A. Holzinger (Ed.): USAB 2008, LNCS 5298, pp. 377–390, 2008.
© Springer-Verlag Berlin Heidelberg 2008
Personalized E-Learning through Environment Design
and Collaborative Activities
Felix Mödritscher and Fridolin Wild
Institute for Information Systems and New Media,
Vienna University of Economics and Business Administration
Augasse 2-6, 1090 Vienna, Austria
{felix.moedritscher,fridolin.wild}@wu-wien.ac.at
Abstract. Over the last century, many theoretical frameworks and technological
solutions for personalized e-learning have emerged. The underlying models,
however, are often based on the practice that domain experts develop an adapta-
tion strategy to personalize content or parts of a learning platform, which leads
to different problematic aspects decreasing the feasibility or utility of such ap-
proaches. After giving a brief overview of the historical development and basic
concepts of personalized e-learning, we outline the shortcomings of the tradi-
tional ‘top-down, ex ante’ models and present an alternative approach which
deals with personal learning environments, web application mashups, learning
activities and learner interactions, as well as pattern-based best practice sharing.
Furthermore, a prototypic implementation for our ‘learner-driven, bottom-up’
approach to personalized e-learning, namely the ‘Mash-UP Personal Learning
Environment’ (MUPPLE), is presented and discussed on the basis of a concrete
scenario.
Keywords: Personal Learning Environments, Learning Environment Design,
Learner Interaction Scripting, End-User Development.
1 Introduction
According to [1], personalized learning aims at “tailoring the teaching to individual
need, interest and aptitude” to ensure the most effective knowledge transfer for each
learner. In fact, the idea of adapting instructions to learner characteristics has been
considered a success factor at least since the 4th century BC, and adaptive tutoring was
a wide-spread method of education in the 18th century [2]. Moreover, a first prototypic
implementation of an adaptive assessment tool was already reported in 1926, while in
the last four decades many technology-based approaches and solutions to personal-
ized learning have emerged [3].
Historically, personalized e-learning is founded on the aptitude-treatment interac-
tion (ATI) research as well as macro and micro-adaptive instructional models. In
practice, these streams lead to technologies like Computer-Managed Instruction
(CMI), Intelligent Tutoring Systems (ITS), Adaptive Educational Hypermedia (AEH),
but they also influenced Learning Management Systems (LMS) and e-learning stan-
dards. Referring to the relevant literature [4, 5, 6], personalized adaptive e-learning

378 F. Mödritscher and F. Wild
typically include four types of models: (1) the domain model to describe learning
resources and knowledge domain, (2) the pedagogical (learner) model to characterize
the learning context and learner states, (3) the didactical model to consider typical
teaching aspects, like learning goals, course sequences, didactical requirements, etc.,
and (4) the adaptation model (rules) to specify the personalization strategies.
Commonly-known frameworks for personalized e-learning address different as-
pects, such as the conceptual idea (cf. the framework for adaptive e-learning by [4]),
the architectural design of an AEH (cf. KnowledgeTree [7]) or a formalization of
adaptive behavior (cf. the first-order logic by [8]). Younger trends also focus on stan-
dardizing personalization rules, e.g. by using XML-based specifications like IMS
Learning Design [9]. However, all these frameworks and their underlying approaches
prevent personalized adaptive e-learning from being utilized or utilizable in educa-
tional practice. In the next section we discuss the shortcomings of current approaches
and examine alternatives based on new concepts and methods from the Internet. Sec-
tion 3 summarizes our idea of a ‘Mash-UP Personal Learning Environment’ and a
first prototypic implementation. Thereafter, section 4 describes personalization of
learning with our prototype on the basis of an exemplary scenario.
2 Shortcomings of Personalized E-Learning and Novel Influences
In learning and research practice, traditional approaches to personalized e-learning
still lack important issues as outlined in the following.
First of all, the current definition of personalized (adaptive) e-learning is often re-
stricted to the context of one user interacting with instructions delivered by one sys-
tem (e.g. the LMS) which, furthermore, contains a pre-defined, up-to-date learner
model and automatically adapts to the learner (cf. [3, 4, 6, 10, 11]). In our opinion,
this definition is not sufficient to meet the requirements of the real world, and the
‘learning environment’ includes all possible entities a learner interacts with and all
influences on the learning process. More provokingly, we would even say that the
learning environment comprises ‘everything but the learner’. For the context of e-
learning we restrict this definition to all tools on the computer utilized by a learner.
Consequently, this point of view widens the scope of adaptivity and personalization.
For instance, the knowledge about a learner, a so-called user model, might be distrib-
uted over several systems and known by a peer or the facilitator only. Additionally,
adaptive behavior is not only observable in one specific system but has to be seen in
connection with a learner utilizing various tools to connect to a learning network and
collaborate with other actors on shared artifacts. As a result, adaptivity and personal-
ization take place within the whole socio-technical system, the learning network, and
not only in one educational system and on the basis of a specific learner model. The
adaptation effects, however, are only visible at the frontend of this socio-technical
system, precisely at the user interface displaying the learning tools.
Secondly, technology-driven personalized e-learning is based on ex ante, top-down
modeling (e.g. modeling of learning styles and cognitive traits depicted in [12]),

