Re: [Corpora-List] TERM EXTRACTION TOOLS

From: Diana Maynard (d.maynard@dcs.shef.ac.uk)
Date: Fri Nov 12 2004 - 14:11:27 MET

  • Next message: Beek L.J.van der: "Re: [Corpora-List] fisher's exact test"

    Hi Lebron
    There's been a lot of work on evaluating different term extraction methods and
    tools in the past few years. Google should come up with a whole bunch of
    references for you. However, a lot of the tools are designed for very
    particular domains/applications, so evaluation can be tricky.

    The C/NC Value method (Frantzi and Ananiadou 99) is worth including in your
    evaluation as a good baseline. You'll probably have to implement the algorithm
    yourself though, unless you can get hold of one from somewhere (many people
    have used it in such evaluations, so it might be worth hunting around).
    Actually if anyone already has a Java implementation of it they'd be happy to
    share, I'd be interested to know.

    Frantzi, K.T. and S.Ananiadou (1999) The C-value/NC-value domain independent
    method for multi-word term extraction. Journal of Natural Language Processing,
    6(3) pp. 145-180

    Regards
    Diana

    > On Thursday, Nov 11, 2004, at 03:48 Europe/Rome, lebron letchev wrote:
    >
    >> Hi,
    >>
    >> I am looking for a good term extraction tools/methods.
    >>
    >> Does anyone know of good term extraction tools/methods? My interest
    >> is to compare some of the existing tools/methodologies to one another
    >> and to evaluate their performances on corpora.
    >>
    >> Thank you in advance
    >>
    >> Sincerely Yours
    >>
    >> Lebron Letchev
    >>
    >>
    >>



    This archive was generated by hypermail 2b29 : Fri Nov 12 2004 - 14:11:29 MET