Knowledge-Based Adaptive Assessment in a Web-Based Intelligent Educational System Ioannis Hatzilygeroudis, Constantinos Koutsojannis, Constantinos Papavlasopoulos Department of Computer Engineering & Informatics, School of Engineering University of Patras, Greece {ihatz, ckoutsog, papavlas}@ceid.upatras.gr Jim Prentzas Department of Informatics and Computer Technology Technological Educational Institute of Lamia, Greece dprentzas@teilam.gr Abstract In this paper, we present an adaptive and intelligent web- based educational system that uses AI techniques for personalized assessment of the learners. More specifically, we focus on a mechanism for on-line creation of a user- adapted test, which can be used alongside the predetermined test. The user can ask for such a test any time he/she is willing to do so, even if he/she has not studied all predetermined concepts of a learning goal. A small rule base is used by an expert system inference engine for making decisions on the difficulty level of the exercises to be included in the test. This is based on the evaluation of the learner during concept studying. Adaptive assessment of the learner can be repeatedly used until there is no further need. Another small rule-base is used for deciding on whether a new test is suggested or not. This is based on the learner’s previous test assessment results. Preliminary experimental results show that the users need less time to study a learning goal when using the adaptive assessment capability of the system. Keywords Intelligent Web-Based Education, Intelligent E-Learning, Adaptive testing, Expert systems, Personalized assessment. 1. Introduction Intelligent Tutoring Systems (ITSs) are systems used for personalized learning [1]. They were usually developed as stand-alone systems. However, the emergence of the WWW gave rise to a number of Web-based ITSs, a type of Web-Based Intelligent Educational Systems (WBIESs) [2]. Adaptive Educational Hypermedia Systems (AEHSs) are also systems that offer personalized education. They are specifically developed for hypertext environments such as the WWW [3]. Enhancing AEHSs with aspects and techniques from ITSs creates a type of Adaptive and Intelligent Educational Systems (AIESs) [2]. Student evaluation or assessment is a basic issue in AIESs. One of the standard ways of achieving that is through testing. Adaptive testing is an interesting and relatively new direction of research [4, 5]. By ‘adaptive testing’, adaptation of a test individually to each student is meant. There have been a number of efforts to this direction. However, most of them make use of the so- called ‘item response theory’ (IRT), a mathematical model and process to create adaptive tests. CAT (Computerized Adaptive Testing) is a well-founded technique that uses IRT [6]. According to CAT, the response to a question determines the next question to be delivered to the learner. In this way, one concept at a time can be examined. On the other hand, this is not always the case. Usually, there is need to examine a number of concepts simultaneously. Furthermore, in most existing student diagnosis models [7, 8], tests refer to a predefined number of concepts; the learner cannot take a part of it, concerning a part of the associated concepts. So, an aspect that has not been paid attention to is the on-line creation of a test that refers to the concepts that the learner has dealt with up to that moment. We constructed an Artificial Intelligence Teaching System (AITS) to assist learning and teaching in the context of the “Artificial Intelligence” course in our Department. AITS is an adaptive and intelligent system. It adapts the course material to the student’s needs as much as possible, based on his/her profile and knowledge level [8, 9]. Knowledge level is evaluated after each concept via an expert system, which takes into account the difficulty level of questions/exercises. Additionally the system provides means to the tutor for constructing questions and tests in a structured way [8]. In this paper, we present a new capability of the system. With the help of a second expert system, AITS on- Proceedings of the Sixth International Conference on Advanced Learning Technologies (ICALT'06) 0-7695-2632-2/06 $20.00 © 2006 IEEE