24 Mar 2005 21:15:20 -0500

Related articles |
---|

Finite State Automaon Question vborkyREMOVE_THIS@yahoo.com (Vinayak R. Borkar) (2005-03-15) |

Re: Finite State Automaon Question torbenm@diku.dk (2005-03-18) |

Re: Finite State Automaon Question vborky@yahoo.com (Vinayak R. Borkar) (2005-03-20) |

Re: Finite State Automaon Question torbenm@diku.dk (2005-03-24) |

Re: Finite State Automaon Question peter.ludemann@gmail.com (2005-03-24) |

Re: Finite State Automaon Question nathan.moore@sdc.cox.net (Nathan Moore) (2005-03-25) |

Re: Finite State Automaon Question vborky@yahoo.com (Vinayak R. Borkar) (2005-03-31) |

Re: Finite State Automaon Question torbenm@diku.dk (2005-04-02) |

From: | peter.ludemann@gmail.com |

Newsgroups: | comp.compilers |

Date: | 24 Mar 2005 21:15:20 -0500 |

Organization: | http://groups.google.com |

References: | 05-03-058 |

Keywords: | theory |

Posted-Date: | 24 Mar 2005 21:15:20 EST |

Take a look at the papers by Lauri Kartunnen et al (referenced from

http://www.xrce.xerox.com/competencies/content-analysis/fsCompiler/fsrefs.html)

... besides the papers, there's a suite of tools for transforming and

combining FSTs, but I don't know what's available outside Xerox

Research. The terminology is somewhat different from FSTs for

compilers, with the emphasis on building FSTs from components and

transforming them (e.g., using composition of FSTs: dictionary "ox",

transformed to "oxs" by plural rule, transformed to "oxen" by

special-case dictionary). Another author to look for: Koskenniemi.

[FSTs for Natural Language Processing tend to have 100s of thousands of

states, so optimization and combining techniques are very important.]

Post a followup to this message

Return to the
comp.compilers page.

Search the
comp.compilers archives again.