Personalized E-Learning through Environment Design and Collaborative Activities 379
requiring technological and pedagogical experts to turn a valid adaptation strategy
into functions and systemic behavior of an e-learning system, primarily following the
paradigm of didactic-awareness [13] and considering aspects of adaptable courseware
[14]. However, such strategies to automated adaptation of learning content and func-
tions might lack of validity, e.g. if learners have to self-assess their learning style [15]
or if wrong pedagogical assumptions have been made, as shown with a study on inter-
active media in [16]. Furthermore, new trends in the field of learning research (cf.
[17]) try to address the perspective of learners and suggest a learner-driven approach
to personalized e-learning, e.g. by analyzing learning behavior and recommending
learning experiences of a learning community to learners.
Thirdly, researcher and developer in the field of personalized e-learning often build
upon Learning Management Systems (LMSs) or Virtual Learning Environments
(VLEs), aiming at adapting parts of such systems, like navigational elements [18] or
systemic behavior by orchestrating services [5]. Some approaches even deal with
opening up LMSs to externally provide adaptive behavior [19]. Concerning the tech-
nological realization, the implementation of personalization in monolithic learning
environment requires high efforts in software development and evaluation and often
ignores existing tools which are used by learners and might be more efficient in a
certain learning context. Younger and more promising approaches focus on the per-
spectives of learners in terms of personal learning environments composing different
learning services into a single user experience [20].
Finally, interoperability is of importance for both e-learning and personalized e-
learning, particularly if learners actively contribute to the learning process. If not pro-
viding interoperability mechanisms [21] or considering standardized learning content
[14, 22], personalisable courseware might get isolated or even be lost within learning
platforms. On the other hand, standardizing adaptable courseware extremely increases
the complexity and the effort of creating and planning a course, as indicated with a
study in [23]. In addition to considering courseware design principles, a heterogeneous
landscape of learning tools and services, collaborative learning activities, and learner
interaction sequences, facilitators have to deal with learning design and, additionally,
with selecting appropriate adaptation models and techniques [18] and extending their
courseware, particularly on the basis of existing standards or beyond [14].
All in all, facilitating personalized e-learning experiences can be characterized with
technological boundaries, restrictions of existing standards and specifications, as well
as with more complexity and efforts for educators. New developments in the Internet,
subsumed with the term Web 2.0 [24], aim at overcoming these problematic aspects
of e-learning. The effects of the Web 2.0 on technology-enhanced learning have al-
ready been examined elsewhere, e.g. in [17, 25]. Following these principles, we pro-
pose that, in analogy to Web 2.0 principles, personalized e-learning should be based
on ‘the Web as a learning space’, allowing learners to use a variety of available tools
and content. By providing ‘rich learning experiences’ through more interactive user
interfaces or community-enabling features, learners can collaborate with peers and
actively participate in the learning process, e.g. using blogging or tagging functional-
ity. On the basis of collaborative learning activities and of open, high-quality content

