SNA (Shetty and Adibi, 2004).
Historically, research in the field has been led by social scientists and physicists (Lorrain and
White, 1971; Wasserman and Faust, 1994) and previous work has emphasized binary interaction
data, with directed and/or weighted edges. Using pure network connectivity properties, SNA often
aims to discover various categories of nodes in a network. Using network properties in SNA, we
can assign “roles” to certain nodes, e.g. Wolfson and Willinsky (1998).
The shape of the social network helps to determine a network's usefulness to its individuals. For
instance, smaller and tighter networks can be less useful to their members than networks with lots
of loose connections (weak ties) to individuals outside the main network. Moreover, networks
with many weak ties and loose social connections, are more likely to introduce new ideas and
opportunities to their members than closed networks with many redundant ties. In other words,
individuals who have similar knowledge, interests, and personal attributes can easily connect and
share knowledge. In addition, those individuals with connections to other network are likely to
have access to a relatively wide range of information and knowledge.

48
Online knowledge-sharing communities are usually small, numbering up to thirty students. In
this paper, we present the challenge of analyzing a large online community, containing more than
one hundred students, in order to determine if it is feasible for knowledge-sharing.
We have deployed social network techniques to analyze patterns of interactions critical to
information- and knowledge-sharing among learners in a virtual community. In accordance with
this, we have determined the characteristics of the networks and roles of the actors.
1.2. Theoretical background
SNA is a methodology for the collection and analysis of relational data. In other words, SNA is
concerned with observations, interviews, and archival records (Wasserman and Faust, 1994).
Additionally, archives of e-mails, chat logs, and web postings provide SNA with rich sources of
information for analysis.
In order to define fundamental concepts in SNA, we use terminology and methods from graph
theory. In this manner, relationships between nodes can be translated to matrices, and then a set
of measures can be calculated using those matrices. Using SNA we can measure many structural
characteristics of the network, such as the existence of subgroups, the relative importance of
individual actors, and the strength of the links (Wasserman and Faust, 1994).
The three essential components of a social network are a set of nodes (or actors) and arcs (or
links) which are a mathematical model of a graph, a sociogram or a graph (created with the nodes
and arcs) and a sociomatrix. For example, in a study of a group of 26 teenagers in suburban Dublin,
the teenagers were asked to name their best friends (Freeman, Webster & Kirke, 1998). The
sociogram can be made by labeling dots with the teenager’s names (nodes) and adding lines
between those dots that represent relationships between them (arcs). The sociomatrix would
have row and column labeled with the teenagers’ names and, in each cell, a 1 would indicate that
they are friends and a 0 that they are not.
Some of the most important conceptual ideas derived from SNA are density, centrality, and
cliques (Scott, 2000). Density is a widely used concept in graph theory that we use to measure the
level of linkage between the nodes of a graph. In a complete graph, each node is connected
directly to every other node, and such a graph has maximal density. For incomplete graphs,
density measures how much a graph differs from the state of completion. While density describes
the general level of cohesion in a graph, the extent to which this cohesion is organized around
particular focal points is measured by the centralization factor. The calculation of the
centralization score will give a value between 0 and 1, with 1 achieved in a graph with “star”
structure (with one central node connected to all the other nodes, none of which is connected to
any other node but the central one), and 0 for a complete graph. It may be useful to identify the
structural centre of the graph, which is a single point or a cluster of points that act as the pivot of
the whole graph. Then it is possible to identify marginal and peripheral nodes which have low
centrality scores (Scott, 2007).
It is important to notice that a node has two kinds of centrality measures: local and global. The
local centrality measure shows how many connections a given node has, and is calculated by the
degree of the node. In directed graphs, it is relevant to talk about in- and out-centrality
(corresponding to in- and out- degree in graph theory). The global centrality measure describes
the position of a node in the overall structure of the network. Some authors (Freeman, 1980)
express global centrality in terms of geodesics (shortest distance between nodes, as opposed to
eccentricity in graph theory). The nodes, whose sums of geodesics to all other nodes are the
smallest, are globally central (this is rather different from the term central nodes in graph theory,
where it stands for the set of nodes with eccentricity equal to the radius, the smallest of all
eccentricities).

