Information Systems and Software Techniques
University of Essen, Universitätsstraße 9, D-45141 Essen, Germany
Published in: Gaul, W.; Locarek-Junge, H., (Eds.): Classification in the Information Age, Springer-Verlag Berlin/Heidelberg 1999. ISBN: 3-540-65855-6
Abstract: Currently, most curricula in university are based on independent courses and do not take into account existing knowledge and learning preferences of the students. We provide a model which tries to deal with these requirements and provides more flexibility. The model is based on the division of courses into self-contained concepts and user-specific navigation structures.
The curricula of many or maybe most German universities are currently under discussion. On the one hand public universities have to face a growing competition from private universities, on the other hand there is a strong desire from government and industries to shorten the length of the studies and to allow higher flexibility and more specializations. As a consequence classes become more heterogeneous. Traditional ways of teaching are not longer sufficient to cover the continuous growing requirements on students.
These requirements result in a number of design goals for a teaching and learning system:
The following sections describe the needed definitions and terms, the resulting structures and an architecture for an implementation. The paper finishes with a discussion of the results and topics for further work.
First, the terms we need to describe our approach are defined:
Course units and courses are conflict-free preselected navigation paths over a set of concepts. A navigation path is conflict-free, if none of the pre-requirements of the covered concepts is violated. An example for the above categorization would be:
When the course material is organized in small, conceptual units the focus on knowledge transfer shifts from course-based knowledge transfer to concept-based knowledge transfer (see Figure 1).
Typically, a student takes several mostly independent courses. Using a concept-based approach, this distinction between different courses vanishes, and the student is able to progress through the course units based on her experience and preferences.
This concept-based organization of the knowledge pool is similar to the proposal by Pilar da Silva et al. (1997). Our emphasis is to provide a scalable and extensible architecture for the realization of such a system based on open Internet standards. The navigation in the system is dynamically computed for every student based on his knowledge and his needs. The cognitive aspects of the actual presentation are in our paper of minor interest, not at least because due to the use of standard web techniques the possible variations in the presentation and linking of the concepts is limited.
The level of experience and the learning preferences of the students have a direct influence on the navigation structure. Learning preferences are expressed as unknown concepts rated with a level of interest and experience is expressed as known concepts rated with a level of confidence. Together, they form the user profile of a student as seen in Figure 2.
In contrast to the course units and concepts which are mostly static, the navigation structure is dynamic. It changes whenever a student gains more knowledge about a concept or decides to change his learning preferences. Additional influences are the prerequisites which some concepts may have, resulting in additional concepts which have to be learned. Figure 3 shows the distinction between the static and dynamic parts of the system.
The part of the navigation structure which is most important are the links from a concept which has been covered completely to the possible following concepts. This set contains all those concepts with fully satisfied prerequisites (prerequisites being concepts already covered with a high confidence value). They are ranked by level of interest.
A possible alternative to this approach would use a predefined navigation structure opposed to a dynamically generated structure. The basic navigation structure could be defined by the lecturer, and modifications would be made by the students. The disadvantage however is that the students are need a certain level of knowledge to modify the predefined structure. A more detailed discussion of different approaches to the definition or generation of navigation structures is subject to further research.
The model presented in this paper is targeted at a web-based environment where the course material is available on a web server or distributed as CD-ROMs. Students are able to read the course material on- or off-line using web-browsers. In our implementation, the Extensible Markup Language (XML, see Bray et al. (1997)) is chosen to structure the course units. XML offers more functionality as HTML (Raggett (1997)) for presentation purposes, especially in conjunction with style sheets like CSS (Lie, Bos, (1997)). Furthermore, it provides an excellent means to define structure and semantics of a document through Document Type Definitions (DTDs). Finally, it has a superior definition of link semantics (Maler, DeRose (1998)). We can identify four types of XML documents:
The structure of each document type is defined in a corresponding DTD, an example for the user profile as seen in Figure 2 is shown in Figure 4.
A user profile is created when the student is enrolled for the first course. This results in an initial listing of concepts in the profile. The concepts to be learned in the course correspond to the CUNKNOWN elements and optionally already known concepts can be added as CKNOWN elements. During the student's progression through the course, the system updates these elements and changes concept references from CUNKNOWN to CKNOWN, as the student passes the corresponding control questions.