380 F. Mödritscher and F. Wild
(‘the next Intel inside for learning’), learners and facilitators can ‘add value to their
learning processes’, for instance by commenting or tagging learning material or
contributing content.
On the other hand, it is also possible to analyze learning behavior in order to ‘har-
ness the collective intelligence of a networked learning community’ if personalized e-
learning is not restricted to learner interactions with one specific system. Hereby,
beneficial semantics for other peers can be provided either by the learners themselves,
e.g. by sharing learning experiences with others, or through automated mining tech-
niques trying to extract and exploit the ‘network effects of a learning community’, e.g.
by recommending tools for learning activities. Finally, from a more technological
point of view, these new influences from the Web 2.0 require new development
methods for ‘software above the level of a monolithic LMS’, being based on ‘light-
weight programming models’ like RESTful architectures [26], going beyond ‘the
software release cycles for LMSs’, and considering complex socio-technical processes
[27] in order to realize personal learning environments [20]. Thus, the ‘long tail of
software’ [28] describes the Web 2.0 idea that learners design their own personal
learning environments on the basis of available learning tools and services and ac-
cording to related learning experiences of peers, if given.
Comparing traditional approaches to our idea of ‘personalized e-learning 2.0’, we
identified significant advantages of this Web 2.0-driven development. Above and
beyond, the personalization strategy is shifted from being implemented by a few do-
main experts and facilitators (ex ante, top-down modeling) to empowering learners to
design their personal learning environments, to collaborate with peers, and to transfer
learning experiences between facilitators and peers, overall leading to a learning net-
work of actors, artifacts, and activities. Based on learner interactions, this networked
community of online learners can be supported by more sophisticated strategies, e.g.
for regulating collaborative learning or for reflecting the learning process (learner-
driven, bottom-up, just-in-time modeling). Particularly, this would decrease the plan-
ning and implementation efforts of personalized e-learning, because modeling is less
deterministic and the necessary models which are distributed and partially even out-
sourced can later be created, through involvement of the learners and on the basis of a
valid learning activity design. Moreover, each model can be developed iteratively
and, if stable enough, evaluated separately from other models.
While traditional personalized e-learning approaches seem to address a declarative
design of procedural experiences, our bottom-up approach aims at designing
procedures for creating declarative artifacts, evolving typical learn-by-heart or know-
how-to-do goals (low-level learning objectives) to the learn-how-to-learn vision
(higher-level learning objectives). Thereby, the idea of the 2.0-driven personalized e-
learning also considers that learners can reuse learning experiences by adaptive shar-
ing, cloning, or prototyping learning activities, instead of implementing and partially
refining them. Last but not least, we foster the possibility that learners can bring in
existing tools and content, instead of working with a given LMS and predefined re-
sources to master teacher-given activities. Concluding this section, we believe the

Personalized E-Learning through Environment Design and Collaborative Activities 381
above-mentioned considerations to present a promising approach to personalized
e-learning, particularly to enable lifelong learning beyond isolated learning contexts
like higher education or workplace learning.
3 The Idea of a Mash-UP Personal Learning Environment
To show personalized e-learning 2.0 in practice, we present the basic concept as well
as a first prototypical implementation of a ‘Mash-UP Personal Learning Environment’
(MUPPLE). Hereby, we start off with low-level aspects, such as the technological
infrastructure, continue with learner interaction issues, and end up with high-level,
learning-related issues behind our idea.
3.1 Technological Infrastructure
As a first step towards a technological infrastructure for a new generation of personal-
ized e-learning solutions, we build upon the Web 2.0 idea of mashups. Hereby, we
propose a so-called web application mashup [29] as one possible infrastructure for
personal learning environments; learners may also use a portal-like platform with
different widgets or different applications on their computer. Extending the idea of
traditional mashups, a web application mashup allows displaying various web-based
tools into one aggregated view within the browser. Such a solution approach needs to
consider the following issues:

− Concluding from mashup visualization techniques [30], the display of different
applications next to each other requires a certain (1) cognitive support for users
(facilitators and peers!) in order to reduce their cognitive load on working with the
system. In accordance with iGoogle, Netvibes or other providers of personalized
websites, we realized a portal-like OpenACS component, namely the XoMashup
application [29], which allows users to arrange tools along a grid layout.
− Addressing (2) controllability in the field of personalized e-learning [31], a web
application mashup has to give the control over the arrangement of and interaction
with the tools to a user. Therefore, our XoMashup component allows a user to rear-
range, minimize, maximize, reload and close each window.
− Furthermore, it is possible to launch web applications and even add new ones to
the mashup space. As usual, browser-based solutions do cause (3) technical re-
strictions. In our case, it is necessary to start full web applications with all its
scripts and style-sheets as a part of the mashup page. Thus, we implemented our
mashup solution on the basis of ‘iframes’. This may be the only way to guarantee
an own environment for each tool but may not be supported by all browsers. Fur-
ther, the usage of iframes enforces the prevention of DOM operations which would
reset the content of an iframe. Consequently, the grid-based windowing system of
XoMashup is realized with absolute positioning and the manipulation of CSS di-
rectives.