49
Another concept, which is related to node centrality, is the concept of betweenness. This
measures the extent to which the role of a node is that of an intermediary. In other words, some
nodes may act as “brokers” or “gatekeepers” between groups of nodes, therefore playing an
important role in the network (corresponding to the term “vertex cut point” in graph theory).
“Betweenness is, perhaps, the most complex of the measures of point centrality to calculate. The
‘betweenness proportion’ of a point Y for a particular pair of points X and Z is defined as the
proportion of geodesics connecting that pair which passes through Y - it measures the extent to
which Y is ‘between’ X and Z” (Scott, 2007, p. 87).
“One of the major concerns of social network analysis is the identification of cohesive
subgroups of actors within a network. Cohesive subgroups are subsets of actors, among whom
there are relatively strong, direct, intense, frequent, or positive ties” (Wasserman & Faust, 1994,
p. 249). For SNA theory, a group is a structure that can be discovered empirically by the
examination of patterns of interactions among members of a population. Those subsets of
members that are highly interconnected constitute groups (Garton, Haythornthwaite & Wellman,
1997).
Besides density, centrality, and cohesion measures, SNA comprises a battery of techniques that
have been used in recent studies related to electronic mediated interactions (Aviv, Erlich, Ravid, &
Geva, 2003; Willging, 2004). The e-mail traffic was the data source used to detect communities of
practice and leadership in an organization. Some authors used a betweenness centrality algorithm
to identify communities of practice from e-mail logs (Tyler, Wilkinson, & Huberman, 2003). In this
study, e-mail data was used to create a graph that showed the 185,773 e-mails exchanged during
almost 3 months by the HP Lab’s 485 employees. Using the betweenness algorithm, 66
communities were identified. Using a colour-coded schema for the graph, organization leadership
inferences were made possible by a visual inspection of the resulting graph.
Visualizations are a key tool for analyzing social networks. Recent works on the visualization of
network activity have made evident the practical use of SNA metrics. Krebs (2003) describes how a
project team was monitored using network metrics and the visualizations of e-mail flow, which
provided an x-ray of how the project worked. The x-ray monitoring started after the team missed a
key milestone in the fourth month of the project. Some of the hubs became bottlenecks since they
were under-performing. The health of the project was monitored each month through network
metrics, which revealed emergent leadership and communities of practice. More integration
between the departments was found necessary by project managers and, therefore, more project
team members were assigned. This intervention improved the information flow by reducing the
communication load on the hubs.
During the subsequent eleven months of the x-ray monitoring, no more milestones were missed
(Krebs, 2003). Viegas, Boyd, Nguyen, Potter, & Donath, (2003) created visualizations of the social
rhythms of e-mail exchange, transforming the intangible online digital experience into concrete
images, which can be displayed and evaluated. These authors say that the value of a visualization
tool lies in its effectiveness as a device for information retrieval, and that these visual objects
provide users with the elements needed to explore and reflect about their personal digital
experiences, providing them with memory prompts and allowing storytelling. “Just as photographs
allow individuals to begin relationships by having a mechanism for sharing information about
one’s past, these visualizations provide a tangible link to one’s digital interactions” (p.10).
1.3. Background
The FIT Community Server (FITCS) has existed for several years. It was originally designed for
distance learning students but, in time, it became the most popular way of communication for all

