Learning regular expressions from representative examples and membership queries

Abstract

A learning algorithm is developed for a class of regular expressions equivalent to the class of all unionless unambiguous regular expressions of loop depth 2. The learner uses one representative example of the target language (where every occurrence of every loop in the target expression is unfolded at least twice) and a number of membership queries. The algorithm works in time polynomial in the length of the input example. © 2010 Springer-Verlag.