Learnable classes of categorial grammars
Learnable classes of categorial grammars
Horn extended feature structures: fast unification with negation and limited disjunction
EACL '91 Proceedings of the fifth conference on European chapter of the Association for Computational Linguistics
Polyadic Dynamic Logics for HPSG Parsing
Journal of Logic, Language and Information
| Hi-index | 0.00 |
A theory of natural language can be evaluated on both extensional and intensional grounds. Systematic investigations of the extension of a theory may, for instance, lead to studies of the invariance properties of such theories. The intentional parameters that I wish to address include complexity, learnability, and monotonicity. The main results, on which my thesis builds, up to this point, include: (i) the universal recognition problem of model-theoretic feature-based grammar formalisms is complete for non-deterministic polynomial time, since such formalisms have the polysize model property, (ii) this result holds also for linearization-based extensions, (iii) the universal recognition problem of strongly monotonic, hybrid feature-based grammar formalisms is decidable in deterministic polynomial time, and (iv) there exists a strongly monotonic unification categorial grammar that is learnable in the limit from positive data. In addition, invariance studies have lead to the identification of a class of modal languages that define common feature-based grammar formalisms. The objective of my studies is to identify a tractable and learnable feature-based formalism.