EFFICIENT LEARNING OF CONTEXT-FREE GRAMMARS FROM POSITIVE STRUCTURAL EXAMPLES

In this paper, we introduce a new normal form for context-free grammars, called reversible context-free grammars, for the problem of learning context-free grammars from positive-only examples. A context-free grammar G = (N, Σ, P, S) is said to be reversible if (1) A → α and B → α in P implies A = B and (2) A → αBβ and A → αCβ in P implies B = C. We show that the class of reversible context-free grammars can be identified in the limit from positive samples of structural descriptions and there exists an efficient algorithm to identify them from positive samples of structural descriptions, where a structural description of a context-free grammar is an unlabelled derivation tree of the grammar. This implies that if positive structural examples of a reversible context-free grammar for the target language are available to the learning algorithm, the full class of context-free languages can be learned efficiently from positive samples. © 1992.