Exercise Generation by Group Models for Autonomous Web-Based Learning

Creating exercises for learners requires significant time. This is one reason, beside difficulties of discussing individualized tasks in a classroom setting, why often only few exercises are created and posed to all learners alike. In Web-based autonomous learning the setting problem is removed, while simultaneously demand for individualized exercises, comparable to adaptive learning material, increases.A model is proposed for assembling exercises from independent elements and creating an exemplary solution alongside. Here the cold-start problem is especially problematic, as opposed to learning material no sensible default-view exists. This difficulty can be reduced by integrating a group model, i.e. the history of results or actions of other learners.This paper presents an implementation of the model for generating cases for learning in the legal area. The Web-based user interface of such an online system is very important to render comparing the learners' solution with the exemplary one simple, allowing correction by peers as well. The current implementation state as well as planned extensions is described.