Re: Corpora: Tranformation-based learners?

C Hogan (compling@juno.com)
Mon, 25 Jan 1999 13:12:22 -0500

>I'm currently working on an implementation of transformation-based
>learning
>('Brill-learning') which I hope will be both a bit faster and a more
>flexible
>than Brill's own distribution (reachable from Eric Brill's homepage
>http://www.cs.jhu.edu/~brill/). I intend to release it to the corpus
>linguistics/NLP community when it is finished.
>
>My question is simply if there are other alternative
>TBL-implementations around,
>and if anyone has performed bench-marks etc. on such systems.

There is an implementation of EDTBL as Finite-State Transducers:

Emmanuel Roche and Yves Schabes. 1995.
Deterministic Part-of-Speech Tagging with Finite-State Transducers.
Computational Linguistics. 21(2) pp. 227--253.

They claim 10 times speed-up over other taggers.

--chris

christopher hogan
computational linguistics
pittsburgh, pa

___________________________________________________________________
You don't need to buy Internet access to use free Internet e-mail.
Get completely free e-mail from Juno at http://www.juno.com/getjuno.html
or call Juno at (800) 654-JUNO [654-5866]