Re: Corpora: brill tagger

From: Klas Prutz (klas.prytz@ling.uu.se)
Date: Fri Mar 15 2002 - 21:12:55 MET

  • Next message: Christophe ASSELIN: "Corpora: Google currently offers many interface languages"

    ...and on the Brown corpus, I believe. I have a version trained on the
    written part of BNC Sampler as well.

    Regards

    Klas Prytz
    Institutionen för lingvistik
    Uppsala universitet

    On Thu, 14 Mar 2002, Adam [iso-8859-2] Przepiórkowski wrote:

    >
    > > I was wondering if anyone new what tagset Eric Brill's Transformation-based learning Tagger used, or whether it has it's own tagset.
    >
    > The tagger can be trained on any a corpus with any tagset, but the
    > pre-trained version was trained on WSJ, as far as I remember, and so
    > it assumes the UPenn Treebank tagset:
    >
    > http://www.comp.leeds.ac.uk/amalgam/tagsets/upenn.html
    >
    > --
    > Adam P.
    >
    >

    This archive was generated by hypermail 2b29 : Fri Mar 15 2002 - 21:28:09 MET