Paper: Finite State Transducers Approximating Hidden Markov Models

ACL ID P97-1059
Title Finite State Transducers Approximating Hidden Markov Models
Venue Annual Meeting of the Association of Computational Linguistics
Session Main Conference
Year 1997
Authors
  • Andre Kempe (Xerox Research Centre Europe, Grenoble France)

This paper describes the conversion of a Hidden Markov Model into a sequential transducer that closely approximates the behavior of the stochastic model. This transformation is especially advantageous for part-of-speech tagging because the re- sulting transducer can be composed with other transducers that encode correction rules for the most frequent tagging errors. The speed of tagging is also improved. The described methods have been implemented and successfully tested on six languages.