You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@opennlp.apache.org by Tommaso Teofili <to...@gmail.com> on 2014/10/15 09:32:13 UTC
Parsing with PCFGs
Hi all,
in a bit of spare time I sketched a basic implementation of (in memory)
probabilistic context free grammars which, if properly trained, can be used
to build the parse tree of a given sentence, however (also looking at the
doc on the website) it's not completely clear what's already implemented in
trunk, I see there are 2 algorithms for parsing, could someone shed some
light on them? And eventually fire an opinion for adding PCFGs as an
additional algorithm?
Regards,
Tommaso
Re: Parsing with PCFGs
Posted by Tommaso Teofili <to...@gmail.com>.
Ok, no problem.
In the meantime I've added the first PCFG implementation in the sandbox,
see http://svn.apache.org/r1632735
Regards,
Tommaso
2014-10-16 11:33 GMT+02:00 Rodrigo Agerri <ra...@apache.org>:
> Hello!
>
> No, unfortunately not :)
>
> Cheers,
>
> Rodrigo
>
> On Thu, Oct 16, 2014 at 9:20 AM, Tommaso Teofili
> <to...@gmail.com> wrote:
> > Hi Rodrigo,
> >
> > thanks a lot for your inputs, do you have insights on the "treeinsert"
> > algorithm [1] too?
> >
> > Thanks,
> > Tommaso
> >
> > [1] :
> >
> http://opennlp.apache.org/documentation/1.5.3/manual/opennlp.html#tools.parser.parsing
> >
> > 2014-10-15 9:38 GMT+02:00 Rodrigo Agerri <ra...@apache.org>:
> >
> >> Hi,
> >>
> >> The main algorithm (called chunking in the trunk) is based on
> >> Ratnapharki's work.
> >> It is best to directly read the paper.
> >>
> >> http://link.springer.com/article/10.1023/A:1007502103375
> >>
> >> This is a shift-reduced parser which incidentally are becoming quite
> >> fashionable again. For example, Stanford CoreNLP recently released a
> >> shift-reduced parser themselves, as an alternative to their PCFGs,
> >> lexicalized parser.
> >>
> >> HTH,
> >>
> >> Rodrigo
> >>
> >> On Wed, Oct 15, 2014 at 9:32 AM, Tommaso Teofili
> >> <to...@gmail.com> wrote:
> >> > Hi all,
> >> >
> >> > in a bit of spare time I sketched a basic implementation of (in
> memory)
> >> > probabilistic context free grammars which, if properly trained, can be
> >> used
> >> > to build the parse tree of a given sentence, however (also looking at
> the
> >> > doc on the website) it's not completely clear what's already
> implemented
> >> in
> >> > trunk, I see there are 2 algorithms for parsing, could someone shed
> some
> >> > light on them? And eventually fire an opinion for adding PCFGs as an
> >> > additional algorithm?
> >> >
> >> > Regards,
> >> > Tommaso
> >>
>
Re: Parsing with PCFGs
Posted by Rodrigo Agerri <ra...@apache.org>.
Hello!
No, unfortunately not :)
Cheers,
Rodrigo
On Thu, Oct 16, 2014 at 9:20 AM, Tommaso Teofili
<to...@gmail.com> wrote:
> Hi Rodrigo,
>
> thanks a lot for your inputs, do you have insights on the "treeinsert"
> algorithm [1] too?
>
> Thanks,
> Tommaso
>
> [1] :
> http://opennlp.apache.org/documentation/1.5.3/manual/opennlp.html#tools.parser.parsing
>
> 2014-10-15 9:38 GMT+02:00 Rodrigo Agerri <ra...@apache.org>:
>
>> Hi,
>>
>> The main algorithm (called chunking in the trunk) is based on
>> Ratnapharki's work.
>> It is best to directly read the paper.
>>
>> http://link.springer.com/article/10.1023/A:1007502103375
>>
>> This is a shift-reduced parser which incidentally are becoming quite
>> fashionable again. For example, Stanford CoreNLP recently released a
>> shift-reduced parser themselves, as an alternative to their PCFGs,
>> lexicalized parser.
>>
>> HTH,
>>
>> Rodrigo
>>
>> On Wed, Oct 15, 2014 at 9:32 AM, Tommaso Teofili
>> <to...@gmail.com> wrote:
>> > Hi all,
>> >
>> > in a bit of spare time I sketched a basic implementation of (in memory)
>> > probabilistic context free grammars which, if properly trained, can be
>> used
>> > to build the parse tree of a given sentence, however (also looking at the
>> > doc on the website) it's not completely clear what's already implemented
>> in
>> > trunk, I see there are 2 algorithms for parsing, could someone shed some
>> > light on them? And eventually fire an opinion for adding PCFGs as an
>> > additional algorithm?
>> >
>> > Regards,
>> > Tommaso
>>
Re: Parsing with PCFGs
Posted by Tommaso Teofili <to...@gmail.com>.
Hi Rodrigo,
thanks a lot for your inputs, do you have insights on the "treeinsert"
algorithm [1] too?
Thanks,
Tommaso
[1] :
http://opennlp.apache.org/documentation/1.5.3/manual/opennlp.html#tools.parser.parsing
2014-10-15 9:38 GMT+02:00 Rodrigo Agerri <ra...@apache.org>:
> Hi,
>
> The main algorithm (called chunking in the trunk) is based on
> Ratnapharki's work.
> It is best to directly read the paper.
>
> http://link.springer.com/article/10.1023/A:1007502103375
>
> This is a shift-reduced parser which incidentally are becoming quite
> fashionable again. For example, Stanford CoreNLP recently released a
> shift-reduced parser themselves, as an alternative to their PCFGs,
> lexicalized parser.
>
> HTH,
>
> Rodrigo
>
> On Wed, Oct 15, 2014 at 9:32 AM, Tommaso Teofili
> <to...@gmail.com> wrote:
> > Hi all,
> >
> > in a bit of spare time I sketched a basic implementation of (in memory)
> > probabilistic context free grammars which, if properly trained, can be
> used
> > to build the parse tree of a given sentence, however (also looking at the
> > doc on the website) it's not completely clear what's already implemented
> in
> > trunk, I see there are 2 algorithms for parsing, could someone shed some
> > light on them? And eventually fire an opinion for adding PCFGs as an
> > additional algorithm?
> >
> > Regards,
> > Tommaso
>
Re: Parsing with PCFGs
Posted by Rodrigo Agerri <ra...@apache.org>.
Hi,
The main algorithm (called chunking in the trunk) is based on
Ratnapharki's work.
It is best to directly read the paper.
http://link.springer.com/article/10.1023/A:1007502103375
This is a shift-reduced parser which incidentally are becoming quite
fashionable again. For example, Stanford CoreNLP recently released a
shift-reduced parser themselves, as an alternative to their PCFGs,
lexicalized parser.
HTH,
Rodrigo
On Wed, Oct 15, 2014 at 9:32 AM, Tommaso Teofili
<to...@gmail.com> wrote:
> Hi all,
>
> in a bit of spare time I sketched a basic implementation of (in memory)
> probabilistic context free grammars which, if properly trained, can be used
> to build the parse tree of a given sentence, however (also looking at the
> doc on the website) it's not completely clear what's already implemented in
> trunk, I see there are 2 algorithms for parsing, could someone shed some
> light on them? And eventually fire an opinion for adding PCFGs as an
> additional algorithm?
>
> Regards,
> Tommaso