50
FIT students. FITCS consists of several units, one of which is reserved for communication on FIT
courses. This unit was designed for knowledge and information sharing. Naturally, one might ask if
students actually use FITCS for knowledge-sharing and, if yes, to what extent. In order to answer
these questions, it is necessary to analyze if FITCS has the characteristics of a social network, to
identify actors in the network according to their roles, and to check if the roles of the principal
actors are sufficient for participation in a knowledge-sharing community.
After identifying the questions, we designed a scientific project in order to answer them. The
first phase of the project consisted of modeling communication with a graph, and the deployment
of social network analysis in order to check the following hypothesis:
H1: Educators and technical staff have proper roles as recommended in a knowledge-sharing
community.
H2: Stars of the social network are the most successful students.
H3: FITCS supports knowledge-sharing.
We used data from FITCS with regard to communication on first year courses in the 2007/8
academic year, and modeled it into three networks (one network and two sub-networks):
N: Overall first year communication.
N1: Communication on Introduction to Programming - fall semester course, sub-network of N.
N2: Communication on Programming 1 - spring semester course, sub-network of N.
First year students were the target population for this research because, even if they were
experienced in communication on various forums, they had never used it for knowledge-sharing
purposes. The other important reason for choosing freshmen was the high dropout rate. We
wanted to identify any possible reasons for this in order to reduce it.
The research was carried out at FIT, Mostar, after the end of the 2007/8 academic year, when
293 freshmen were enrolled. We collected data from FITCS for the period from 1.10.2007 until
30.9.2008, and modeled them into three networks (N, N1 and N2). Network N had 273 vertices,
N1 143, and N2 99.
We accessed data from the FITCS database and exported it into an MS Excel spreadsheet, added
the necessary node identification, exported data into Matlab, and created adjacency matrices.
One of the biggest problems in knowledge-sharing at FITCS is the existence of so-called
spammers (spammers are actors with a very high vertex-degree who tend to write posts irrelevant
to the topic). Results of a previous study (Bijedic and Burak, 2006) show they are among the most
interactive actors. Additionally, the authors have found that other students do not support such
behaviour, and that spammers often cause other actors to give up communication on a topic. Both
students and educators expressed their disapproval by petitioning administrators to ban
spammers.
Another problem we recognized is that not all of the educators actively participated in
knowledge-sharing on the FITCS (results of internal quality assurance surveys). On the other hand,
students relied very much on communication with the educators, especially for help in the
interpretation of the syllabus, problem solving, and homework. Again, the results of internal
quality assurance surveys indicated that students wanted confirmation that their work was well
done. The authors found that such behaviour was appropriate for freshmen, especially in terms of
their average age.
In order to formulate a general picture, we investigated overall first year communication, and
then focused on the particular course networks, N1 and N2. These courses were chosen because
students generally found them difficult, and often teamed up in order to complete them. The
choice of one fall and one summer semester course was to enable us to follow the development of
the social network.

51
The purpose of the research was to analyze the state of the art of the FITCS as a virtual learning
community, in order to make recommendations for more adequate use. In particular, the goal was
to determine proper roles for teaching and support staff.
2. METHOD
In order to visualize the patterns of interaction among the participants, we mapped such
interactions into a two-dimensional matrix. A matrix of a network of size n is a square matrix (n x
n) whose elements represent ties (links) among actors in the network.
The network is modeled as a graph with nodes representing actors and edges representing
relationships (ties/links) among them, based on a relational data model. The relational dimension
between nodes A and B is recorded as 1 in the cells (A, B) and (B, A) if there exists an edge
between them; and as 0 if there is no such edge. On the other hand, if the relation is directional,
an arc (flow) from source A to link B and vice versa is recorded as 1 in cell (A, B), and a 0 in cell (B,
A). The existence of an edge or an arc is also referred to as adjacency. Adjacency is the graph
theoretic expression of the fact that two nodes are directly related, tied, or connected with one
another (Robinson and Foulds, 1980). Formally, adjacency is defined as:
Let N
n
n j
i Î
, denote actors i and j in a set of N actors. Let ij
a denote the existence of a
relation (arc) from actor i to actor j. Actors i and j are adjacent if there exists either of the two arcs,
ij
a or ji
a . Given a graph ( )
V
E
G ,
= , its adjacency matrix ( )
G
A is defined by ( ) ( )
ij
a
G
A = ,
where 1
=
ij
a if either ij
a or ji
a are non-zero, and 0 otherwise.
2.1. Mathematical model of networks
We modeled communication with a graph (undirected), as follows: nodes of the graph are
active FITCS actors (students, educators and the administrator). There exists an edge between two
nodes if the two actors wrote posts on the same topic. The graph is undirected because we
assumed that all who participated in a topic had read all the previous posts, and consequently had
decided to start communication. That is, a post is not only an answer to the previous post, but an
answer to a part of the previous communication. The edges are not weighted because it was
more important to find out if the actors had communicated and with how many others, then to
quantify communication between two actors (Bijedic and Burak, 2006).
2.2. Social network analysis
Social network researchers often measure network activity for a node by using the concept of
degrees—the number of direct connections a node has.
The degree of centrality in social network theory is the most intuitive network
conceptualization of centrality, and it has a simple theoretical relationship with accuracy. The
centrality of an individual is simply the number of people that person is directly tied to. A node
with a high degree of centrality suggests a high proportion of connectivity with other nodes in the
network.
To determine whether a community had formed out of the interaction, we determined group
density. Density is a measure of how connected individuals are to others in a group, and the idea is
that a higher degree of connection is a positive indicator of a community. A group’s density is
“…the ratio of the actual number of connections observed, to the total potential number of
possible connections" (Johnson, Palinkas & Boster, 2003). It is calculated by using the following

