Docstoc

Adaptive Authoring of Adaptive Hypermedia

Document Sample
Adaptive Authoring of Adaptive Hypermedia Powered By Docstoc
					                           Adaptive Authoring of Adaptive Hypermedia

Yes, you’ve read it correctly! There are two ‘adaptive’s in that title. ‘ Adaptive Authoring of Adaptive
Hypermedia’ is my current research field and interest. What does this mean? Well, let’s start from the very
beginning.

Adaptive hypermedia (AH) is a relatively new but exciting field, opening many possibilities, especially in
the educational domain, but now-a-days also for commercial applications.

How did adaptive hypermedia emerge? Well, its forefathers are hypermedia and adaptivity, as the name
says. But the actual source of AH lies in web-based educational systems. These systems arose to take
advantage of the potential inherent in the Web; systems such as WebCT and Blackboard, and they are
extremely popular around the world. They reflect both the good and the bad facets of current ‘popular’
educational objectives – they reach many potential learners, but fall into the ‘one approach for all’ trap.
They assume that all learners are the same, that their educational background, abilities and motives are also
the same. Hence a single topic/subject/discipline is taught (or more accurately ‘presented’) in a uniform
manner.
Educationalists have known for years that which the Web community at large is only now appreciating:
different students learn differently. They have different abilities and backgrounds that should be addressed
in a personalised and individual manner. The discipline of Adaptive Hypermedia and more specifically
Adaptive Educational Hypermedia (AEH) has arisen to address this fundamental deficiency in many of the
current Web-based education systems. AEH draws upon experience in the areas of pedagogy, psychology,
intelligent tutoring systems and hypermedia. The ultimate objective of AEH is to understand and describe
how to create the ‘perfect’ online lesson for every learner – utilising a common set of learning resources.

Similarly, the ultimate objective for adaptive hypermedia is to create the ‘perfect’ online material for every
user. 1

    So, the second ‘adaptive’ from the title means adaptation to the end user: to the learner using an
educational site, or to the consumer using a commercial one. It may mean adding ‘intelligent’ behaviour to
our sites, but as we have serious trouble defining what this ‘intelligence’ actually means (see the TU/e
Intelligent Systems course) we contain ourselves with calling them adaptive sites.

The above history lead to the fact that today there are many research groups that have created their own
AEH or AH systems, each to their own unique specifications, making it difficult to reuse material created
for one system in another. This means that any system change determined by system extensions or
departmental politics would require a full rewrite of any created material. As yet however no combined
effort has been made to extract the common design paradigms and adaptation patterns from these systems.

Therefore, one of our main research goals is to find adaptation patterns for adaptive hypermedia. This
research is done under the umbrella of a Minerva Socrates programme called ADAPT. More information
can be found at: http://wwwis.win.tue.nl/~acristea/HTML/Minerva/index.html
As you may guess, these patterns can be reused to interface AH systems, or to author new AH contents and
behaviour, or even to filter AH information from the world wide web.

Back to the educational domain, one of the possible common design paradigms for AEH is, learning styles.
Learning Style models have been researched and used for decades. They address the fundamental
psychological issue that there are substantial differences between individuals and their cognitive
mechanisms, by which we all learn. Recently a small subset have been implemented in current AEH
systems. For example: WHURLE, CS383 and ILASH all implement different parts of the Felder-Solomon
ILS (Index of Learning Styles). Others such as INSPIRE uses Kolb’s theory of experiential learning; or the
Dunn and Dunn model as used in iWeaver.

1
 This may mean that a commercial site is only suggesting to one to buy the goods that one really
needs/desires, for instance.
Our research on adaptive hypermedia has two main axes:
     one is finding a more appropriate and more general data storage and manipulation model for
        adaptive response, partially by extending AHAM (Wu 2002), an existing adaptive hypermedia
        reference model. Our modeling efforts generated LAOS, a five-layer authoring model for adaptive
        hypermedia and LAG, a three-layer adaptation model;
     the other direction is to implement an authoring tool for adaptive hypermedia, which is to
        instantiate the created models. This is based on the fact that we have found in previous research
        that the main reason that adaptive hypermedia is not being not more widely implemented lies in
        the difficulty of authoring for such environments. We are trying to solve these problems by
        finding methods and techniques for the transforming of the authoring process into an easier task,
        mainly by automating many of the common authoring tasks As well as extending this automating
        towards a new version of adaptation: designer adaptation. Research towards this goal has
        generated MOT, a designer adaptive authoring environment for WWW adaptive hypermedia
        authoring.

The LAOS model, a five-layer adaptive hypermedia authoring model, specifies a flexible framework for
(collaborative) adaptive hypermedia authoring. Figure 1 shows LAOS with its:
        domain model (DM) for the resources of the sites,
        goal and constraints model (GM) for the extra information about the intention of the presentation,
        user model (UM) to capture user characteristics such as user knowledge or interests,
        adaptation model (AM), further refined in the LAG model (Figure 2), and finally
        presentation model (PM), showing adaptation information about the machine/ device (laptop,
         palmtop, etc.) the user will see the information on.