382 F. Mödritscher and F. Wild
All in all, the web application mashup solution allows learners to reuse existing
(web-based) tools and services and can be considered as a technological infrastructure
for our approach. Moreover, web application mashups are very flexible and, therefore,
useful for many other application areas as well. Nevertheless, without some kind of
underlying semantics like necessary models for personalizing the learning process, the
XoMashup component would be nothing more than a personalizable (customizable)
portal system, lacking pedagogical support for learners such as guidance or reflection.
3.2 Learner Interaction Model
Therefore, we built up a learner interaction model for describing how learners can
design their personal learning environment and interact with it. The following aspects
were taken into consideration for this model:

− In order to be (4) independent of a subject domain, we applied the Activity Theory
model in a similar way as manifested for the INCENSE system [32]. Basically, we
broke down the learning context into situations which describe the physical and so-
cial environment of learners. In such a situation, a learner is engaged in a so-called
activity which consists of actions and objects (artifacts or other outcomes) and in-
cludes tools (or tool combinations) and other actors (facilitators or peers, even in
multiple roles). Such a learning activity is meant to be our basic instructional entity
where learners actively experience a domain and construct knowledge.
− This notion of a learning activity is (5) simple and understandable for learners,
thus is considered to be of importance for a (6) scrutable systemic behavior [33]
and a good basis for experiencing further personalization strategies.
− To enable (7) reflective learning [34](p.7), we decided to bind each action to one
specific tool and one specific object in order to produce one outcome. Although
different actors can work on the same action and even produce the same outcome,
each learner only sees her own actions, and all started actions are visualized to-
gether with the corresponding object and tool.
− Additionally, these action-object-tool triples are recommended to peers on defining
and starting new actions as a certain (8) learner support. However, learners are
able to overwrite decisions and recommendations given by the system and may
build up their personal learning environment by defining own actions and objects,
bringing in own tools, and going through the actions in their own sequence to
achieve the outcomes.
− Addressing (9) learnability and efficiency, our learner interaction model is imple-
mented in the form of a domain-specific language called ‘Learner Interaction
Scripting Language’ (LISL). Table 1 shows an example of a simple activity con-
sisting of two actions. First, the learner is expected to record a short self-
description with the tool VideoWiki (http://distance.ktu.lt/videowiki/), whereby the
REST-based call for this action has to be specified and the URL for the object
‘self-description’ is determined by completing the action. Second, the learner
should go through the self-descriptions of the peers by accessing a predefined
URL, e.g. the collection containing all self-descriptions.

Personalized E-Learning through Environment Design and Collaborative Activities 383
Table 1. Exemplary LISL code for the activity ‘Getting to know each other’ consisting of two
action statements, each one bound to one object and one tool
¾ define action Compose with url http://[...]
¾ define action Browse
¾ define object ‘self-description’
¾ define object ‘descriptions of peers’ with url […]
¾ define tool VideoWiki with url http://[...]
¾ Compose ‘self-description’ using VideoWiki
¾ Browse ‘descriptions of peers’ using VideoWiki

Fig. 1. Web-based LISL interpreter displaying the interpreted LISL script from Table 1 (red
lines indicate errors, whereby the error message is shown below the command input field)
We build a web-based interpreter for LISL code, as can be seen in Fig. 1. This
scripting approach allows experienced users to code their learning activities very
efficiently, while novices can use web-based control widgets and dialogs which act as
a wrapper for the LISL statements and are shown in the upcoming section.
3.3 Higher-Level Learning Paradigms
On top of the mashup infrastructure and the learner interaction model, higher-level
learning paradigms address important issues for building and sustaining networked
communities of learners:

− With regard to [35], the constructivistic-collaborative approach to adaptive e-
learning deals with aspects of the (10) active participation of learners, motiva-
tional factors like self-esteem and (11) collaborative activities. Technologically,
such considerations lead to a high degree of interactivity also found in games and
simulations, to adaptation strategies for motivating learners, e.g. by pedagogical
agents, or to adaptivity through collaboration, most prominently addressed by the
field of Computer-Supported Collaborative Learning (CSCL). As already men-
tioned, the MUPPLE approach stresses a method to build and sustain a learning
network of actors, artifacts, and activities, which increases the motivation to learn
and aims at developing more complex competencies [36]. Similar to Web

