The Web has revolutionized the way information is delivered to people throughout the world. It did not take long for learning material to be delivered through the Web, by using electronic textbooks. The use of hypertext links gives the learner a lot of freedom to decide on an order in which to study the material. This leads to problems in understanding the textbook, which can be solved by using methods and techniques. In this chapter we describe how the field of educational hypermedia benefits from and . We also show that the information gathered about the learners and their learning process can be used to improve the quality of the electronic textbooks.

This paper deals with a new challenge in Adaptive Hypermedia (AH) and web-based systems: finding the adaptation language to express, independently from the domain model or platform, the intelligent, adaptive behaviour of personalised web courseware. The major requirements for the ideal language are: reuse, flexibility, high level semantics, and ease of use. To draw closer to this ideal language, we compare two such language proposals: LAG, a generic adaptation language, and a new XML adaptation language for Learning Styles (LS) in AHA!, LAG-XLS.

In this paper we present the second experiment in combining teaching and research: the testing of MOT and its new adaptive patterns. MOT is an adaptive hypermedia authoring tool based on the LAOS adaptive hypermedia authoring framework. The patterns are implemented via an adaptive language that uses a low granularity domain model, LAOS, to extract the adaptation alternatives. The tests were performed in a class of over thirty students enrolled in the fourth year of the University "Politehnica" of Bucharest, taking a two week intensive course in Adaptive Hypermedia. The focus of this paper is on the experiment itself, and its parameters: the setting and initial planning, the implementation and the results. Finally, we will comment on the results and interpret them.

Adaptive Hypermedia (AEH) is considered, in principle, superior to regular Hypermedia, due to the fact that it allows personalization and customization. However, the creation of good quality AEH is not trivial. Nowadays, a lot of research concentrates on the authoring challenge in adaptive hypermedia. We previously introduced the LAOS model, a five-layer adaptive hypermedia authoring model that describes AEH in a detailed way, to allow flexible re-composition of its elements, according to the personalization requirements. However, such a detailed structure claims a lot of time to populate with AEH instances. Alternatively, we propose semi-automatic authoring techniques that populate the whole structure based on a small initial subset that has been actually authored by a human. We analyze here the different possible initial subsets, and the resulting structures, based on the LAOS architecture. Moreover, we examine if the flexibility of the whole was in any way affected by the replacement of human authoring with automatic authoring. We see the latter as yet another step towards adaptive hypermedia that 'writes itself'.