Knowledge Management for a Tutoring System

---

Knowledge is a principal factor that makes personal, organizational, and societal intelligent behavior possible (Wiig, 1995). Knowledge management consists of activities focused on the organization gaining knowledge from its own experience and from the experience of others, and on the judicious application of that knowledge to fulfill the mission of the organization (Wiig, 1995). In the context of a learning environment, such an organization consists of a group of students. These activities are executed by integration of technology, organizational structures, and cognitive-based strategies to convey existing knowledge and produce new knowledge. The critical step is the enhancement of the cognitive system in acquiring, storing, and utilizing knowledge for learning, problem solving, and decision making.

Knowledge management is stated as the management of the organization (an individual student or a group of students in our context of learning environment) toward the continuous renewal of the organizational knowledge base, which may include the creation of supportive organizational structures, facilitation of organizational members, application of IT instruments, with an emphasis on teamwork and diffusion of knowledge (as in groupware) (Bertel). As such, knowledge management is a strategy that turns an organization’s intellectual assets (recorded information and the talents of its members) into greater productivity, new value, and increased competitiveness.

For a tutoring system, obviously we need a framework that can support knowledge management: a framework that offers a computational environment in which well-represented knowledge can serve as a communication medium between students and their activities. The indicated framework can consist of a shared knowledge representation and mechanisms for customized routing of knowledge to appropriate students (De Diana).

Models, methods, tools, and techniques for effective knowledge management become increasingly available, which is very important for education, because learning is a highly interactive process, and different kinds of knowledge are transferred among learners, tutoring systems, and human tutors.

An essential aspect of knowledge is that it is contextualized and dependent. This is the reason why knowledge is so difficult to acquire, represent, access, and transfer. Bruff and Williams pointed out that intelligent tutoring systems have to provide mechanisms to deal with the following interrelated knowledge- modeling problems (Bruff, 1999):

• Uncertainty of knowledge

• Conflicts among knowledge

• Dependency among knowledge

• The problem of knowledge granularity

• Incompleteness of knowledge, i.e., all relevant knowledge may not be known

• Fusion of knowledge, where knowledge is merged from different sources

• Revision of existing knowledge base when new knowledge is obtained (This new knowledge may be inconsistent with the existing knowledge base.)

Knowledge that is uncertain or incomplete may need to be revised and refined over time. Therefore, revision of a knowledge base is closely related to modeling both the uncertainty and the incompleteness of information. If readers are interested in these topics, more references can be found in Further Readings section at the end of this chapter.

Besides the problems of knowledge incompleteness, updating, conflicts, granularity, and uncertainty, one of the problems from knowledge modeling in an agent-based tutoring system is dealing with different kinds of knowledge. We discuss how to manage these different kinds of knowledge existing in a tutoring system in the following subsections, which are based mainly on the research by Morin (1998):

1. Domain knowledge (conceptual and procedural) (DK)

   Domain knowledge is the real knowledge we want to teach a student; it contains all conceptual and procedural aspects of the knowledge of one topic or area. Different topics or courses may have domain knowledge with different structures. Usually, domain knowledge may include concepts and relations among concepts, and often, these relations will organize concepts into a hierarchical structure, which will help the learning process greatly and provide a foundation for problem solving or inferential knowledge. For example, concepts can be basic entities like the binary tree or binary search tree. And, there is a “subclass of” relation between them.

2. Problem-solving knowledge (inferential) (PSK)

   Problem-solving knowledge is the knowledge that a student uses to learn domain knowledge. It is usually modeled and stored as procedures, and it contains inferential processes used to solve a problem using relation information from domain knowledge (Lelouche, 1997).

3. Tutoring knowledge (TK)

   Tutoring knowledge includes information about common student errors and misconceptions. Tutoring knowledge is the most important knowledge, because it is the key for us to use to build a customized learning system for each student. This customized learning system can deliver appropriate individualized instruction to help students learn more effectively and efficiently. This ability depends heavily on the availability and accuracy of the information held about the student in the student agent, which holds different types and levels of sophistication of the knowledge and also includes methods with which to elicit and incorporate the new information into the student model.

   Tutoring knowledge is usually session-based, because it varies from topic to topic, from student to student, and from time to time, even for the same student. Moreover, to make a learning process more interesting and efficient, a tutoring system should use a variety of stimuli, such as multimedia techniques, to present a topic in different ways, even to the same student, and to change the ways of presentations of the explanations or answers provided to the student.

---