=Paper=
{{Paper
|id=None
|storemode=property
|title=Self-modeling and Self-reflection of E-learning Communities
|pdfUrl=https://ceur-ws.org/Vol-709/paper11.pdf
|volume=Vol-709
|dblpUrl=https://dblp.org/rec/conf/ectel/Petrushyna10
}}
==Self-modeling and Self-reflection of E-learning Communities==
https://ceur-ws.org/Vol-709/paper11.pdf
Self-modeling and Self-reflection of E-learning
Communities
Zinayida Petrushyna
RWTH Aachen University
Information Systems and Databases Chair
Ahornstrasse 55, 52066, Aachen, Germany
{petrushyna}@dbis.rwth-aachen.de
Abstract. Numerous e-learning communities die over the course of time
as they do not manage to adapt themselves to the changes of the com-
munity environment. We support the communities by providing possi-
ble solutions to problems the communities deal with. Firstly, we model
communities and their environments and analyse them. As a result, the
comparison of communities, environments and their evolution is possi-
ble. Afterwards, we propose to identify what solutions of communities
were successful and enable communities to survive. These solutions are
vital for communities that find themselves in similar situations. We sug-
gest to share the solutions with other communities to keep them fit. To
evaluate our approach, we propose to simulate communities and their
environment to estimate whether a proposed solution effectively helps
communities to survive.
Keywords: community of practice, community patterns, fitness of communities
1 Introduction
Communicating, sharing knowledge and achieving goals are the fundamentals of
communities [16]. Humans are social beings. Thus, participating in communities
is an unavoidable part of human life. Communities have different participants,
different topics of interest and different goals but in general the structure of
communities is similar.
Communities mature or die over the course of time. Therefore, we aim to
preserve the knowledge of communities by discovering and exploiting patterns
of community fitness so as to support them in reacting on changes inside the
community and in their environment.
2 Discovering Patterns of Community Fitness
As we require to find the most fit communities, we need to compare them with
each other. We need an approach for modeling communities so that we can
compare models of communities with each other. Firstly, a modeling approach
61
must be extendable: as soon as a disturbance, an unexpected event, appear, a
community has to adopt to the change. Furthemore the approach has to be able
to represent desired situations, i.e. community goals. Goals of communities have
potentially positive or negative impacts on a community. Most of the modeling
techniques coming from process modeling, business process reengineering and
requirements engineering focus on technical systems that support work. Actors
of technical system models support functionalities that cover users’ needs and,
thus, the users are interacting with the systems only for covering the needs.
Hovewer, no indication is done on user motivations and goals [17]. Even Learning
Process Modeling that includes a reengineering concept and considers learning
goals [12] cannot give an appropriate view on a learning community as it consider
needs that a learning module cover but not goals that a community / a leaner
has. The IMS Learning design can express learning activities that are performed
by a learner using a learning object. The activities can be shared and reused.
However, as in the case with process modeling the focus of the IMS Learning
design is on user actions but not on user goals. We find that i* from Yu covers
all requirements we defined.
As soon as an appropriate approach for community modeling is found, com-
ponents of a community as well as techniques for computing the components
should be clarified. Afterwards, a database schema which represents a commu-
nity model should be created. Using the schema, community parameters are
stored in a database. Moreover, the computation of the changes of a community
over the course of time and their storage should be solved. We are interested
in finding patterns of fitness in the stored data. The patterns are solutions for
repeatable events, where a problem doen’t mean . Thus, a community is char-
acterized by a set of components and sets of a community on different time
intervals may be different. Investigating the sets, we can discover patterns, e.g
repeatable events, and solutions, e.g. adoptation, of communities . We support
a community to survive as we compare its parameters with parameters of other
communities. As soon as we know that the community has a pattern, same that
the other communities have, we can suggest the community solutions that the
other communities used. It is upon the community if it wants or donnot wants
to take suggestions into consideration.
3 The loop of community survival
As it was mentioned in the previous section, we observe life of a community
in a changeable environment. Figure 1 depicts a circulation that normally hap-
pens if something unexpected appears, i.e. disturbances. The disturbances are
positive, negative or neutral events that are unexpected and appear inside of
a community or outside it. The Self-modeling phase stands for mining commu-
nity parameters and changes of those parameters over the course of time. The
Self-reflection phase stands for analysing parameters of communities within all
time periods when parameters where captured. Moreover, patterns of commu-
nities are discovered and stored in the phase. Last but not least, the support
62
of communities is perfomed by finding similar disturbances that other com-
munities met and by proposing solutions the other communities have applied.