384 F. Mödritscher and F. Wild
2.0-driven platforms like Facebook, learners require facilities to get involved into
collaborative activities and to regulate collaboration and social interactions with
peers.
− Particularly for attracting new users, the success of MUPPLE highly depends on
recognizable benefits for learners. Hereby, we foster the paradigm of (12) best
practice sharing on the basis of activity patterns which can be provided by facilita-
tors and learners to be shared within the community. Similarly to the idea of script-
ing collaborative activities [37], we apply the LISL scripting language to describe
these activity patterns. As learning activities are encoded in form of LISL scripts,
they can be exported into activity patterns and shared with other learners; vice
versa, peers can use available best practices and create their own activities out of
these patterns.
− Beside learner-driven best practice sharing, the bottom-up approach of MUPPLE
supports the (13) analysis of former learning scripts in order to personalize differ-
ent aspects of learning, e.g. by recommending action-object-tool triples to inexpe-
rienced learners or suggesting tool landscapes for certain activities. Such
personalization strategies also address the learnability and efficiency of applying
MUPPLE in practice.
4 Personalized Learning with MUPPLE
Based on our wider understanding of personalized e-learning, we build a first proto-
type considering the issues mentioned so far. Fig. 2 shows a screenshot of the
learner’s view on our exemplary activity ‘Getting to know each other’ which consists
of the two actions ‘Compose self-description’ and ‘Browse descriptions of peers’ (cf.
table 1). On the top, the header displays the activity currently opened. To the left-hand
side, learners are supplied with an overview of their own activities and can navigate
through them. By clicking on it, a learning activity is loaded and displayed in the
content area; additionally, a branch with all action-object-tool triples included is
opened simultaneously. The content area provides three different view modes of a
MUPPLE page. By choosing one of the three tabs, a learner has a view on the web
application mashup (‘preview’), an editor for the LISL code of this page (‘code’) or
the LISL interpreter (‘log’). This structure of a MUPPLE page is related to principles
of end-user development, which is closer examined in [38].
The LISL interpreter does not only show the interpreted code, but also highlights
possible errors with detailed explanations and allows entering single lines of code (cf.
Fig. 1). The preview mode, on the other hand, comprises an integrated view of all
learning tools launched so far. Each tool is located within an own window (a so-called
‘Mupplet’) with the control elements mentioned in the last section (‘reload original
URL’, ‘minimize’, ‘maximize’, and ‘close’) on the upper right side. Creation facilities
on the left-hand side allow creating new activity pages from blank or from given
patterns. Furthermore, it is possible to add action-object-tool triples to an opened
MUPPLE page, whereas possible values are recommended on the basis of all other

Personalized E-Learning through Environment Design and Collaborative Activities 385

Fig. 2. User view on activity ‘Getting to know each other’, where the screen consists of a
header (top), facilities for navigation and activity management (left), and the content area with
its three tab-views (the mashup space, the LISL code editor, and the command-line interpreter).
activity pages. A ranking strategy has not been implemented for these recommenda-
tions so far. However, features like the closeness between activities (e.g. through
derivation from the same pattern), the action-object-tool binding or social relations in
the learner network may present useful factors to rank and filter values if there are too
many of them.
In the following, a simple scenario is described to explain how personalization of
the learning process takes place on working with the MUPPLE prototype: Consider a
group of students distributed over various universities in different countries. To intro-
duce the students to each other, the facilitator decides to use MUPPLE and predefines
an activity pattern for this purpose. Based on former experiences with such a sociali-
zation exercise, this pattern includes the following action statements:

− Compose ‘self-description’ using VideoWiki
− Browse ‘descriptions of peers’ using VideoWiki
− Contact ‘two peers’ using WebMail
− Delete ‘spam entries’ using VideoWiki

Navigation and
Creation Facilities
Tabs
Header
‘Mupplet’
Action-Object-Tool Recommendation

