KBS-HYPERBOOK - An Open Hyperbook System for Education Peter Fröhlich, Wolfgang Nejdl, and Martin Wolpers Institut für Rechnergestützte Wissensverarbeitung University of Hannover, Lange Laube 3, 30159 Hannover, Germany {froehlich, nejdl, wolpers}@ kbs.uni-hannover.de Abstract. Educators are currently investing much effort into the conversion of their textbooks and lectures into hypertext documents. Within our KBS Virtual Classroom project, we have recently developed a data-base oriented method for systematically designing such hyperbooks. However, converting theoretical material into a hyperbook is only the first step towards a modern, project-oriented, computer-supported learning environment. To realise such an envi- ronment, the static course materials (regardless whether they are printed textbooks or hyper- books) should be replaced by an open hypermedia information system, to which both the edu- cators and the students contribute. In this spirit we have redesigned our hyperbook modeling technique to make it suitable for both capturing theoretical lecture material and providing a structure for the student projects. Introduction In our KBS Virtual Classroom Project we work on providing students with complete teaching and learning envi- ronments for several subjects of computer science (artificial intelligence, software engineering and introduction to computer science I (CS1)). To encourage project-oriented education, our lecture notes consist of a complete project accompanied by the necessary theoretical background. The structure and content of these lecture notes are described by a set of hyperbook data models forming a skeleton for the presented project and the accompanying theoretical background and providing a documentation of content and structure of our hyperbook. Additionally, these models serve as a skeleton for the student projects, leading to an open hyperbook knowledge base which is continuously enhanced and extended each year by student projects. Previous Modeling Approaches The most frequently used approach to hypermedia modeling is to use no models at all. Web pages are written and hyperlinks are hard coded among them. Hypertext documents designed in this way have a poor structure, in which users get lost quickly. Recently, the structured design of hyperbooks (and hypermedia in general) has re- ceived increasing attention. We identify two research directions relevant for the design of educational hyper- books: 1. Systematic object-oriented modeling techniques developed for generic hypermedia systems like web servers and database front-ends. The Hypertext Design Method (HDM) [Garzotto et al. 93], the Relationship Man- agement Methodology (RMM) [Isakowitz et al. 95], and the Object-Oriented Hypermedia Design Model (OOHDM) [Barbosa et al. 96] are the most prominent examples. These methods structure a domain into a set of types or classes and their relationships. By classifying the pages into these types, navigation links can be inferred automatically from the relationships. 2. Specific approaches for educational systems have been investigated by the adaptive hypertext community [Brusilovsky et al. 96a,b], [Kay et al. 94], [Dale et al. 96], [Cleary et al. 94]. Usually the lecture pages, which can have an arbitrary structure, are indexed by the high level concepts of a user model. The knowledge de- rived from the user model can either modify the basic navigation structure of the document, e.g. the system