Reducing student hint use by creating buggy messages from machine learned incorrect processes

Abstract

The goal of this research is to improve existing forms of help in tutoring systems by using "Buggy" messages, which are a simple text message specific to the incorrect answer. Buggy messages are created from machine learned incorrect processes based on the student's incorrect answer. A randomized control trial is run in ASSISTments to determine if the buggy messages were effective. © 2014 Springer International Publishing Switzerland.