52
formula: Density = 2a/N (N-1), where "a" is the number of observed interactions between
participants, and "N" is the total number of participants.
In social network graphs applied to virtual learning community settings, a tie or relation
between two actors has both strength and content. The content might include information,
advice, or friendship, shared interest or membership, and typically some level of trust. The level of
trust in a tie is crucial in a virtual learning community. Two aspects of social networks affect trust.
One is “relational”— having to do with the particular history of that tie, which produces
conceptions of what each actor owes to the other. The other is “structural”: some network
structures make it easier than others for people to form trusting relationships and avoid
malfeasance. For example, a dense network with many connections makes information on the
good and bad aspects of one’s reputation spread more easily.
We performed social network analysis in Ucinet 6 for Windows (version 6.204). The parameters
we calculated and interpreted can be divided into parameters of the network and those describing
a particular node. To gain insight into the overall communication of the analyzed networks, we
determined network components, and calculated the average distance, density, the clustering
coefficient, cohesion, the average closeness and the average betweenness. For a description of
communication between the actors under consideration we determined egocentric networks, and
calculated degree, closeness, and betweenness (Scott, 2007). As additional information, we
calculated distributions of degree frequencies, and determined the position of actors in focus with
regard to these measures.
3. RESULTS AND DISCUSSION
In this section, we present the results for the three analyzed networks, N1, N2 and N3. Results
are grouped into network characteristics and characteristics of nodes representing particular
actors. Apart from network characteristics, in order to check hypotheses, we present results for
nodes representing educators and selected successful students. The results for other students are
not presented in the tables due to the lack of space, but we have commented on them also where
necessary. We have not presented results for any of the identified spammers, for they are not
relevant in terms of knowledge-sharing.
Table 1 shows the results necessary for understanding the overall level of communication in the
analyzed networks. The two components recognized in all three networks, are the consequences
of the existence of isolated vertices (four in network N, one in network N1 and four in network
N2).
With regard to the number of enrolled freshmen in the 2007/8 academic year, we can say that
the majority engaged in communication using the FITCS. Bearing in mind that not all actors in N
were freshmen, we estimate that 85-90% freshmen actually communicated via the FITCS.
Therefore, we can say that students are interested in such communication, and consequently, that
the FITCS has a role to play inknowledge-sharing.
Results presented in Table 1 show low density. In SNA, density represents the general level of
linkage among the points in a graph, and can take values between 0 and 1. From Table 1 we see
that the actual values vary from 0.17 to 0.25, and therefore we can conclude that none of the
observed networks shows a high level of linkage among the actors. The observed small average
distance, varying from 1.80 to 1.93, indicates the small-world property of the networks. In
addition, quite high cohesion, varying from 0.53 to 0.63, supports the previous assumption of the
small-world property of all three networks. Such characteristics are typical of human
communication. Therefore, we can conclude that FITCS is a typical online society.
There are four isolated vertices in N (1.4 %), one in N1 (0.7 %), and four in N2 (4 %). Even if
these percentages are small, the existence of isolated actors raises a question: “Why has no one
answered their posts?” Possible answers are that the questions already existed, or were not