386 F. Mödritscher and F. Wild
After introducing MUPPLE to the students, the facilitator invites them to create a
MUPPLE page from the given pattern. Now, each student instantiates a learning ac-
tivity ‘Getting to know each other’ from the pattern and, furthermore, customizes it
according to her own need. For instance, if students do not need the ‘delete’ action
and might not even have appropriate permissions within the VideoWiki tool, they
simply can remove this action from their pages. A possible result could be the MUP-
PLE page shown in Fig. 2. On the other hand, the facilitator who obviously has a
different role in this collaborative activity uses the action ‘delete’ to clean up the
collection containing the self-descriptions from spam regularly. With respect to con-
trollability, a student might want to use a web-based chat tool to communicate with
peers. So, she could bring in this tool and remove the original one, either by modify-
ing the LISL code or using web-based control widgets. Finally, personalization takes
place on creating new action statements if MUPPLE recommends action-object-tool
triples from peers. Consequently, MUPPLE also might provide observation facilities
to inform users about changes in the activities of peers or the facilitator, which are
currently not realized.
Beside personalization within the learning activities, MUPPLE and its pattern-
based best practice sharing method allow personalized learning in a broader sense.
First, improved patterns can be derived from the learning activities experienced in
practice, which, in our opinion, might increase the quality of the content and activities
within MUPPLE. Second, the activity patterns are valuable for other learners and peer
groups as well. For instance, other facilitators could reuse the learning experiences of
the above-mentioned scenario if the actors share them in terms of an activity pattern.
Hereby, the pattern-based approach might be useful to avoid the cold-start problem of
learning platforms. Third, this best practice sharing method is also advantageous with
respect to scrutability and privacy, two important issues for personalization. Activity
patterns and their underlying semantic model are simple to understand, and exporting
a LISL script easily allows filtering learner-specific lines of code. Here, the learners
can regulate the amount of a (successful) activity they want to share. Above and be-
yond, personalization and (automated) adaptation effects take place within the learn-
ing environment (i.e. the learning network) while learners (peers and facilitators) use
tools to collaborate on shared artifacts within their common activities.
Despite of the possibilities and strengths of our MUPPLE approach, a few disad-
vantages have to be outlined here. Primarily, these problems concern technological
issues. First of all, it is necessary to have a high degree of interoperability between
web applications, which now is not always the case. This specifically relates to single-
sign-on procedures and communication channels to transfer both data and events from
one application to the other. For example, the WebMail client requires authentication,
so, currently, learners have to login separately in each application. Regarding com-
munication channels, [21] propose a specification how to realize distributed feed
networks with buffered-push capabilities. We intend to further investigate these
means and will gain experiences on how they can be incorporated into LISL. It is
planned to introduce additional ‘connect’ statements for combining tools with the
abovementioned feed-based interoperability mechanism. We can think of other

Personalized E-Learning through Environment Design and Collaborative Activities 387
approaches, though, and we do not have a solution for the efficient communication of
events. Secondly, the utilization of iframes causes problems in cross-domain scripting
(cf. [39]). Finally, we are also aware that LISL and MUPPLE still lack important
functions, especially in the area of regulating collaboration and privacy, and a com-
prehensive evaluation.
5 Conclusions and Future Work
In this paper, we stated that personalization and adaptivity is much more complex in
real world learning situations, particularly if learners connect to a network of actors,
artifacts, and activities. Therefore, we consider traditional approaches for personal-
ized e-learning as not sufficiently to provide personalized learning experiences on the
computer. As a consequence, we introduced our idea of ‘personalized e-learning 2.0’,
taking into consideration that learners connect to a socio-technical network and col-
laborate on shared artifacts and outcomes. Although our prototypical implementation
of a Mash-UP Personal Learning Environments (MUPPLE) is work in progress, we
described how the learning process is adapted through environment design and

collaborative activities in networked communities. As a conclusion, we underline the
following three success factors being comprised in our model:

− First, we break up with traditional personalized e-learning models and consider the
learning environment not to a pre-condition for, but an outcome of personalized e-
learning. Therefore, we build upon learning environment design and our learner in-
teraction scripting language to be able to describe, understand, and reproduce the
outcome of learning.
− Second, the pedagogical model behind MUPPLE is very simple, so that learners
can understand how personalization works. Additionally, this activity model is a
solid basis for learning environment design and further personalization strategies,
like the automated analysis of user behavior and network effects as well as the pro-
vision of recommendations or advanced regulation facilities.
− Third, the structure and domain-independency of these learning activities address
higher-level learning objectives, independently of a subject domain, and enable
best practice sharing as well as reflective learning, i.e. aiming at paradigms like
learn-how-to-learn instead of learn-by-heart.

In total, we believe that personalized e-learning will proceed from an instructional
design and top-down, ex ante modeling to a learner-driven, bottom-up, just-in-time
adaptation of learning by considering the principles of Web 2.0. This is meant to be
‘personalized e-learning 2.0’. Our future work will address interoperability issues of
learning tools, regulation facilities for collaborative activities in learning networks as
well as experiences in real-world learning settings, particularly to evaluate the utility
of the MUPPLE approach for higher education and lifelong learning. Concerning
personalized e-learning, we have to think of further strategies to recommend (adver-
tise) pattern-based best practices to peers and to support learners in their collaborative
activities within their learning network.

