NLP parser in Haskell?
dokondr
dokondr at gmail.com
Sat Oct 29 19:37:28 BST 2011
2011/10/29 Grzegorz Chrupała <pitekus at gmail.com>
...
> May I ask why you need your parser to be written in Haskell? Why not
> just use the Stanford parser to do your parsing, read the output into
> your Haskell program, and do whatever processing you need to do in
> Haskell?
>
I think the answer is obvious:
1) FFI with Java is more theoretical then practical thing.
2) It is much easier to work with syntax tree as Haskell data type, then to
parse it from text into this data type first. Also in case you have to
parse not a single, but thousands of sentences, then you have to code
special Haskell-call-Java-for-text-output synchronization logic. Or
Haskell-HTTP(REST) / JMS -Java Web_Service-Fun_Stuff.
Too much work to do a natural thing.
Of course I will have to go through all this fun staff once again. In fact
I already did this once - Web APP (Java GUI / GWT) + Haskell everything
else.
:(
>
> On Sat, Oct 29, 2011 at 17:10, dokondr <dokondr at gmail.com> wrote:
> > I asked this question several month ago, and ask again in hope for some
> new
> > development.
> > Ideally I would like to find Haskell libraries similar to:
> > The Stanford Statistical Parser
> > http://nlp.stanford.edu/software/lex-parser.shtml
> > or:
> > NLProcessor by Infogistics
> > http://www.infogistics.com/textanalysis.html
> >
> > Both of the above do: tokenization, capitalized word normalization,
> sentence
> > segmentation, part-of-speech tagging and syntactic chunking.
> >
> > Again ideally it would be possible to pass NLP parser a sentence and get
> > back an output, similar to what Stanford Parser provides. For example:
> > http://nlp.stanford.edu:8080/parser/
> >
> > Your query
> >
> > Colorless green ideas sleep furiously.
> >
> > Tagging
> >
> > Colorless/JJ
> > green/JJ
> > ideas/NNS
> > sleep/VBP
> > furiously/RB
> > ./.
> >
> > Parse
> >
> > (ROOT
> > (S
> > (NP (JJ Colorless) (JJ green) (NNS ideas))
> > (VP (VBP sleep)
> > (ADVP (RB furiously)))
> > (. .)))
> >
> > Typed dependencies
> >
> > amod(ideas-3, Colorless-1)
> > amod(ideas-3, green-2)
> > nsubj(sleep-4, ideas-3)
> > advmod(sleep-4, furiously-5)
> >
> > Typed dependencies, collapsed
> >
> > amod(ideas-3, Colorless-1)
> > amod(ideas-3, green-2)
> > nsubj(sleep-4, ideas-3)
> > advmod(sleep-4, furiously-5)
> >
> > Statistics
> >
> > Tokens: 6
> > Time: 0.050 s
> >
> >
> > _______________________________________________
> > NLP mailing list
> > NLP at projects.haskell.org
> > http://projects.haskell.org/cgi-bin/mailman/listinfo/nlp
> >
> >
>
--
All the best,
Dmitri O. Kondratiev
"This is what keeps me going: discovery"
dokondr at gmail.com
http://sites.google.com/site/dokondr/welcome
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://projects.haskell.org/pipermail/nlp/attachments/20111029/84caf690/attachment.htm>
More information about the NLP
mailing list