You are on page 1of 10

Knowledge Sharing over Social Networking Systems:

Architecture, Usage Patterns and Their Application

Tanguy Coenen, Dirk Kenis, Céline Van Damme, and Eiblin Matthys

Vakgroep MOSI, Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussels, Belgium


{tanguy.coenen,dirk.kenis,
celine.van.damme, eiblin.matthys}@vub.ac.be

Abstract. The recent success of social networking sites like mySpace, Friend-
ster, Orkut, LinkedIn, Ecademy and openBC indicates that extending one's so-
cial network through a virtual medium is a popular proposition. The social
networking paradigm can be integrated with the knowledge management field,
in which sharing knowledge with others is a central issue. This paper investi-
gates how the social networking concept can be applied to support knowledge
sharing between people. Together with common features in the architecture of
social networking systems, a number of platform independent usage patterns are
discussed that can support knowledge sharing between members. Finally, we
present the open source KnoSoS system, which integrates the discussed archi-
tecture and usage patterns.

Keywords: social networking, knowledge management.

1 Social Networking Systems


New computer mediated technologies, like instant messaging, voice-over-ip and video
telephony have significantly lowered the cost of communication and, when compared
to email, have made computer-mediated communication much richer. It is now possi-
ble to talk hours at a stretch to someone at the other side of the world at no cost be-
yond the basic Internet connection fee. Yet these technologies still do not provide all
the cues that are available in a face-to-face conversation. When meeting a person over
the Internet and interacting with him, it is difficult to find out if you are really inter-
acting with who the person claims to be. Furthermore, finding people on the Internet
with whom you are likely to have interesting and useful conversations is not some-
thing that is likely to occur by chance. To make sure that people can represent their
identity and to allow people with mutual or compatible interests to find each other,
social networking systems have been created. The first social networking systems
were dating systems, to which people turned to find a partner. Yet dating systems are
not the subject of this paper, as they offer little to support knowledge sharing.
Through systems like Orkut, MySpace, LinkedIn, O’Reilly Connection, Ecademy
and OpenBC, the social networking paradigm is spreading quickly on the Internet as a
way for people to develop an online social life. Spearheading this change are Ameri-
can social networking systems like Friendster, Orkut and MySpace. In May 2006, the
latter had 76.524.752 members, which is roughly equivalent to the population of

R. Meersman, Z. Tari, P. Herrero et al. (Eds.): OTM Workshops 2006, LNCS 4277, pp. 189 – 198, 2006.
© Springer-Verlag Berlin Heidelberg 2006
190 T. Coenen et al.

countries like France or Germany. On 19/06/2005, mySpace was sold to Rupert Mur-
doch’s News Corp for 580 million $, indicating that social networking systems are
very actual material and that, if they are not yet the business of today, they are proba-
bly the business of the future.
Considering the success of systems like MySpace, it is obvious that young people
are currently spending a lot of their time online, using computer mediated communi-
cation and social networking systems to express their personality and meet with
friends. It is likely that, as these people grow older and enter the work force, they will
be accustomed to these technologies and will expect to be able to use them in their
work environment.
By presenting the common architecture of social networking systems and a number
of general usage patterns, this paper indicates how social networking systems can
support knowledge sharing. This is relevant both within large organisations, between
organisations or between individuals without organisational affiliation. Yet before in-
dicating how social networking systems can benefit knowledge sharing, it is neces-
sary to present an overview of some issues related to the sharing of knowledge.

2 Knowledge Sharing
One of the central issues in the knowledge management field has always been the
sharing of knowledge [12]. Knowledge sharing can occur in what we call the passive
and the interactive mode. In the passive mode, the source, who owns the knowledge,
externalises his knowledge and stores it as information. The receiver, who wishes to
use the knowledge, assimilates the knowledge but has no way of formulating feed-
back to the source. Unlike what is the case for passive knowledge sharing, interactive
knowledge sharing involves a possibility for the receiver to provide the source with
feedback. From a constructivist perspective, individuals are seen as possessing their
own unique understanding of the world. As a consequence, communication is by
definition complicated as it confronts different mental models. [7]. The possibility to
produce feedback can thus be essential in situations where the receiver does not un-
derstand the information, provided by the source. The source can then re-formulate
his knowledge in a way that is more suited to the needs of the receiver1
Passive knowledge sharing has the great benefit of being highly reusable. The
source externalizes his knowledge once and the resulting information can be reused
many times by different receivers. Yet the knowledge, which is made available
through passive knowledge sharing, can go quickly out of date and there is a motiva-
tional problem. Indeed, people find it hard to contribute their knowledge to a vague
audience if they do not have a clear view on the "return on investment" which they
will obtain from sharing their knowledge [11].
Therefore, knowledge sharing must not only focus on passive knowledge sharing,
but should also support interactive knowledge sharing. Both modes are useful and
should therefore be present in and between organisations or between independent

