[developers] QUERY REGARDING GRAMMAR COMPILATION

Woodley Packard sweaglesw at sweaglesw.org
Fri Oct 28 16:06:16 CEST 2016


Naman,

Please keep the delphin developers list on the CC for this thread.  Unfortunately I can't help with DMRS questions or details of the German grammar, but someone on that list probably can.

Regards, Woodley



> On Oct 28, 2016, at 12:52 AM, Naman Deep Singh <nmndeep at gmail.com> wrote:
> 
> Sir, 
> I have compiled the german grammar from the source told by Mr. Michael.
> But in the ACE generated MRS of any german sentence there seems to be an additional '--' preffixed to few attributes as PSV, ZU which is leading to error generation on attempt to generate DMRS for the same.
> 
> Also, On removing this '--' the DMRS is generated fine.
> Please suggest me a remedy for this issue.It would be highly appreciated.
> With Regards
> NAMAN DEEP SINGH 
> 
> 
> 
> 
> 
>> On Mon, Oct 24, 2016 at 11:28 AM, Naman Deep Singh <nmndeep at gmail.com> wrote:
>> Sir,
>> Regarding Hindi,
>> We create our own DMRS for Hindi which are then mapped to ACE generated DMRS.
>> As Hindi and other Indian languages are less complex than English the user inputted data is sufficient enough for us to generate all the relations in the input.Hindi has less of implicit relationships as that of English.
>> 
>> With Regards
>> NAMAN DEEP SINGH
>>  
>> 
>>> On Mon, Oct 24, 2016 at 5:40 AM, Woodley Packard <sweaglesw at sweaglesw.org> wrote:
>>> Hi Naman,
>>> 
>>> There is a Spanish grammar (SRG), but its support for generation using ACE is limitted at the moment.  I’m not aware of a large scale HPSG Russian grammar that is compatible with ACE, but maybe someone else on the DELPH-IN developers list (CC’d) is.
>>> 
>>> I’m also curious what grammar you are using to parse Hindi?
>>> 
>>> Regards,
>>> Woodley
>>> 
>>>> On Oct 23, 2016, at 10:32 PM, Naman Deep Singh <nmndeep at gmail.com> wrote:
>>>> 
>>>> Thank you for your quick and precise response sir.
>>>> Here we are trying to generate Hindi sentences into multiple languages and in the above mentioned link Spanish,Russian grammars for ACE were not found.
>>>> If you could help me again in the same I would be highly obliged.
>>>> With Regards
>>>> NAMAN DEEP SINGH
>>>> 
>>>>> On Sat, Oct 22, 2016 at 6:00 AM, Woodley Packard <sweaglesw at sweaglesw.org> wrote:
>>>>> Hi Naman,
>>>>> 
>>>>> The only thing I would add to what Mike already wrote is that (I believe) you should to compile GG using ace/config.parse.tdl for parsing and ace/config.generate.tdl for generating, to take maximum advantage of statistical models.
>>>>> 
>>>>> Regards,
>>>>> Woodley Packard
>>>>> 
>>>>>> On Oct 21, 2016, at 10:54 PM, Michael Wayne Goodman <goodmami at uw.edu> wrote:
>>>>>> 
>>>>>> Hello Naman,
>>>>>> 
>>>>>> I'm adding the DELPH-IN developers list [1] since I'm not a developer for ACE nor GG, and others might have more information for you. I'd be happy to answer pyDelphin questions, however, and for that I encourage you to either file an issue [2] or ask on the developers list.
>>>>>> 
>>>>>> But I think I can solve your immediate problem.
>>>>>> 
>>>>>> In order to get GG's DAT file for ACE to use, you'll need to compile it (see [3]). The repository you are using appears outdated, as it doesn't have the relevant files for ACE. Try the one in the LOGON repository, instead: http://svn.emmtee.net/trunk/dfki/gg
>>>>>> 
>>>>>> Navigate to the gg directory and run the following:
>>>>>> 
>>>>>>     ace -g ace/config.tdl -G gg.dat
>>>>>> 
>>>>>> When the command terminates, you should be able to use the gg.dat file with ACE:
>>>>>> 
>>>>>>     echo "Das Haus ist kelin." | ace -g gg.dat
>>>>>> 
>>>>>> I hope that helps you get started.
>>>>>> 
>>>>>> [1] http://lists.delph-in.net/mailman/listinfo/developers
>>>>>> [2] https://github.com/delph-in/pydelphin/issues
>>>>>> [3] http://sweaglesw.org/linguistics/ace/download/ace-0.9/doc/options.wiki
>>>>>> 
>>>>>> 
>>>>>> On Fri, Oct 21, 2016 at 10:09 PM Naman Deep Singh <nmndeep at gmail.com> wrote:
>>>>>> SIR, 
>>>>>> I am  member of a team currently working on Machine translation and we are using ACE and py-Delphin.
>>>>>> 
>>>>>> I am trying to implement a German grammar with ACE but it's file composition is not same are ERG or JACY, this grammar is available at http://gg.opendfki.de/ on unpacking the archive a lot of .tdl files come up but as in ERG not .dat file comes up.
>>>>>> Would you please look into my problem and help me to resolve the issue.
>>>>>> 
>>>>>> With Regards
>>>>>> NAMAN DEEP SINGH
> 
> <out2.mrs>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.delph-in.net/archives/developers/attachments/20161028/74c735d9/attachment.html>


More information about the developers mailing list