Semi-automatic generation of competency maps based on educational data mining

Abstract

We propose a semi-automatic method for the generation of educational-competency maps from repositories of multiple-choice question responses, using Bayesian structural learning and data-mining techniques. We tested our method on a large repository of responses to multiple-choice exam questions from an undergraduate course in Languages and Automata Theory at Spain’s national distance-learning university (UNED). We also draw up guidelines and best practices, with a view to defining an educational data-mining methodology and to contributing to the development of educational data-mining tools.