53
interesting to other actors. In all, we can conclude that there was no deliberate isolation, and that
the communication was adequate.
Additionally, the small average distances also indicate that the networks may have a scale-free
property. In order to investigate this, we analyzed degrees of nodes, and came up with results in
favour UK SPELLING of our assumption. Therefore, we expected that there exist hubs in all the
networks. As expected from our previous research, the identified hubs were some of the most
successful students, some of the educators, the administrator and some of the spammers. Bearing
in mind the vulnerability of networks to the exclusion of hubs, we can say that the FITCS is safe as
long as educators communicate.
The observed betweenness, varying from 7.83 in N to 56.23 in N1, shows differences in the
analyzed networks. We can conclude that there were no gatekeepers in network N, and that
network can be regarded as being safe for communication, no matter which actor is excluded. On
the other hand, the observed betweenness in networks for specific subjects, N1 and N2, was quite
high. Therefore, we could conclude that those networks were in danger of falling apart if the
gatekeepers had decided not to communicate any more. On the other hand, high closeness,
varying from 52.66 to 56.78, speaks in favour of the safety of all the networks, in terms of
vulnerability to the removal of the "busiest" actors.
To summarize, we are sure that we have a typical social network, with its actors exchanging
plenty of information. Since one of the authors is an educator involved in the courses under
consideration, we can also say that the majority of actors shared information and knowledge that
was closely related to the curriculum. It follows that there are strong indications that knowledge-
sharing exists. In order to be sure, we have to analyze the characteristics of the most interesting
nodes (actors).
3.1. Social network roles
The centrality measures presented in Table 2 can be used for the analysis of the roles of the
actors in networks N, N1 and N2.
In order to analyse the positions of actors, we constructed egocentric networks for six
educators and several students. It turns out that educators 1, 2, 5 and 6 are stars in their
egocentric networks (network presenting overall communication of the particular actor), while
educators 3 and 4 have the roles of moderators. All three students shown in Table 2 were stars in
their egocentric networks.
The distribution of degree frequencies in networks N, N1 and N2 was far from normal, and
could be classified as scale-free, even if the power, p, is quite low (N1: p=-1,84, R2
=0,79; N1: p=-
1,31, R2
=0,79; N2: p=-1,99, R2
=0,87).
The FITCS administrator is the ultimate hub and star in network N, not a hub in network N1 but
still with above average communication, and absent from N2. Such a role is probably the
consequence of personality, but is inappropriate and unnecessary in a knowledge-sharing
community. The only explanation for such a role could be compensation, for some of the first year
educators did not take part in the FITCS at all (Willging, 2005).
As for educators 1 and 2, they shared responsibilities for the courses analysed through N1 and
N2. From Table 2 it is obvious that educator 1 is the star of N1, while educator 2 has a marginal
role, that educator 2 is the star of N2, while educator 1 has marginal role. The role of educator 1 in
N1 can be justified to some extent, for N1 is describing communication on a first semester course,
and freshmen had no previous experience with online knowledge-sharing. The roles of these
educators are identical to their in-class roles, so we can conclude they were both very enthusiastic.
Of the four remaining educators, another two are stars in their egocentric networks, and two have
roles of moderators.

54
Therefore, we can neither accept nor reject H1 for all of the educators regarded as a group. All
we can conclude is that it seems that the role in communication depends very much on the
personality of the actor. We accept H1 for three educators (if we accept that the role of educator
1 in network N1 is close to proper), but we still have to reject H1 for the administrator and for the
remaining three educators. Nevertheless, these very enthusiastic educators strongly support the
assumption of knowledge-sharing at the FITCS, in spite of the population of freshmen that are not
mature enough to prefer knowledge-sharing to other means of socialization.
Table 1: Parameters of social networks
N N1 N2
Average distance 1.90 1.80 1.93
Components 2 2 2
Density 0.18 0.25 0.17
Clustering coefficient 0.23 0.80 0.78
Cohesion 0.58 0.61 0.53
Closeness 54.37 56.78 52.66
Betweenness 7.83 56.23 43.94
Naturally, most of the stars of the networks are students. Some of them are more and some less
successful. A typical example is Student 3 from Table 2, who was successful in the course
corresponding to N1, and less successful in the other (corresponding to N2). This indicates that
students communicate if they are preparing for the exam, and tend to be less communicative if
their knowledge is less. The other two students (1 and 2) from Table 2 were among the most
successful students in all the courses, and their roles as stars in N, N1, N2, and the educators’
egocentric networks, are evidence of the importance extrovert personalities. The roles of all three
students strongly support the assumption of knowledge-sharing. We also want to emphasize that
there were very successful students who were not so extrovert, and are not such obvious stars,
but their communication was above 75% of the overall communication.
Table 2: Centrality measures in networks N, N1 and N2. In this table * indicates maximal value, while #
indicates central position in the network, which is equal to the role of a star in the social network.
Degree Betweenness Closeness Degree Betweenness Closeness
Network N Network N1
Administrator # 113* 295.35* 59.44* Administrator 39 61.28 36.22
Educators Educator 1 # 119* 881.17 * 46.40*
Educator 1 # 58 0 42.61 Educator 2 13 0 33.89
Educator 2 # 65 0.07 48.67 Student 1 # 90 281.22 42.39
Educator 3 40 16.44 53.20 Student 2 # 100 287.17 43.69
Educator 4 # 57 276.63 54.97 Student 3 # 103 520.08 44.10
Educator 5 17 11.06 48.50 Network N2
Educator 6 # 64 0 49.55 Administrator - - -
Students Educator 1 28 151.85 17.59
Student 1 # 72 24.77 57.44 Educator 2 # 73* 819.86* 19.18*
Student 2 # 98 108.54 60.62 Student 1 10 0 16.93
Student 3 48 38.52 53.94 Student 2 # 52 221.74 18.39
Student 3 16 0 17.10