1
Also, complex knowledge, which is highly intertwined with other knowledge components, re-
quires an interactive mode of knowledge sharing, as the receiver may need to obtain some
context knowledge in order to correctly understand the knowledge of the source.
Knowledge Sharing over Social Networking Systems 191

individuals. Support for the passive mode has been around for over a decade in the
form of information management approaches, focussing on storing information in da-
tabases. We argue that a focus on the interactive mode of knowledge sharing is neces-
sary and that applying concepts of social networking systems in addition to new rich
computer-mediated communication technologies can support this. In the next section,
the common architectural features of social networking systems are presented.

3 Common Features of Social Networking Systems

A study of existing social networking systems (mySpace, Friendster, Orkut, LinkedIn,


Ecademy and openBC) has revealed a common architecture, depicted in figure 1
[2][3]. In the individual space, the user is allowed to create a personal profile, con-
taining structured and unstructured information. The structured information part of the
profile contains information on different facets of the individual's personality. These
facets vary between systems. Indeed, where some systems concentrate more on creat-
ing friendship relationships (e.g. mySpace, Friendster and Orkut), others are focused
on creating business relationships (e.g. LinkedIn, Ecademy and openBC). Information
on for example one's musical taste would be interesting to have in a friendship-
oriented system, but less interesting in a business-oriented system. In practice, each
social networking system contains different fields in which structured information can
be entered. The unstructured fields allow people to create a free and rich representa-
tion of their own identity. Wysiwyg editors are provided to create webpages contain-
ing text, images, movies and sound clips. In addition, some systems allow people to
create blog entries, which further enrich a person's profile.
In the dyadic2 space, users create contacts with whom communication can be un-
dertaken over the internal messaging system. This internal messaging system is very
similar to an email system, but has the advantage of being able to shield a user's ex-
ternal email address. The message, which is sent over the internal messaging system,
is forwarded to the user's external email account, without the need for the sender of
the contact to know the email address of the receiver. This is a necessary measure to
prevent social networking systems from becoming vehicles for email spamming.
Another element of the dyadic space is the possibility for both members of a dyad
to create feedback on the other member of the dyad. In different analysed social
networking systems, this can be done by rating certain characteristics of the other
member of the dyad, or by writing testimonials on this person [2]. We call such sig-
nals, "identity feedback" and have found it to be important in the creation of a sense
of trust in social networking systems. Indeed, such systems create many relationships
that are purely virtual in nature, which results in a need for the members of the system
to evaluate the genuineness of the system’s users.
Finally, the group space contains tools, which allow knowledge sharing between
multiple people. This space constitutes the overlap between the areas of knowledge
management, social networking systems and community informatics. In most sys-
tems, the tools, which are available in the group space, are limited to a forum on

2
A dyad is a relationship which has been acknowledged by both members of a relationship.
192 T. Coenen et al.

Fig. 1. General structure of social networking systems

which members can post and read messages. There is still much room for improve-
ment in the group space, as it exists in current social networking systems, in order
to better support knowledge sharing between multiple people.

