Automatic Generation of Exercises for Selftesting in Adaptive E-Learning Systems.pdf

Automatic Generation of Exercises for Self-testing in Adaptive E-Learning Systems: Exercises on AC Circuits Paul CRISTEA, Rodica TUDUCE University “POLITEHNICA” of Bucharest, Faculty of Electrical Engineering Spl. Independentei 313, Bucharest, Romania, pcristea@dsp.pub.ro Abstract. The paper presents aspects of knowledge assessment in the framework of adaptive e-learning systems, focussing on aspect related to self-testing, as a way of providing the input necessary for both system adaptation and user informed decisions. A model of up-ward propagation of evaluation credits and down-ward propagation of validation and acceptance is presented. A prototype example of automatic test generation is described in some details for the special case of electric ac circuits analysis. Introduction The spreading of e-learning tools usage is steadily advancing, but it is still lagging far behind expectations. E-learning does not seem to fulfil its promise to become the most important learning methodology, especially in the context of the increased role of continuous and life-long learning. This is often explained by the fact that, despite their recent impressive developments, most of the currently available e-learning tools and environments are still less appealing than the traditional face-to-face teaching methods for both students and tutors [1-2]. From the student’s point of view, there are two quite opposite main reproaches: either the complaint about the “lack of human touch”, the rigidity of most e-learning tools resulting in the same web pages presented to any user intending to acquire a certain item of knowledge, or the protest against the “over- coaching”, the system behaving as knowing better than the user what are his/her needs. The problem of what should be part of a “learner’s profile”, what and how much of the learner’s specific objectives, interests, preferences, and even current state of mind should be tracked as part of the c