55
The above discussion indicates that we should accept H2. In the end, we think that there is enough
evidence to support the assumption of knowledge-sharing in N1 and N2. The fact that actors have
similar roles in N, leads to the conclusion that knowledge-sharing existed in N. Therefore, we can
accept H3 for all three networks.
4. CONCLUSIONS and RECOMMENDATIONS
The results of this research are not in favour of H1 (that educators have proper roles, as is usual
in a knowledge-sharing community). Nevertheless, this online community is specific for it consists
of mostly young freshmen. Therefore, the research results for smaller groups of Masters students
may not be applicable to the community we analyzed. On the other hand, hypothesis H2 (Stars of
the social network are the most successful students) is confirmed, since some of the most
successful students, with extrovert personalities, were the stars of the three analyzed networks.
Finally, we can say that H3 (the FITCS supports knowledge-sharing) is also confirmed. The fact
that 85-90% of the students had accepted this means of communication, and that there existed
very small numbers of isolated vertices, together with the properties of the three analyzed
networks and the determined roles of the identified actors, all support the statement that the
FITCS is a knowledge-sharing community.
Recommendations based on this research are as follows: those educators should be authorized
to control content and ban spammers, that educators should be motivated to support online
knowledge-sharing, that educators should be educated with regard to their proper role in such a
community, and that successful students with introverted personalities should be motivated to be
more active in knowledge-sharing.
REFERENCES
Bijedic, N. Gimbert, J.. Miret, J.M. Valls, M. (2007); Elements of Discrete Mathematical Structures
for Computer Science; Univerzitetska knjiga; Mostar.
Bijedic, N.and Burak, S. (2006). Cluster Analysis of a Scale-Free Network", TMT 2006 Conference
Proceedings.
John Scott (2007). Social Network Analysis, A Handbook (Second edition). Sage Publications:
London.
Johnson, J. C., Palinkas, L.A. & Boster, J. S. (2003). Informal Social Roles and the Evolution and
Stability of Social Networks. Prestige, Centrality, and Learning: A Social Network Analysis of
an Online Class. 2003 International Sunbelt Social Network Conference Proceedings.
Koschmann, T., Kelson, A. C., Feltovich, P. J., & Barrows, H. S. (1996). Computer-Supported
Problem-Based Learning: A principled approach to the use of computers in collaborative
learning. In T. Koschmann (Ed.), CSCL: Theory and practice of an emerging paradigm (pp. 83-
124). Mahwah, NJ, 1996
Lee, M. and Lee, Y. (2004). How to Measure Team Communication in Web-based Collaborative
Learning: The Application of SNA (Social Network Analysis). In G. Richards (Ed.), Proceedings
of World Conference on E-Learning in Corporate, Government, Healthcare, and Higher
Education 2004 (pp. 361-366). Chesapeake. VA: AACE
Lorrain, F., & White, H.C. (1971). Structural equivalence of individuals in social networks. Journal of
Mathematical Sociology, 1, pp.49-80.
Robinson, D. F. and Foulds, L.R. (1980) Digraphs: Theory and Techniques. New York: Gordon and
Breach.