4 Pattern Languages
Patterns describe common ways of solving problems. A pattern language is an or-
dered set of patterns that are tied together in a logical way. Problems arise in all do-
mains of human activity, making pattern languages a useful way to structure problems
and solutions in different disciplines. Whereas the term was first introduced in archi-
tecture [1], pattern languages have been applied to many other fields, like education
and object-oriented programming. In this paper, we do not claim to develop a pattern
language, but present a number of loose patterns, which are the result of our analysis,
development and usage of social networking systems for knowledge sharing.
Each pattern has a name, a context, a system of forces and a solution. The context
describes the conditions that must be taken into account to fully understand the prob-
lem and its solution. The system of forces describes the problem at hand and the solu-
tion describes how the problem is solved. An important point concerning patterns is
that they are meant to be generic, applying to many different instances of the same
problem. Translated to social networking systems for knowledge sharing, this means
that we aim to formulate patterns that are applicable to social networking system in
general, and should therefore be taken into account when designing such systems. The
added value of a pattern does however not only lie in the solving of a problem. In-
deed, many others, like [13], see messy situations, which are the characteristic of
human activity, as also problematic in their definition of the problem, besides the so-
lution. In other words, identifying the right problem is a challenge in itself when deal-
ing with human activity.
The patterns which are presented next are the result of our experience with social
networking systems (mySpace, Friendster, Orkut, LinkedIn, Ecademy and openBC)
Knowledge Sharing over Social Networking Systems 193

and our insight in the issue of knowledge sharing. Other patterns have been developed
or are in development (e.g. identity representation, rating of content, project manage-
ment), but cannot be discussed here due to a lack of space.

4.1 Pattern 1: Creating Group Boundaries

Context3
Previous research [2][3] has established that there is a statitistically significant differ-
ence in the amount of exchanged messages between open and closed groups. The dif-
ference lies in the boundaries that exist around these groups. In an open group, every-
one can participate in the communications that are a part of the group's knowledge
sharing activity. In a closed group, one needs to be a member4 to participate in the
group's knowledge sharing activity. In the 1196 open groups of the Ecademy social
networking system, 3 times less messages where exchanged on average, compared to
the 304 closed groups in the system.
This is in line with [9] who proposes that knowledge sharing will be more common
in groups with a stable membership, due to a lower expectation of free-rider behav-
iour by other members of the group. Another point, which can support this higher
knowledge sharing activity in closed groups, is the argument in [2] that the value of
knowledge lies in the scarcity of the capacity to act that it produces. If one gives away
one's knowledge to the whole world, as is the case in open groups, the scarcity of this
capacity to act can seriously decrease, as others can readily assimilate the knowledge.
Yet contributing knowledge to a group of people who share a certain interest can pro-
duce reciprocation benefits, originating from generalized social exchange between the
members of the group [9]. If this group is closed, the contributor will feel he may re-
ceive these benefits, while only sharing his knowledge with a small subset of human-
ity. Thus, the benefits in the case of knowledge sharing in a closed group are more
likely to outweigh the costs, due to the targeted nature of the knowledge-sharing act.
Still, not all knowledge is essential to the economic position of the holder of the
knowledge. Thus, some knowledge can be easily donated without harming the posi-
tion of the source. If this is the case, it can be in the interest of all the participants of
the system and not only the members of a certain group to be able to access the posted
content. In this case, posting to an open group seems opportune.
In sum, both open and closed groups are necessary in social networking systems
for knowledge sharing, making the creation of boundaries a central pattern in their
functioning. The context, described in this paragraph, allows us to formulate 2 pat-
terns: creating groups with or without boundaries and creating content in groups.
These patterns are discussed next.
Problem
How to create groups with and without boundaries?
Solution
Every user should be able to set up a group if he wishes to do so. At the time of crea-
tion, the person who takes the initiative for the creation of the group decides if the
group should have an open or closed membership. In a closed group, mechanisms

3
The context, described in this section, is relevant to the first two patterns which are presented.
4
Membership is often obtained by requesting it explicitly to one or more administrators.
194 T. Coenen et al.

should be provided which allow outsiders to apply for group membership. One or
more group administrators should then be able to screen the application and grant or
deny access to the group.

4.2 Pattern 2: Tracking Content

Context
A social networking system, like the Internet in its totality, is an environment in
which content is produced at a very high rate. The lowering of the content production
threshold is what probably best defines the web2.0 paradigm. Indeed, systems that al-
low users to easily create and publish text, movies; audio clips etc are producing an
explosion of content. This has accelerated the rate at which content is produced, but
has also lead to an increased heterogeneity in content. This combination of high pro-
duction rate and content heterogeneity has resulted in a situation in which the structur-
ing of content by means of hierarchies or ontologies has become unfeasible.
Problem
How to keep track of heterogeneous content, which is produce at a high pace?
Solution
Allow users to add freely chosen tags to describe the content in the system in order to
retrieve it later. In doing so, they create metadata, expressed in their own terms. A
consolidation of all these tags constitutes a bottom-up taxonomy, created by the
members of the social networking system. This taxonomy is a reflection of the vo-
cabulary, prevalent in the minds of the users. The absence of a controlled vocabulary
and the immediate return on investment of retrieving tagged content by using one’s
own words, is a real incentive for tagging content and tagging it a conscientious way.