As mentioned above, the system will be used via a web browser. The resulting infrastructure is shown in Figure 5. In this scenario, the web server dynamically generates the user-specific navigation structures based on a unique user ID (implemented using the cookie mechanism, see Kristol, Montulli (1997)) and a corresponding user profile. The generation of the navigation structure is handled by CGI scripts. The resulting documents are XML documents, but it is possible to convert them to HTML documents before transmission to the client. This allows the usage of browsers which are not XML/CSS-enabled.
A more flexible approach however is the usage of a web browser which is able to execute scripts. For this purpose, we are using the Cineast browser (Köppen et al. (1997)). It is written in OTcl (Wetherall, Lindblad (1995)) and handles HTML and XML documents along with CSS style definitions. Its main advantage over commercial browsers is that the Cineast browser is a prototyping environment for web-based applications which has furthermore the ability to execute embedded scripts, permitting access to the document structure and the network functionality provided by the Cineast browser through a programming interface. This interface covers on the one hand most of the Document Object Model (DOM, see Byrne (1997)) and allows the generation and modification of XML documents. On the other hand, the networking layer of the Cineast browser can be accessed through OTcl objects, which makes it possible to fetch and store documents across the Internet.
Figure 6 shows the architecture usage of the learning environment when the Cineast browser is used. In this approach, the generation of the navigation structure is realized through scripts embedded in the course material. The user profile and the concept pool can be used off-line if the material is stored on the client side.
The static course material itself is created using XML/SGML enabled editors like Emacs, or generated from existing material, e.g. LaTeX or HTML sources.
It is obvious that static course material cannot meet the requirements which result from differing existing knowledge and learning preferences, increasing demand for more flexible course structures and changing course contents. We presented an approach that divides courses into the underlying concepts and connects these concepts using a dynamically created navigation structure. As a result, we gain the ability to create individual courses which reflect the experience and preferences of the students. They can be used for on- and off-line learning using web techniques such as XML, CSS and embedded scripts together with complying browser and server software. From the point of view of a lecturer, a pool of concepts is built rather than complete courses. This pool of concepts can be used as a basis for courses not only of a single department but of a whole faculty.
However, open for further work and discussion are the following aspects:
BRAY, T., PAOLI, J., SPERBERG-McQUEEN, C. M. (1998): Extensible Markup Language (XML), W3C Recommendation, http://www.w3.org/TR/REC-xml.
BYRNE, S. (1997): Document Object Model (Core) Level 1, W3C Working Draft, http://www.w3.org/TR/WD-DOM/level-one-core-971009.
KÖPPEN, E., NEUMANN, G., NUSSER, S. (1997): Cineast - An extensible Web Browser. Proc. WebNet 97, Toronto.
KRISTOL, D., MONTULLI, L. (1997): HTTP State Management Mechanism. Standards Track Protocol, Network Working Group, RFC2109, ftp://ftp.nic.de/pub/doc/rfc/rfc-2100-2199/rfc2109.txt.
LIE, H.W., BOS, B. (1997): Cascading Style Sheets, level 1, W3C Recommendation, http://www.w3.org/TR/REC-CSS1.
MALER, E., DeROSE, S. (1998): XML Linking Language (XLink), http://www.w3.org/TR/WD-xml-link.
PILAR DA SILVA, D., VAN DURM, R., HENDRIKX, K., DUVAL, E., OLIVIE, H. (1997): A Simple Model for Adaptive Courseware Navigation, Proc. WebNet 97, Toronto.
RAGGETT, G. (1997): HTML 3.2 Reference Specification, W3C Recommendation, http://www.w3.org/TR/REC-html32.html.
WETHERALL, D., LINDBLAD, C.J. (1995): Extending Tcl for Dynamic Object-Oriented Programming, Proc. of the Tcl/Tk Workshop '95, Toronto.
This document was generated using the LaTeX2HTML translator Version 98.1p1 release (March 2nd, 1998)
Copyright © 1993, 1994, 1995, 1996, 1997, Nikos Drakos, Computer Based Learning Unit, University of Leeds.
The command line arguments were:
latex2html -split 0 -no_navigation Adaptiv.tex.
The translation was initiated by Eckhart Koeppen on 1999-01-19