Some classes of regular languages identifiable in the limit from positive data

Abstract

Angluin defined in [Ang82] the classes of k-reversible regular languages and showed that these classes are identifiable in the limit from positive data. We introduced in [DLT01b] a class of automata based on properties of residual languages (RFSA) and showed in [DLT00] and [DLT01a] that this class could be interesting for grammatical inference purpose. Here, we study properties of 0-reversible languages that can be expressed as properties of their residual languages and that are useful for identification from positive data. This leads us to define classes of languages which strictly contain the class of 0-reversible languages and are identifiable in the limit from positive data.