A SEMANTIC COMPONENT IN AUTOMATIC SENTENCE ANALYSIS.
Final scientific rept. 30 Dec 66-30 Dec 68,
CAMBRIDGE UNIV (ENGLAND) LITERARY AND LINGUISTIC COMPUTING CENTRE
Pagination or Media Count:
A procedure is described for automatically associating formulas exhibiting content structure to natural language English sentences. The input is that of an exiting automatic syntax analyzer. This study provides the autonomous semantic component of this analyzer. The formulas produced act as a semantic control and reduce multiple outputs from the syntax analyzer. They are also potentially relevent to problems involving automatic handling of data from natural language text, including the automatic recognition of logical relations between sentences. Results should be regarded as provisional until the procedure can be programmed and tested. These data are needed to enable establishment of all of the special semantic control routines expected to be needed. The postulation of this semantic component level of analysis shares some concepts in common with that of deep structure used in generative transformational grammar. Fundamental considerations about the relationship between descriptive grammars and interpretive algorithms are discussed briefly. An example is given of use of the formulas for disambiguation. Author