A PROBABILISTIC METHOD FOR PHRASE DETERMINATION

Abstract

The model is stochastic but does not require any Markov assumption, although an implication of that approach is discussed. It is assumed only that a sequence of parts of speech of the words of a large sample text or corpus is the realization of some stationary stochastic process. Under this assumption, with the system of parts of speech suitably defined, it is shown that phrase and sentence boundaries tend to occur where prediction of future grammatical forms, given immediate forms, is most difficult. (Author)

Document Details

Document Type
Technical Report
Publication Date
Mar 01, 1962
Accession Number
AD0275465

Entities

People

  • Edward Gammon

Organizations

  • Lockheed Martin Missiles and Space

Tags

DTIC Thesaurus Topics

  • Boundaries
  • Data Science
  • Information Science
  • Mathematics
  • Probability
  • Sequences
  • Stationary
  • Stationary Processes
  • Stochastic Processes

Readers

  • Computational Linguistics
  • Mathematical Modeling and Probability Theory.
  • Theoretical Analysis.