Dear Colleagues,
Thanks for people who respond. It varies between
training Brill tagger and the follow E-mail.
I am resendg it as per request of many persons.
***** It seems that I have to start preparing manully tagged set for
Arabic first which what I intend to do. So, Here comes
my next question: I sthere any annotation tools that can help
in tagiing Arabic manually.
Thanks again,
MOhamed F. Noamany
On Tue, 17 Jul 2001, Oliver Mason wrote:
> Dear Mohamed,
>
> I have a language-independent tagger, QTag, which can be trained using a
> pre-tagged sample text as input. It is implemented in Java, so it should
> handle Arabic texts alright, though I have never tested it. However, I'm
> happy to assist you in adapting the tagger to work with Arabic!
>
> What you would need to have is either a (machine-readable) lexicon, or a
> tagged sample text. This can be used to create a resource file for the tagger,
> which you can then use to tag other (larger) texts with. If you then correct
> any errors in the tagging you can repeat the process with the new (larger)
> training set, and you will then end up with fewer errors.
>
> In an evaluation with Romanian we used a few 10,000 words as training data
> and got a rate of about 98+% correct tag assignments.
>
> Regards,
> Oliver Mason
>
> --
> //\\ lecturer | centre for corpus linguistics | dept. of english | school of
> //\\ humanities | the university of birmingham | edgbaston | birmingham b15
> \\// 2tt | united kingdom | phone +44(0)121-414-6206 | fax +44(0)121-414- /\
> \\// 5668 | web http://www.clg.bham.ac.uk | email o.mason@bham.ac.uk \/
>
This archive was generated by hypermail 2b29 : Wed Jul 18 2001 - 15:38:59 MET DST