MathDox Adaptive: adding user diagnosis and course adaptivity to the MathDox system

Abstract

This thesis reports on the MathDox Adaptive System, a prototype intelligent tutor system based on existing MathDox software, designed to adaptively teach math to high school students. MathDox Adaptive maps the missing skill of each of its users by letting them take a diagnostic test. After the test, MathDox Adaptive makes an e ort to teach the users their missing skills by providing training courses. A training course contains training exercises along with hints and background theory. The courses also adapt the diffi culty of their exercises to the skill level of the user by actively measuring the progress of the user made during the training and increasing or decreasing the diffi culty accordingly. In order to obtain educational content for our prototype, we used knowledge acquisition techniques to interview three experts in the field of teaching mathematics. We used the results of these interviews to generate the content of the diagnostic test, training courses, hints and the background theory. After the successful completion of the prototype, it was evaluated on user satisfaction. A disappointing average satisfaction measurement of "slightly satisfied" was caused mainly due to the slow processing speed of the MathDox system. After the evaluation we discuss how we can solve the problem of the slow processing speed by changing the framework of the MathDox Adaptive system from a XML based framework to a PHP based framework.