4.3 Pattern 3: Grasping Perspectives

Context
During a knowledge sharing process, it is important to be able to grasp the perspective
of an individual or a group of individuals [5]. By providing users with a mechanism
for creating boundary objects, it is possible to create a perspective that can then be
used by others to quickly find out what the cognitive schemes of other people look
like. Indeed, a boundary object by definition is an object that can be shared across
perspectives [4]. This can facilitate knowledge sharing, as, according to social science
perspectives like constructivism [7], different people hold different views of what
constitutes reality. Therefore, visualising the perspectives of people through boundary
objects can increase the insight that different parties of a communication have in each
other's perspectives. This allows them to conduct a communication that is more tar-
geted to the perspective of the other participants in the communication.
Problem
How to create a visual representation of the cognitive perspective of a person or of a
group of persons?
Solution
Use the tags which where attributed as a result of the previous pattern (content track-
ing) to create a cognitive network view which represents the way different tags are
Knowledge Sharing over Social Networking Systems 195

related. This can be done, based on the principle of co-occurrence [8], by which two
tags that reference a same resource, accessible through a URL, are linked by a rela-
tionship of strength 1. The strength of the relationship between the tags is incre-
mented by 1 each time the 2 tags are used together to reference a concept. Thus, if
knowledgeSharing and socialNetworking have been used together 20 times together
to reference a resource, their association’s strengths will be 20. This approach can be
used to create visualisations of independent perspectives, but can also be used to
merge individual perspectives into a group perspective.

5 Knosos

In this section, we present the open-source KnoSoS system and the way in which it im-
plements the architecture and usage patterns which have been developed earlier. As the
KnoSoS system is open-source and freely available, it is possible to use the system for
knowledge sharing and to quickly develop and test new additions to the social network-
ing paradigm. Hence it can be deployed in and customized to different environments,
like multinationals, governmental or non-governmental organisations. Furthermore,
KnoSoS implements the architecture and all the patterns which have been discussed.,
which is not the case for all social networking systems (cf. in tables 1 and 2).

Table 1. Support of the 3-layer architecture by different social networking systems

System Individual space Dyadic space Group space


OpenBC X X X
Ecademy X X X
Orkut X X X
mySpace X X X
O’Reilly Connection X X -
linkedIn X X -
KnoSoS X X X

Table 2. Support of the 3 patterns by different social networking systems

System Pattern 1 Pattern 2 Pattern 3


OpenBC X - -
Ecademy X - -
Orkut X - -
mySpace X - -
O’Reilly Connection - - -
linkedIn - - -
KnoSoS X X X
196 T. Coenen et al.

Table 1 shows that most social networking systems support the 3-layered architec-
ture. In table 2, however, it is clear that of the proposed patterns, only pattern 1 seems
to be supported by most social networking systems. Patterns 2 and 3 are only sup-
ported by KnoSoS. Yet these patterns are important to the support of knowledge shar-
ing. Therefore, KnoSoS is probably more suited for the support of knowledge sharing
than are other existing systems.
Subsequently, the implementation of the 3 usage patterns in the KnoSoS system is
addressed briefly. For a detailed description of KnoSoS and how this system imple-
ments the 3-layer architecture as well as present and other patterns, the reader can
visit and test the system5.

5.1 Pattern 1: Creating Group Boundaries

KnoSoS, like a number of other social networking systems, allows the administrator
of the group to make it open or closed. If a user wants to access an open group, this
can be done without extra effort. To become member of a closed group, however, the
user needs the approval of the administrator.

5.2 Pattern 2: Tracking Content

The tracking of content is done by allowing users to add tags to the different types of
content which are available in KnoSoS. The basic content types are blogs, forum
posts and books. The latter contain collaborative writing features. Additional custom
content types can be added by the administrator of the system. This is useful for creat-
ing repositories of hyperlinks, papers, idea proposals, etc… Each of these content
types can be tagged by the user who visits the specific content. In addition to allowing
the user to keep track of the content in the system, the tagging functionality provides
metadata which will allow subsequent development of more advanced features, like
the matching of content to users, the matching of users to each other and the devel-
opment of bottom-up ontologies. Whereas these topics require further research, one
use of the metadata created by users has already been implemented in the form of
boundary objects. This is discussed subsequently.

