InterBook: an Adaptive Tutoring SystemJohn Eklund
Access Australia Cooperative Multimedia Centre
Carnegie Mellon University
A more detailed paper on InterBook was presented by the authors at the Apple University Consortium Conference, "Flexible Learning", September, 1998.
Most Computer Mediated Communication (CMC) tools rely heavily on a human teacher to individualise the lesson through personal email or phone contact between instructor and student. The content is almost always static. However, many web-based educational applications are expected to be used by very different groups of users with minimal assistance from a human teacher, and hence there is a need for systems which can themselves adapt to learners with very different backgrounds, prior knowledge and learning goals. An electronic textbook is one of the most promising varieties of web-based educational systems, and is appropriate for the delivery of structured textbook-style content. In this paper we describe an approach for developing adaptive electronic textbooks through InterBook, an authoring tool based on this approach which simplifies the development of adaptive electronic textbooks on the web.
Teaching and knowledge
Before InterBook in particular or adaptive systems in general are introduced, it is worthwhile considering issues of teaching and knowledge, and how they are represented in instructional systems. One of the most fundamental and persistent questions in teaching and learning in both real and virtual environments is that of learner control. Jacobs1 in a review of hypermedia and discovery-based learning demonstrates that there is a long history of this issue before the advent of hypertext or computers. How closely should a learner be monitored and directed? What freedoms should they be given to follow their interests and learn of their own accord? How should their progress be modelled and what kind of intervention strategies should be used to guide them in their learning? Instructional Technologists call it the instructionist verses constructivist debate, in educational computing it is known as the question of the locus of control.
Certainly the potential of courseware to interact with the student has improved over the years, largely because designers of software can write for more powerful computers. Entry-level computers can provide excellent graphics, mass hard-disk storage, several megabytes of memory, and sound capabilities. Developments in peripheral devices such as CD-ROM provide scope for developers of educational software to encode more complex routines using this more interactive media. Multimedia has enhanced the cognitive flexibility of computer based learning through its ability to restructure knowledge presentations to meet changing situations.
These new features of microcomputers have become available in a very short time, and affordable hardware continues to develop at a startling rate. There is therefore good reason to think that the capabilities of tutorial software written for the next generation of computers will far exceed today's standards, and provide a platform for a far more natural discourse between the student user and the computer. However, the success of the media as an instructional aid relies as much on its ability to parallel the cognitive processes of the learner through tracing the student's acquisition of knowledge as it does on its ability to present the lessons in a stimulating and interactive way2. The recent improvements in computer hardware, interface design and multimedia presentation have not addressed the problem of providing the computer with an understanding of the user, which we maintain is a critical attribute of an effective teacher. The complex field of user modelling is another new research area that has arisen from attempts to address exactly this problem.
Since microcomputers were introduced into education in the late 1970's one of their primary uses has been as a teaching tool through the use of educational software. Some of the tutorial software, or courseware, over the past decade has been little more than a screen by screen presentation of content with questions which, when correctly answered, progress the student to the next screen of information. A common complaint, among many others related to implementing tutorial software in educational settings, has been an incapacity to adapt to the particular learning needs of the student user. An experienced teacher would know from the individual responses of the student which aspect of the work requires repetition, how explanations can be restated and what questions to put to the student before additional material can be given. The teacher can be considered to have developed a cognitive model of the student, that is, an understanding of the learner's current knowledge in the subject domain, and has at his/her disposal a variety of teaching strategies to be deployed depending on student feedback and the nature of the information to be presented. The tutor works with a student model, a teaching model containing strategies for effective teaching, and an understanding of the lesson content including the relationships between the component parts in the knowledge domain, and is able to make adjustments to the flow of the tutorial by responding to the student.
While it has been generally argued that one of the strongest reasons for using computers in education is their ability to provide individualised instruction in tutor mode3,4, one of the problems in making computer software more individualised and the student-computer discourse more natural is that as more branching capabilities are required to meet the enormous number of possibilities in student responses, it soon becomes impossible to implement. This is one of the reasons for the development of intelligent tutoring systems and subsequently adaptive educational systems.
Adaptive Educational Systems
While adaptive hypermedia5 is a new direction of research within the area of adaptive and user-model based interfaces, the goal of adaptivity has featured in the design of intelligent systems for a considerably longer period. It is worthwhile at this point to ask "What is adaptivity?" Adaptive hypermedia systems are capable of altering the content or appearance of the hypermedia on the basis of a dynamic understanding of the individual user, to adapt the content or presentation to certain characteristics of the user. The noted psychologist Piaget described intelligence as the ability of an organism to assimilate and to adapt its environment, and it is generally accepted that there are organisms which are more or less intelligent than others.
