Adaptive Rückmeldungen im intelligenten Tutorensystem LARGO

Abstract

The Intelligent Tutoring System LARGO is designed to help law students learn argumentation skills. The approach implemented in LARGO uses transcripts of oral arguments as learning resources: Students annotate them and create graphical representations of the argument flow. The system encourages students to reflect upon arguments proposed by the attorneys and helps students detect possible weaknesses in their analysis of the dispute. Technically, graph grammar and collaborative filtering algorithms are employed to detect these weaknesses. This article describes how “usage contexts” are determined and used to create adaptive feedback in LARGO. On the basis of a controlled study with the system that took place with law students at the University of Pittsburgh, we discuss to what extent the automatically calculated usage contexts can predict student’s learning gains.