In Figure 1 is depicted
that Community of Practice
(CoP) [16], Activity Theory
[4] and Actor Network Theory
[11] create a consolation for
the theoretical basis for the
Self-modeling phase. The CoP
suits better in our methodol-
ogy than learning networks as
the CoP focuses on commu-
nities that are informal and Fig. 1. The loop of self-modeling and self-reflection
not institutionalised as ex- for e-learning communities
plained by Cumming and Zee
[3]. Wenger claims that a CoP consists of three main components. These are
about interactions of community members, same context and possessing a sim-
ilar knowledge domain. As we consider communities’ actual and desired states,
we need to extract the goals communities have. A CoP stresses collaborative
work of learners while Engeström explains learning through goal-directed ac-
tivites of learners. As we are interested in both, communities and goals, we find
correlating points between both theories and use them for our investigations.
For creating a model of a community, we apply not only the learning theo-
ries but Actor Network Theory (ANT) which considers environment as a set of
agents. It states that all elements in environment as an actor. Hence, all mem-
bers of a community are actors as well as Media and Artefacts they use. We
need to create a model of a community that is extandable as environment and
the community change and we do not know what actor should be added to the
model over the course of time. According to the ANT, we can add any event or
resource as an actor if it is still not defined in the model.
4 The repository of community fitness patterns
We observe (in the Self-modeling) and support (in the Self-reflection) the life
of learning communities in changeable environment and care about survival of
communities.
According to CoP and Activity Theory, learning communities components
should be computed. Collaborations between learners can be examined using
Social Network Analysis (SNA) [2], users influence measures or measures of col-
laborations are explicified through SNA parameters. Observing the parameters,
we can suggest roles a learner plays and understand a state of a community
collaboration in general. Moreover, we can extract patterns based on the param-
eters. The example of a pattern: If 2/3 of community members organizes cliques,
isolated groups of learners, and a number of members in the cliques is no more
63
than 2/10 of all community members than the community needs to refine the
structure of its network and increase communication between the cliques. The
density will increase, clustering affect will descrease and connections between
isolated groups will be established.
Technologies of the Semantic Web [1] allow to get a clear idea about con-
cepts and contexts of knowledge that are mentioned in a community-generated
content. This information can be used as a set of community parameters that
answer for a community knowledge domain. Using an API like OpenCalais 1 , it
is possible to clarify main topics of discussions and documents in communities
and to discover emergent themes. The example of a pattern: A topic cannot be
supported by community members involved in the discussion, so that the mem-
bers are unsatisfied. Members of the community are need to be found that have
expertise in the topic.
Moreover, activities performed by learners over the course of time express
goals of learners and their communities. Goal mining is required to understand
which goals communities have and how do they reach them, e.g., how do they
deal with disturbances and what solutions do they have. To define goals, we refer
to the phases explained in [5]. During the Plan, Learn and Reflect phases a goal is
set, achieved and future goals are set. During the Self-modeling phase we collect
community parameters over the course of time when a community achieves its
goals and during the Self-reflection phase we define patterns that were applied
for the community as well as we suggest solutions of other communities if the
community hasn’t still achieve a goal.
Summarizing all, a community model includes community parameters that
are computed by techniques expalined in previous 3 paragraphs. So that dif-
ferent states of a community over the course of time are saved in a database.
The pattern dicovery should be then performed under community parameters
in the database. The challengable task will be to form patterns and combine
different community patterns in them as parameters devoted to communication
and parameters devoted to concepts are not correlating [14].
As soon as patterns are discovered, they should be evaluated on a test set of
communities. Particularly, the simultions of communities implemented as multi-
agent systems [15] can be used to define changes within communities if a partic-
ular pattern is applied.
The results that are achieved so far:
– Modeling of communities with i* and utilizing extracted parameters of the
communities in models to define the changes over the course of time and
adoptations to the changes [8, 13].
– A Community-oriented database, the Mediabase, was reengineered accord-
ing to Actor Network Theory principles [7]. The Mediabase is a community-
oriented knowledge repository. The Web 2.0 interface for the Mediabase, the
Mediabase commander 2 , allows users to add different Web media to share
within their communities. Furthermore, the resources are analyzed with the
1
http://www.opencalais.com
2
http://www.prolearn-academy.org/mediabase
64
OpenCalais API and important categories and concepts of the resources are
defined. Users can see the most popular media that is used within their com-
munities as well as the most popular tags users attach to media. Moreover,
media stored in the Mediabase can be visualized with PALADIN II which
pictures interactions between community members within media and defines
different roles of community members [9] as well as represents link networks
of blogs. On the example of the Mediabase and the Mediabase commander,
we apply ANT for the Mediabase model and we try out SNA technique and
OpenCalais API for extracting community parameters.
– SNA experiments on different Wikipedias extract roles of community mem-
bers with consideration of the culture of contributors. We conclude about
cultural differences in knowledge creation and differences between people of
the same culture but physically located at different places in the world. For
example, the amount of contributions of Turkish diaspora from Germany is
higher than Turkish diaspora living in other countries. We used SNA tech-
nique for defining roles of Wikipedia contributors and check the applicability
of those roles over different communities.