388 F. Mödritscher and F. Wild
Acknowledgements
This work has been produced in the context of iCamp, a research and development
project financially supported by the European Union under the ICT programme of the
6th Framework Programme (Contract number: 027163).
References
1. Heller, J., Steiner, C., Hockemeyer, C., Albert, D.: Competence-Based Knowledge Struc-
tures for Personalised Learning. International Journal on E-Learning 5(1), 75–88 (2006)
2. Sampson, D., Karagiannidis, C.: Kinshuk: Personalised Learning: Educational, Techno-
logical and Standardisation Perspective. Interactive Educational Multimedia 4, 24–39
(2002)
3. Park, O., Lee., J.: Adaptive Instructional Systems. In: Jonassen, D.H. (ed.) Handbook of
Research on Educational Communications and Technology, pp. 651–684. Lawrence Erl-
baum Associates, Mahwah (2004)
4. Shute, V., Towle, B.: Adaptive E-Learning. Educational Psychologist 38(2), 105–114
(2003)
5. Conlan, O.: The Multi-Model, Metadata Driven Approach to Personalised eLearning Ser-
vices. University of Dublin, Dublin (2005)
6. Mödritscher, F.: Adaptive E-Learning Environments: Theory, Practice, and Experience.
VDM, Saarbrücken (2008)
7. Brusilovsky, P.: KnowledgeTree: A Distributed Architecture for Adaptive E-Learning. In:
Proceedings of the World Wide Web Conference (WWW), pp. 104–113 (2004)
8. Henze, N., Nejdl, W.: A Logical Characterization of Adaptive Educational Hypermedia.
New Review of Hypermedia and Multimedia 10, 77–113 (2004)
9. Specht, M., Burgos, D.: Modeling Adaptive Educational Methods with IMS Learning De-
sign. Journal of Interactive Media in Education (Adaptation and IMS Learning Design), 1–
13 (2007)
10. Henze, N., Dolog, P., Nejdl, W.: Reasoning and Ontologies for Personalized E-Learning in
the Semantic Web. Educational Technology & Society 7(4), 82–97 (2004)
11. Chen, C.-M., Lee, H.-M., Chen, Y.-H.: Personalized e-learning system using Item Re-
sponse Theory. Computers & Education 44(3), 237–255 (2005)
12. Graf, S., Lin, T., Kinshuk, J.: The relationship between learning styles and cognitive traits
- Getting additional information for improving student modelling. Computers in Human
Behavior 24(2), 122–137 (2008)
13. Helic, D.: A Didactics-Aware Approach to Management of Learning Scenarios in ELearn-
ing Systems. Graz University of Technology, Graz (2006)
14. Mödritscher, F., García-Barrios, V.M., Gütl, C.: Enhancement of SCORM to support adap-
tive E-Learning within the Scope of the Research Project AdeLE. In: Proceedings of the
World Conference on E-Learning in Corporate, Government, Healthcare, & Higher Educa-
tion (E-Learn), pp. 2499–2505 (2004)
15. Phillips, M.: Cognitive Style’s influence on Media Preference: Does it matter or do they
know? In: Proceedings of the World Conference on Educational Multimedia, Hypermedia
& Telecommunications (ED-MEDIA), pp. 1023–1028 (2005)
16. Holzinger, A., Kickmeier-Rust, M., Albert, D.: Dynamic Media in Computer Science Edu-
cation; Content Complexity and Learning Performance: Is Less More? Educational Tech-
nology & Society 11(1), 279–290 (2008)

