Role of constrained computational systems in natural language processing

摘要

The use of constrained formal/computational systems just adequate for modeling various aspects of language—syntax, semantics, pragmatics and discourse, among others, has proved to be an effective research strategy leading to deep understanding of these aspects, with implications to both machine processing and human processing. This approach enables one to distinguish between the universal and stipulative constraints. This is in contrast to an approach where we start with the most powerful formal/computational system and then model the phenomena by making all constraints stipulative in a sense. The use of constrained systems for modeling leads to some novel ways of describing locality of structures and brings out the relationship between the complexity of description of primitives and local computations over them. These ideas serve to unify theoretical, computational and statistical aspects of natural languages processing in AI. It is expected that this approach will be productive in other domains of AI.