An Intelligent Tutoring System is typically strongly adaptive, working in a well-structured information space; gathering data about the user's movements and using this information to dynamically modify the presentation and functionality of the system in clearly defined ways. It is also important to remember that adaptivity is not a technology, but a goal. Adaptivity is a common functional goal of intelligent systems. In summary, for a system to be called an adaptive hypermedia system it must have the following characteristics:
Adaptivity may be at the content level or at the link level. Content-level adaptivity is the dynamic generation of content based on a user model. Link level adaptivity on the other hand assumes a static content, and the appearance or prominence of the links connecting elements of this hyperspace is altered. This is what is termed adaptive navigation support6. Adaptive navigation support includes the use of user model based maps, link sorting, hiding and annotation. Adaptive annotation augments the links with a comment which provides the user with information about the current state of the nodes behind those links7,8,9,10. This method has been shown to be especially efficient in educational hypermedia11,12,13 and this is the particular technology used in InterBook. Link annotations can be provided in textual form or in the form of visual cues, for example, using different icons, or colours, font sizes, or font types and so forth.
InterBook - An adaptive educational system using link annotation
As a web-based computer mediated courseware delivery tool, InterBook has no conferencing facilities, and may be best described as an environment in which structured textbooks could be presented in a multiply-navigable interface. Any knowledge base that contains reasonably specific and identifiable knowledge elements that can be organised hierarchically into sections, subsections and indexed in detail is suitable for delivery through the InterBook system. Technical and software manuals are an excellent example of suitable material. InterBook takes the structures commonly found in such a textbook (such as tables of content, indexes and glossaries) and delivers them on the web with navigation support, providing individualised assistance to each learner. All InterBook-served electronic textbooks have a generated table of content, a glossary, and a search interface. In InterBook, the structure of the glossary resembles the pedagogic structure of the domain knowledge in that each node of the domain network is represented by a glossary entry. Likewise each glossary entry corresponds to one of the "domain concepts". All sections of an electronic textbook are indexed with "domain-model concepts".
The knowledge about the domain and about the textbook content is used by InterBook to serve a well-structured hyperspace. In particular, InterBook generates contextual links between the glossary and the textbook. Links are provided from each textbook section to corresponding glossary entries for each of the involved background or outcome concepts. Similarly from each glossary entry which describes a concept InterBook provides links to all textbook units that can be used to learn this concept. This means that an InterBook glossary integrates features of an index and a glossary. These links are not stored in an external format but generated 'on the fly', in other words dynamically, by a special module that takes into account the student's current state of knowledge represented by the user model.
InterBook uses coloured bullets and different fonts to provide adaptive navigation support (Figure 1) through link annotation. Annotations are evident in the individual section of the text (in the textbook window) as well as in a local overview map (on the navigation bar). Wherever a link appears on a page (in the table of content, in the glossary or on a regular page), the font and the colour of the bullet informs the user about the status of the node behind that link. Currently four colours and three fonts are used. A green bullet and bold font means "ready and recommended", i.e. the node is ready-to-be-learned but still not learned and contains some new material. A red bullet and an italic font warn about a not-ready-to-be-learned node, while white means "clear, nothing new" (i.e. all concepts presented on a node are known to the user). Violet is used to mark nodes which have not been annotated by an author. A check mark is added for already visited nodes. A node is annotated green when all of the prerequisite concepts for that node have been met. In other words, the particular user has previously visited a node or nodes which have those prerequisite concepts listed as outcome concepts. Obviously the initial node will have no prerequisite concepts, only outcome concepts. A node is annotated red (a not-recommended node) when it contains prerequisite concepts that have similarly not been met. Outcome concepts are 'met' when a node is simply visited.
Figure 1. Adaptive link annotation in InterBook using the traffic-light metaphor
Certainly, in order for this form of markup to be useful, the textbook needs to be used in a particular way: It assumes that the path of a user will be approximately linear (in the sense that 'linear' means the same path through the information space as is the optimal path sequenced by the author). Suppose for example that a learner decides to enter the textbook at section 2 rather than section 1. They may decide from the headings that this general topic is rather simple and not want to begin at section 1. In this case they would receive 'not recommended' annotations throughout their subsequent movement through the hyperspace. This problem, and the fact that InterBook assumes that a page is learned when it is simply accessed, prompted the development of embedded tests in later versions. Later versions of InterBook integrate all three methods of annotation: history-based (on the basis of where the user has been), prerequisite-based (on the basis of what prerequisite nodes the user has visited), and knowledge-based (on the basis of the user's demonstrated understanding of the content).
The user model in InterBook is initialised from the registration page via a stereotype model, and is modified as the user moves through the information space. The user model consists of an individual file in a folder called "users", which is updated as the student progresses through the material.
