Personalised Intelligent Tutoring for Digital Libraries

Computer-based training is a fast-growing multi-billion dollar industry. The possibilities for systems that o er personalised training or tutoring, that dynamically adapt to the training needs of individual students, are immense. This not only means the personalisation of training content but perhaps even the personalisation of exams and student evaluations. In this paper we focus on ways of personalising multiple-choice exams and describe a technique for predicting the exam answers for individual students based on their previous exam history. We describe an evaluation of a collaborative ltering prediction system and demonstrate that accurate predictions can be achieved with limited pro ling information.