Misconception elicitation from the logs of an educational system

Abstract

Large volumes of educational data are being collected by online learning environments. This data can be used to track progress of learners and adjust their learning experience in an individually-optimised way. One important challenge in this regard is to effectively recognise typical misconceptions responsible for frequent patterns of erroneous learning behaviour. The purpose of this thesis is to develop a method for identifying such misconceptions in the stream of educational activity. The developed method is evaluated in a data-mining experiment conducted over a large dataset of learners’ responses to arithmetic exercises.