Regular grammatical inference: A genetic algorithm approach

Abstract

Grammatical inference is the problem of inferring a grammar, given a set of positive samples which the inferred grammar should accept and a set of negative samples which the grammar should not accept. Here we apply genetic algorithm for inferring regular languages. The genetic search is started from maximal canonical automaton built from structurally complete sample. In view of limiting the increasing complexity as the sample size grows, we have edited structurally complete sample. We have tested our algorithm for 16 languages and have compared our results with previous works of regular grammatical inference using genetic algorithm. The results obtained confirm the feasibility of applying genetic algorithm for regular grammatical inference.