                                               • lowest level: direct adaptation techniques/ rules
                                                                                 techniques/
                                                  – adaptive navigation support & adaptive presentation
                                                  – implem.: AHA!; expressed in AHAM syntax
                                                  – techniques usually based on threshold computations of variable-value
                                                    pairs.
                                               • medium level: adaptation language                                          Adaptation
                                                  – more goal / domain-oriented adaptation techniques: based on a           Assembly
                                                    higher level language that embraces primitive                            language
                                                  – low level adaptation techniques (wrapper)
                                                  – new techniques: adaptation language

                                               • high level: adaptation strategies                                   Adaptation
                                                  – wrapping layers above                                           Programming
                                                  – goal-oriented
                                                                                  Adaptation                          language
                                                                                 Function calls

                                                                     Figure 2. The three layers of adaptation


                                                The LAG model introduces a three-layer model and
                                                classification method for adaptive techniques: direct adaptation
                                                rules, adaptation language and adaptation strategies. The
                                                benefits of this model are twofold: on one hand, the granulation
  Figure 1. The five level AHS authoring model.
                                                level of authoring of adaptive hypermedia can be precisely
                                                established, and authors therefore can work at the most suitable
level for them. On the other hand, this is a step towards standardization of adaptive techniques, especially
by grouping them into a higher-level adaptation language or strategies. In this way, not only adaptive
hypermedia authoring, but also adaptive technique exchange between adaptive applications can be
enabled.

To make the authoring burden lighter, automatic authoring techniques are being researched. To do this we
are aiming at exploiting the LAOS structure. These techniques consist of automatic transformation (and
interpretation) of rules between the different layers of the model. This results of which populate some
layers based on the contents of others. This generates automatic linking based on semantics, or automatic
content labeling, etc. For this particular line of research various techniques can be used, e.g., machine
learning.

   This is where the first ‘adaptive’ in the title comes from: the adaptation to the designer or author, the
automatic processing of what the system ‘thinks’ might be his requests. These transformation rules
represent designer-goal oriented adaptation techniques. Therefore, our research represents yet another step
towards an adaptive hypermedia that ‘writes itself’. Again, this may be called another type of ‘intelligence’.
For the practical part, we have built and are continuing to develop MOT, an AH authoring system. You can
have a look at the latest online MOT version:
      for static authoring at: http://e-learning.dsp.pub.ro/mot/
      for dynamics authoring at: http://e-learning.dsp.pub.ro/motadapt/

For testing MOT’s implementation of patterns and interfacing capabilities, we use it as an authoring
system. MOT represents in this way what authors might be expecting. As example delivery systems, we
have started interfacing to two famous systems: WHURLE and AHA!, to see what current state of the art
AH systems can provide. The aim is to establish a "write once, use many" methodology for content and
adaptation creation for current AH environments.

For testing the actual usability with real users, we have already done two interesting experiments of
combining teaching and research: the testing of MOT, an adaptive hypermedia authoring tool based on the
LAOS adaptive hypermedia authoring framework, via a class of about twenty graduate students from the
Eindhoven University of Technology, taking a two week intensive course in Adaptive Systems and User
Modeling; and a class of about thirty five undergraduate students from the ‘Politehnica’ University of
Bucharest, Romania, taking a class on ‘Adaptive Hypermedia’. The results of these tests were fed back and
are valuable for our further research.

Concluding, we can say that Adaptive Hypermedia (AH) could be the answer to the imperative need of
personalization for the web. For everyone to be able to easily create AH, however, robust and flexible AH
authoring tools must be designed, giving various choices of ready-made, easy-to-use delivery solutions to
various possible authoring problems. We believe this is an interesting new domain with many open
questions and much work to be done. Therefore we’re inviting interested students to come join us and do a
final Master project or an internship within the above framework.


                    Alexandra I. Cristea received her IS Dr. title and worked as Research Associate at the University
                    of Electro-Communications, Tokyo, Japan. She is presently Assistant Professor in the IS Group,
                    Faculty of Mathematics & Computer Science, TU/e. Her research interests include Adaptive
                    Hypermedia authoring, User Modeling, Semantic Web, AI, Neural Networks, Adaptive Systems,
                    Concept Mapping, ITS, Web-based Educational Environments. She authored and co-authored
                    over 90 research papers and course booklets. She is a member of IEEE, was program committee
                    member of Hypertext, AH, ICCE, ICAI, IKE (a.o.) and was reviewer or session chair for many
                    conferences; she is executive peer reviewer of the ET&S Journal.
                    More info: http://wwwis.win.tue.nl/~acristea/

				
DOCUMENT INFO