Automatically creating personalised exercises based on student profiles

Course Marker (CM - previously called Course Master), is a CAA system that marks programming code and diagrams and is a reengineered version of the Ceilidh system. It has been used in many higher institutions around the world, including at its developers' base, the School of Computer Science & IT, at the University of Nottingham. CM allows teachers to set exercises, and students to submit their solutions to these exercises. Upon each submission, students obtain immediate marks and detailed feedback from the system [4].The system has been well received and is excellent at marking exercises. However it can be extended in a variety of ways. This paper describes such an enhancement based on customising the questions to individual students by generating a personalised set of questions. Currently, all students are given the same set of exercises.To improve the learning environment, the idea is to use tailored multi-modal questionnaires. Firstly new types of question have been added to CourseMarker functionality. Initially these are the fixed response type of question such as multiple choice and multiple answer. Later more sophisticated types based on the full power of CourseMarker will be added. These are combined with currently available programming exercises and diagram based questions into a multi-modal questionnaire consisting of multiple types of question.Initially the questions in a given questionnaire are fixed by the teacher and pre-selected from a question bank. The next step is to allow the system to choose a set of questions dynamically from the question bank so as to give each student a different questionnaire.A major enhancement to this, and the main thrust of the research, is to tailor each questionnaire individually to each student at an appropriate level of difficulty for that students' ability. That is we will generate personalised exercises for each student, according to some individual student profile.Metadata for the students' profile, such as their background, their current level of knowledge on pre-requisite and current topics and subjects, and their preferred or best learning modes, are to be stored in a user model [2]. All this information can be obtained by interacting with applications such as Knowledge Tree [3] and WHURLE [1] and by using the outcome of earlier questionnaires.This poster presentation will explain the work involved to incorporate these enhancements.