56
Russo, T. C and Koesten, J. (2005). Using Social Network Analysis to Understand Sense of
Community in an Online Learning Environment. Communication Education, 54(3), pp.254-
261 July 2005.
Shetty, J., and Adibi, J. (2004). The Enron email dataset database schema and brief statistical
report. Technical Report, Information Sciences Institute, USC.
Suthers, D. (1999). "Representational support for collaborative inquiry.," in Proceedings of the
32nd Hawaii International Conference on the System Sciences. Los Alamitos, CA: IEEE
Computer Society Press, 1999.
Wasserman, S., Faust, K. (1994). Social Network Analysis: Methods and Applications. Cambridge
Cambridge University Press.
Wenger, E. (1998). Communities of practice. Learning meaning and identity. Cambridge:
Cambridge University Press.
Willging, P. A., (2005) Using Social Network Analysis Techniques to Examine Online Interactions,
US-China Education Review, 2(9), pp.46-56
Wolfson, L. & Willinsky, J. (1998). What service learning can learn from situated learning. Michigan
Journal of Community Service Learning, 5, pp.22-31.
Zhu, E. (2006). Interaction and cognitive engagement: An analysis of four asynchronous online
discussions. Instructional Science, 34(6) pp.451-480.
Turkish Abstract
Öğrenme topluluğu'nun bir sosyal ağ olarak analizi
Öz: Online bilgi paylaşım toplulukları genellikle küçüktürler. Bu çalışmada, büyük online
toplulukları ın bilgi paylaşı ı için uygun olup olmadıkları sunulmaktadır. Ayrıca çalışmada sosyal
paylaşım ağı tekniklerini belirlenerek etkileşim ve sanal ortamda öğrencilerin pilgi paylaşı ı analiz
edilmiştir. Analiz için ağ özellikleri ve katılımcıların rolleri belirlenmiştir. Araştırma 2007/8
akademik yılı sonunda Bosna Hersek Mostar Bilgi Teknolojileri fakültesinde 293 birinci sı ıf
öğrencileriyle birlikte yürütülmüştür. Veriler ekim 2007-Eylül 2008 tarihleri arasında Bilgi
Teknolojileri Fakültesi Topluluk serverinden toplanarak 3 ağ içerisinde modelleştirilmiştirler (N, N1,
N2). N ağı (ayrıntılı iletişim) 273, N1 (Güz döneminde Programlama dersi) 143, ve N2 (Yaz
okulunda programlama dersi) 99. Kayıtlı birinci sı ıf öğrencilerinin sayısı ı göz önüne alındığı
zaman 85-90% ‘ı ın Bilgi Teknolojileri Fakültesi Topluluk serveri aracılığıyla iletişim kuracakları
tahmin edilmiştir. Çalışma sonucunda elde edilen bulgular bazı eğitimcilerin online bilgi paylaşı ı
topluluklarında uygun rollerinin olmadığı ortaya çıkmıştır, ancak bunun sebebinin katılımcıların
birinci sı ıf olmasından kaynaklanabildiği düşünülmektedir. Diğer taraftansa, bazı başarılı
öğrencilerin dışa dönük kişiliklerinin, analiz edilen 3 ağ içerisinde göze çarptıkları görülmektedir. Bu
nedenle, sistemin analiz edilen bölümünün bir bilgi paylaşım topluluğu olduğu sonucuna
varabiliriz. Çalışmadan elde edilen sonuçlara bağlı olarak eğitimcilerin online bilgi paylaşı ı
sağlamaları için motive edilebileceği, eğitimcilerin toplulukta uygun roller için eğitilebileceği ve
başarılı öğrencileri motive ederken içe dönük kişiliklere sahip öğrencilerin aktif rol almaları
sağlanabileceği, önerileri getirilmiştir.
Anahtar kelimeler: Online topluluk; sosyal ağ Analizi; bilgi paylaşı ı

Analysis Of A Learning Community As A Social Network

Recommended

Recommended

More Related Content

Similar to Analysis Of A Learning Community As A Social Network

Similar to Analysis Of A Learning Community As A Social Network (20)

More from Maria Perkins

More from Maria Perkins (20)

Recently uploaded

Recently uploaded (20)

Analysis Of A Learning Community As A Social Network