[Corpora-List] Trainable constituency parsers

[Yannick Versley]

One that you should try is the BerkeleyParser:
http://code.google.com/p/berkeleyparser/

For a variety of languages, the latent-variable model works better than
head-lexicalized PCFGs.
It doesn't require a head table like Bikel's parser, but train/test split
is (or was, when I looked) hard-coded, so you still need to go in and tweak
some bits.

Best,
Yannick

On Tue, Sep 8, 2009 at 10:45 AM, Sergio Castro <sergio.castro at di.fc.ul.pt>wrote:

> Dear corpora members,
>
> I'm looking for tools and trainable constituency parsers for
> PCFG's and HPCFG's, preferably ones that support POS input.
> I've used Dan Bikel's parser, but was looking for alternatives,
> any help would be great.
>
> Thank you for your time,
> Sérgio Castro
>
>
> _______________________________________________
> Corpora mailing list
> Corpora at uib.no
> http://mailman.uib.no/listinfo/corpora
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/corpora/attachments/20090908/1f63294f/attachment.htm>
-------------- next part --------------
_______________________________________________
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora