Exercise generation on language specification

Abstract

Exercise generation on language specification is a challenging problem, because of the richness of the objects in the domain. In this paper we discuss Mgbeg (Meta-Grammar-Based Exercise Generator) a toolkit for exercise generation on context-free languages. Mgbeg approach is based on a meta-grammar formalism and tool, used to define a set of similar exercises. Mgbeg is simple attributed grammar used to describe the set of valid exercise (and randomly generate one of them). Each exercise typically contains several attributes calculated during the generation steps: namely, one or more formal specification of the language (context free grammar); the exercise statement; other information such as examples, common mistakes, validation data, to be used in the construction of the exercise statement, solution, and assessment steps. Complementary the toolkit provides a grammar module, with functionality for grammar comparison, sentence generation and recognition; a template engine (to help in textual attributes calculation).