– Collaborations between teachers of European SchoolNet 3 were simulated
with the purpose to find a perfect partner so that the teachers, forming a
partnership, benefit from each other. Simulations were based on teacher pro-
files based on teacher competences. The simulations were performed with a
small amount of data and the algorithm should be found to efficiently com-
pute benefits of teachers againt other teachers. We tried simulation tech-
niques for having experience to simulate communities and their behaviour
based on pre-defined models.
5 Related work
The Multi-method approach considers different aspects of learning in CoPs [10].
Laat et al. concentrate on analysis and definition of roles within learning com-
munites, however they do not consider modeling of learning communities and
support communities in their survival. Glahn et al. support learning communi-
ties reflection by providing visualizations of learner interactions but they do not
consider if learners achieved the goals they set [6].
References
1. Tim Berners-Lee, James A. Hendler, and Ora Lassila. The Semantic Web. Scientific
American, 5 2001.
2. U. Brandes and T. Erlebach. Fundamentals. In Ulrik Brandes and Thomas Er-
lebach, editors, Network Analysis: Methodological Foundations. Springer, 2005.
3. S. Cummings and A. van Zee. Communities of practice and networks: reviewing
two perspectives on social learning. KM4D Journal, 1 (1):8–22, 2005.
4. Y. Engeström. Learning by expanding. Orienta-Konsultit Oy, Helsinki, 1987.
3
www.eun.org
65
5. Karin Fruhmann, Alexander Nussbaumer, and Dietrich Albert. A psycho-
pedagogical framework for self-regulated learning in a responsive open learning
environment. In In Proceedings of the International Conference eLearning Baltics
Science, Rostock, Germany, 1-2 July 2010 2010.
6. Christian Glahn, Marcus Specht, and Rob Koper. Perspective and contrast; design
principles for supporting self-directed and incidental learning. In K. Tochtermann
& H. Maurer (Eds.), editor, 9th International Conference on Knowledge Manage-
ment and Knowledge Technologies (I-KNOW’09) and 5th International Conference
on Semantic Systems, pages 299–308, September, 2-4, 2009, Graz, Austria, 2009.
Verlag der Technischen Universität Graz.
7. R. Klamma and Z. Petrushyna. The troll under the bridge: Data management
for huge web science mediabases. In J. Ohlbach C. Scheideler H.-GT. Hegering,
A. Lehmann, editor, Proceedings of the 38. Jahrestagung der Gesellschaft für Infor-
matik e.V. (GI), die INFORMATIK, pages 923–928. Köllen Druck+Verlag GmbH,
Bonn, 2008.
8. R. Klamma and Z. Petrushyna. Pattern-based competence management: On the
gap between intentions and reality. In 11th IFIP Working Conference on VIRTUAL
ENTERPRISES, Saint-Etienne, France, 11-13 October, 2010.
9. R. Klamma, M. Spaniol, and D. Denev. PALADIN: A pattern based approach to
knowledge discovery in digital social networks. In K. Tochtermann and H. Maurer,
editors, Proceedings of I-KNOW ’06, 6th International Conference on Knowledge
Management, Graz, Austria, September 6 - 8, 2006, J.UCS (Journal of Universal
Computer Science) Proceedings, pages 457–464. Springer, 2006.
10. M. D. Laat, V. Lally, L. Lipponen, and R.-J. Simons. Online teaching in networked
learning communities: A multi-method approach to studying the role of the teacher.
Instructional Science, 35:257–286, 2007.
11. B. Latour. Technology is Society Made Durable. in J. Law (ed.), A Sociology of
Monsters: Essays on Power, Technology and Domination, London: Routledge:103–
31, 1991.
12. A. Naeve, P. Yli-Luoma, M. Kravcik, and M.-D. Lytras. A modeling approach to
studying the learning process with a special focus on knowledge creation. Inter-
national Journal of Technology Enhanced Learning (IJTEL), Vol.1, Nos. 1/2:pp.
1–34, 2008. paper is based on the PROLEARN Deliverable D5.3 from 2005.
13. Z. Petrushyna, R. Klamma, and M. Kravcik. Designing during use: Modeling of
communities of practice. In IEEE International Conference on Digital Ecosystems
and Technologies, 2010.
14. Zinayida Petrushyna and Ralf Klamma. No guru, no method, no teacher: Self-
classification and self-modelling of e-learning communities. In Pierre Dillenbourg
and Marcus Specht, editors, Times of Convergence. Technologies Across Learn-
ing Contexts, volume 5192 of Lecture Notes in Computer Science, pages 354–365.
Springer Berlin Heidelberg, 2008.
15. Yoav Shoham and Kevin Leyton-Brown. Multiagent Systems: Algorithmic, Game-
Theoretic, and Logical Foundations. Cambridge University Press, 2009.
16. E. Wenger. Communities of Practice: Learning, Meaning, and Identity. Cambridge
University Press, Cambridge, UK, 1998.
17. Eric Siu-Kwong Yu. Modelling strategic relationships for process reengineering.
PhD thesis, Graduate Department of Computer Science, University of Toronto,
1995.
66