A METHODOLOGY FOR DESIGNING AUTOMATIC EVALUATORS USING GLMP PARADIGM

Abstract

Formative assessment improves human learning processes as it provides learners with information about what they need to work on. To automate formative assessment, we built a general model (GLMP) that allows the design of systems aimed at reproducing instructor's reasoning for learning assessment and generate a natural language assessment report. This paper presents a methodology for designing these automatic evaluators, highlighting the main points to be taken into consideration by the designer.