A lot of course material is nowadays available in specific e-learning systems. Most of these systems are realised as Web sites where the content of the courses is published in the form of documents, available either as static HTML pages or PDF files. Finding particular information about a topic in these sometimes large document repositories is usually done by using search engines, but they hardly yield the expected elements.
Furthermore, the results found by keyword searches arempresented in lists of more or less relevant references, users then have to go through such lists and check if the content corresponds to their needs. Likewise, finding an exercise or an example of a subject of interest is hopeless in many cases. This is so, because the heterogeneity and the non- semantically structured nature of the Web persist in e-
learning systems. Therefore, it seems necessary to involve documents in specific systems where authors can not onlymgroup them according to different topics, but also extract important concepts, definitions, exercises and examples and integrate them into the system in a formal manner.
They should have the possibility to structure information according to semantically relevant aspects easily accessible to readers.
Using search engines means following a bottom-up knowledge acquisition process while structured information in hyperbooks introduce top-down elements like hierarchies and taxonomies, which is more adaptable to human users. Besides, redundancy of document collections can be dramatically reduced.
Our aim is to develop a simple hyperbook system that enables the users to organise the stored content by hierarchies, multiples views and some automatically generated relations between concepts. Such a system has to be simple in regard of the requirements of e-learning platforms.
First, content will be under a continuous evolution process and modifications should be executable in a simple manner. Second, if users understand not only the content, but also the structure of the system, their performance will improve. The system should also enable the collaborative writing of text materials.
Over several years, we have asked students of a computer science course to collaborate in the creation of a hyperbook about the topics of the course. Teachers have built the main concepts, and students were asked to create small informational fragments and link them with these concepts.
The idea behind this approach was to get students more familial to
the domain. In the first years, such hyperbooks were built without either technical or semantic guidelines, with the result that students didn't created really meaningful links between the fragments. The resulting hyperbook never reached the state of a coherent information framework.
For this reason, in recent years, we began to define more precisely the concept structure and introduced automatic link inference. The current system has a server-side and database-backed architecture, the dynamically generated Web pages can be browsed using a standard browser and content updates are taken into account immediately.
The article is organized as follows: In section 2, we describe our virtual hyperbook model. Next in section 3, we present the role of the domain ontology for link inference and we show how adaptive mechanisms can be built based on the notion of points of view. In section 4, we explain the interface of the hyperbook and, finally, in
section 5, we conclude with an outlook about future research.
https://www.researchgate.net/publication/4104786_Adaptive_mechanisms_in_a_virtual_hyperbook