The InterBook approach uses two kinds of knowledge: knowledge about the domain being taught (represented in the form of a domain model) and knowledge about the students (represented in the form of individual student models). The domain model serves as a basis for structuring the content of an adaptive Electronic Textbook.
The simplest form of domain model is just a set of domain concepts. What we call 'concepts' are named differently in different research papers - attributes, topics, knowledge elements, objects, learning outcomes, but in all cases they are just elementary pieces of knowledge for the given domain. A more advanced form of the domain model is a network, with nodes corresponding to domain concepts and links reflecting several kinds of relationships between concepts. This network represents the structure of the domain covered by a hypermedia system. The domain model provides a structure for representation of the student's knowledge of the subject.
For each domain model concept, an individual student's knowledge model stores some value which is an estimation of the student knowledge level of this concept. This type of model (called an overlay model) is powerful and flexible: it can independently measure the student's knowledge of different topics. The overlay student model can be kept up-to-date relatively easily. All student actions (page visits, problem-solving, quizz answering) are tracked and used to increase or decrease knowledge levels for involved concepts. Another component of the student model is the model of a student's individually assigned learning goal. Adaptive guidance mechanisms will ensure that the student achieves a sequence of assigned learning goals.
Authoring an adaptive electronic textbook can be divided into 5 steps. In brief, an Electronic Textbook is prepared as a specially structured Word file and the task is to convert this file into InterBook format. InterBook recognises the structure of the document through the use of headers using predetermined text styles. The second step in the authoring process involves concept-based annotation of the Electronic Textbook to let InterBook know which concepts stand behind each section. An annotation is a piece of text of special style and format inserted at the beginning of each section (between the section header and the first paragraph). For each unit the author provides a set of outcome and background concepts. In this way, each section is annotated with a set of prerequisite concepts (or terms which exist in other sections which should be read before the current section), and a set of outcome concepts (terms which will be assumed known once the reader has visited the section).
Once the annotations are complete the file is saved in RTF format. The RTFtoHTML program (http://www.sunpack.com/RTF/) with some special settings is used to convert the Electronic Textbook into HTML format. Then the ".html" extension on the file is manually altered to ".inter" so that it can be recognised by the Interbook system. Lastly, when the InterBook server starts, it parses all InterBook files in its "Texts" folder (i.e. all files with extension .inter) and translates it into the list of section frames. Each unit frame contains the name and type of the unit, its spectrum, and its position in the original HTML file. The obtained LISP structure is used by InterBook to serve all the available textbooks on the WWW providing the advanced navigation and adaptation features. The content which is presented to the student is generated on-the-fly using the knowledge about the textbook, the student model, and HTML fragments extracted from the original HTML file. These features of InterBook are based on the functionality of the Common Lisp Hypermedia Server14.
Application of adaptive systems to flexible delivery in higher education
We believe this tool significantly simplifies the design of an adaptive electronic textbook on the web. It provides full support in preparation and serving such a textbook for authors who need only to be familiar with a word processor.
Brusilovsky, Eklund and Schwarz15 made the case for the use of knowledge-based systems in higher education as one which provides a unique solution to the fact that students are entering web-based instruction with very different knowledge, goals and backgrounds. Most courseware is aimed at the 'average' student, and does not account for minority groups, such as those with language difficulties, exceptional skill or poor subject knowledge. Accordingly, three steps in courseware development were suggested: first, the development of the content implemented with a variety of instructional strategies such as questions, interactive examples and problems; second, the refinement of the materials to suit the requirements of the current student population; and third, the use of adaptive mechanisms which personalise the hyperspace of individual students to account for individual knowledge and preferences. Personalising course materials on these three levels of granularity is not only part of the process of authoring effective courseware, but also critical in utilising the computer as an instructional medium offering some of the benefits of one-to-one human tutoring.
This paper supports the use of knowledge-based systems, in particular adaptive educational systems, for the flexible delivery of course materials in higher education. These systems recognise the importance of individual learner knowledge, as adaptive systems can customise courseware and take part of the role of the human teacher in individualising instruction. We have based our arguments on a practical model of the role of knowledge in teaching and learning, and demonstrated adaptivity in computer based educational environments through InterBook, a tool for delivering adaptive textbooks on the World Wide Web. InterBook uses adaptive annotation technology, a form of adaptive navigation support, to augment hyperlinks with a comment which informs users about the current state of the nodes behind the annotated links, and does this on an individual basis by adapting links through a user-model, an individual record of a student's progression through the courseware.
UniServe Science News Volume 12 March 1999
Page Maintained By: PhySciCH@mail.usyd.edu.au