Improving the Efficiency of a Natural Language Processor

摘要

NLP is a processor being developed at the Naval Postgraduate School for research in natural language man-machine communication. With this system text can be translated into an entity-attribute-value information structure, and such a structure can be translated into text. These two processes, called decoding and encoding, respectively, are specified by writing 'rules' in a language designed for this system. The thesis reports on a scheme for storing these rules in the computer in a compact fashion, and describes the related routines. The savings in core storage and CPU time achieved by using this scheme are given for a particular application of NLP. (Author)