Personalized E-Learning through Environment Design and Collaborative Activities 389
17. Rollett, H., Lux, M., Strohmaier, M., Dösinger, G., Tochtermann, K.: The Web 2.0 way of
learning with technologies. International Journal of Learning Technology 3(1), 87–107
(2007)
18. Brusilovsky, P.: Methods and techniques of adaptive hypermedia. User Modeling and
User-Adapted Interaction (UMUAI) 6(2-3), 87–129 (1996)
19. Gütl, C., Mödritscher, F.: Towards a Generic Adaptive System applicable for Web-based
Learning Management Environments. In: Proceedings of the Annual Workshop of the SIG
Adaptivity and User Modeling in Interactive Systems (ABIS), pp. 26–31 (2005)
20. Van Harmelen, M.: Personal Learning Environments. In: Proceedings of the International
Conference on Advanced Learning Technologies (ICALT), pp. 815–816 (2006)
21. Wild, F., Sigurdarson, S.E.: Distributed Feed Networks for Learning. European Journal for
the Informatics Professional (UPGRADE) 9(3) (2008)
22. Santos, O.C., Boticario, J.G.: Meaningful pedagogy via covering the entire life cycle of
adaptive eLearning in terms of a pervasive use of educational standards: The aLFanet ex-
perience. In: Tomadaki, E., Scott, P. (eds.) EC-TEL 2006. LNCS, vol. 4227, pp. 691–696.
Springer, Heidelberg (2006)
23. Mödritscher, F.: e-Learning Theories in Practice: A Comparison of three Methods. Journal
of Universal Science and Technology of Learning (JUSTL) 0(0), 3–18 (2006)
24. O’Reilly, T.: What is Web 2.0. O’Reilly Media Inc. (2005) (2008-07-21),
http://www.oreilly.com/pub/a/oreilly/tim/news/2005/09/30/wha
t-is-web-20.html
25. Ebner, M., Holzinger, A., Maurer, H.: Web 2.0 Technology: Future Interfaces for Tech-
nology Enhanced Learning? In: Stephanidis, C. (ed.) HCI 2007. LNCS, vol. 4556, pp.
559–568. Springer, Heidelberg (2007)
26. Lathem, J., Gomadam, K., Sheth, A.: SA-REST and (S)mashups: Adding Semantics to
RESTful Services. In: Proceedings of the International Conference on Semantic Comput-
ing (ICSC 2007), pp. 469–476 (2007)
27. Jones, S., Maiden, N.A.M.: RESCUE: An Integrated Method for Specifying Requirements
for Complex Socio-Technical Systems. In: Mate, J.L., Silva, A. (eds.) Requirements Engi-
neering for Socio-Technical Systems, pp. 245–265. Idea Publishing (2004)
28. Kraus, J.: The long tail of software: Millions of Markets of Dozens. Bnoopy: An entrepre-
neurship blog (2005) (2008-07-21), http://bnoopy.typepad.com/bnoopy/
2005/03/the_long_tail_o.html
29. Mödritscher, F., Neumann, G., García-Barrios, V.M., Wild, F.: A Web Application
Mashup Approach for eLearning. In: Proceedings of the OpenACS and.LRN Conference,
pp. 105–110 (2008)
30. Spoerri, A.: Visual Mashup of Text and Media Search Results. In: Proceedings of the In-
ternational Conference Information Visualization, pp. 216–222 (2007)
31. García-Barrios, V.M., Mödritscher, F., Gütl, C.: Personalisation versus Adaptation? A
User-centred Model Approach and its Application. In: Proceedings of the International
Conference on Knowledge Management (I-KNOW), pp. 120–127 (2005)
32. Akhras, F.N., Self, J.A.: System Intelligence in Constructivist Learning. International
Journal of Artificial Intelligence in Education 11(4), 344–376 (2000)
33. Kay, J.: Stereotypes, Student Models and Scrutability. In: Proceedings of the International
Conference on Intelligent Tutoring Systems (ITS), pp. 19–30 (2000)
34. Boud, D., Keogh, R., Walker, D. (eds.): Reflection: Turning Experience Into Learning.
Routledge, Abingdon (1985)
35. Koper, R., Olivier, B.: Representing the Learning Design of Units of Learning. Educa-
tional Technology & Society 7(3), 97–111 (2004)

390 F. Mödritscher and F. Wild
36. Mödritscher, F., García-Barrios, V.M., Gütl, C.: The Past, the Present and the Future of
adaptive E-Learning: An Approach within the Scope of the Research Project AdeLE. In:
Proceedings of the International Conference on Interactive Computer Aided Learning, ICL
(2004)
37. Dillenbourg, P., Jermann, P.: Designing integrative scripts. In: Fischer, F., Mandl, H., Ha-
ake, J., Kollar, I. (eds.) Scripting computer-supported collaborative learning: Cognitive,
computational and educational perspectives, pp. 277–302. Springer, New York (2007)
38. Wild, F., Mödritscher, F., Sigurdarson, S.E.: Designing for Change: Mash-Up Personal
Learning Environments. eLearning Papers 9 (2008)
39. Jackson, C., Wang, H.J.: Subspace: Secure Cross-Domain Communication for Web Mash-
ups. In: Proceedings of the International Conference on World Wide Web (WWW), pp.
611–620 (2007)