5.3 Pattern 3: Grasping Perspective

The tags, produced through the application of pattern 2, are used to create a map of
the perspective of a user. This is done by means of a Java applet which we developed,
called TagViz6. The relationships between the tags are inferred based on the co-
occurrence method, described in section 4.3.
At the time of writing, only the perspectives of single users could be visualised, but
efforts are under way to visualise the perspectives of sets of users. In this way, it will
be possible to visualise the perspective of a whole group. This visualisation of

5
The system can be tested at www.knosos.be. KnoSoS is a distribution of the open-source
Drupal content management systems to which custom modules have been added.
6
TagViz is available for testing at http://www.knosos.be/sandbox/tagviz/index.html
Knowledge Sharing over Social Networking Systems 197

Fig. 2. Visualisation of part of a personal perspective through a boundary object, constructed


from the tags in a KnoSoS account

perspectives is expected to facilitate knowledge sharing between users and the local-
isation of users with specific knowledge. Whether this is truly the case offers an inter-
esting opportunity for future research.

6 Conclusion
This paper proposes that passive knowledge sharing should be complemented with in-
teractive knowledge sharing approaches and that the social networking paradigm is
well suited to allow this. From experiences by participating in -and designing- social
networking systems, 3 patterns have been distilled to enhance social networking sys-
tems for knowledge sharing. Furthermore, it was discussed that most present social
networking systems share a common architecture based on 3-layers. .
The open-source KnoSoS system builds on this common architecture and imple-
ments the presented patterns At the moment it acts as a platform in which new
techniques can be tested. One particular improvement which will be pursued in the
198 T. Coenen et al.

immediate future is the matching of users to each other and the matching of users to
content. The tags which are produced as a consequence of using the systems will play
a major role in this. A beta version of the KnoSoS system can be tested on
www.knosos.be.

References
1. Alexander, 1977, A Pattern Language: Towns, Buildings, Construction, New York: Ox-
ford university press
2. Coenen, T, 2006, Knowledge sharing over social networking systems, forthcoming PhD
dissertation
3. Coenen, T, 2006, Structural aspects of social networking systems, Proceedings of Web-
based communities 2006, San Sebastian, Spain
4. Carlile, P R, 2002, A Pragmatic View of Knowledge and Boundaries: Boundary Objects in
New Product Development, Organisation Science, Vol 13, No 4, p 442-455
5. Boland, R J, Tenkasi, R V, 1995. Perspective making and perspective taking in communi-
ties of knowing. Organisation Science, Vol 6, p350-372.
6. Daft, R, Lengel, R, 1986, Organisational Information requirements, media richness and
structural design, Management science, Vol 32, p554-571
7. Fosnot, C.T., 1996, Constructivism: a psychological theory of learning, in Fosnot, C.T.,
1996, Constructivism – theory, perspectives and practice, , New York: Teachers Press
College
8. Heylighen, F, 2001b, Mining Associative Meanings from the Web: from word disam-
biguation to the global brain in Temmerman, R (ed.), Proceedings of Trends in Special
Language and Language Technology, Brussels: Standaard Publishers, http://pespmc1.
vub.ac.be/Papers/MiningMeaning.pdf
9. Kollock, P, 1999, The economies of online cooperation, in Smith, M A, Kollock, P (eds),
Communities in cyberspace, London: Routledge, p220-239
10. Markus, M L, 2001, Toward a Theory of Knowledge Reuse: Types of Knowledge Reuse
Situations and Factors in Reuse Success, Journal of Management Information Systems,
Vol 18, No 1, p57-93
11. Nahapiet, J, Goshal, S, 1998, Social capital, intellectual capital and the organisation ad-
vantage, Academy of Management review, No 23, p242-266
12. Nonaka, I, Takeuchi, H, 1995, The knowledge-creating company, Oxford University press,
New York
13. Vennix, J, 1996, Group model building: facilitating team learning using system dynamics,
Chichester: Wiley

You might also like