Hi Jonathan,
1.
Like Vs Enjoy
We're just trying to create a hierarchy of features to include since this
is a big project, we'd like to know what are the things we'd implement
first, and then if time permits what are the features we'd add to the API
next.
2.
- Blackbox example
You have high level functions,
lttoolbox.make_transducer(input)
This kind of implementation probably best for users who haven't used C++
version of the code
- A customizable model with dictionary preprocessing would be
Intermediately close to c++ implementation, but with abstraction on
dictionary preprocessing.
Here one doesn't have control over the conversion of .dix files to any
other format. (can't work with letter transducers freely, no XML parsing
functions available to users etc, etc)
But fairly good control over other functionalities of lttoolbox like the
making changes in the current state of the transducer, having available the
class for the transducer for the users to modify,
- The third option is intuitive, and probably the most convenient for users
who have been using the C++ version the module.
The reason for such a question is the time to develop would be directly
proportional to the level of abstraction. And hence if I could determine
the most wanted trend, I could focus on other things in the module.
3.
I really haven't added that question and haven't dug deep into the Output
Formats, It was a question added by one of my mentors, I'll inquire more
about it and get back to you regarding this. However, If there are any
assumptions you are making (that would be your expectations from the
module) you could just make a short note in the question about any
additional feature that you specifically want.
Regards
Arghya Bhattacharya
On Mon, 30 Apr 2018 at 18:26 Jonathan Washington <
jonathan.n.washing...@gmail.com> wrote:
> Hi Arghya,
>
> Could you clarify a few things about the survey?
>
> - How should I understand the difference between "like" and "enjoy"?
>
> - Could you give a hypothetical coding example of each of the lttoolbox
> models?
>
> - Does "UD-compatible style" refer to CoNLL-U format, or does it have to
> do with tag names and the like? Is it safe to assume that CG format will
> also be supported?
>
> --
> Jonathan
>
>
> On Sun, Apr 29, 2018, 16:01 arghya bhattacharay <arghyati...@gmail.com>
> wrote:
>
>> Hello everyone,
>>
>> Thanks for all your inputs regarding the previous enquiry.
>>
>> Following that, we have designed a Form to materialize the survey and to
>> find out more details as to what is desired in the python API for Apertium.
>> I request all to kindly fill in the form (link below) so that we can
>> cater to the needs of the users.
>>
>> LINK: https://goo.gl/forms/k4pGJicXIgsVOMG83
>>
>> Regards
>> Arghya Bhattacharya
>>
>> On Wed, 18 Apr 2018 at 20:50 Tommi A Pirinen <
>> tommi.antero.piri...@uni-hamburg.de> wrote:
>>
>>> [Replies inline]
>>> On Wed, 18 Apr 2018 14:41:33 +0530
>>> arghya bhattacharay
>>> <arghyati...@gmail.com> wrote:
>>>
>>> > It would be really helpful to get to know what are the python NLP
>>> > libraries that people generally prefer to use so as to gain some
>>> > insight on what is desirable in the API being built and explore the
>>> > various implementation types that users find convenient.
>>> >
>>> > I would request one and all to kindly give their inputs on their
>>> > choice of python NLP libraries.
>>>
>>> One thing that I have to say I found particularly nice is the pyfst[1]
>>> for OpenFST, this is after having programmed with the OpenFST bridge to
>>> HFST in c++, it mostly just works and has a real pythonic feel to it.
>>> And the ipython notebook addition is very nice.
>>>
>>> As a short example I would probably want to use apertium python module
>>> e.g. like this:
>>>
>>> import apertium
>>>
>>> ape_fin = apertium.load("fin-morph")
>>> analyses = ape_fin.analyse("kissa")
>>> for analysis in analyses:
>>> print(analysis)
>>> ape_findeu = apertium.load("fin-deu-debug")
>>> translations = ape_findeu.translate("puutarhassa")
>>> for translation in translations:
>>> print(translation)
>>>
>>> Details about package/class/functions and names are unimportant to
>>> me, but it should be simple and not require messing about with
>>> irrelevant datatypes (headers/streams/alphabets/pipelines/...) and
>>> underlying complexities like many c++-to-python bindings tend to be.
>>>
>>> [1] <https://pyfst.github.io/>
>>> --
>>> Doktor Tommi A Pirinen, Computational Linguist,
>>> <https://flammie.github.io/purplemonkeydishwasher/>, Universität
>>> Hamburg, Hamburger Zentrum für Sprachkorpora <http://hzsk.de>. CLARIN-D
>>> Entwickler. President of ACL SIGUR SIG for Uralic languages
>>> <http://gtweb.uit.no/sigur/>.
>>> I tend to follow inline-posting style in desktop e-mail messages.
>>>
>>>
>>>
>>>
>>> ------------------------------------------------------------------------------
>>> Check out the vibrant tech community on one of the world's most
>>> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
>>> _______________________________________________
>>> Apertium-stuff mailing list
>>> Apertium-stuff@lists.sourceforge.net
>>> https://lists.sourceforge.net/lists/listinfo/apertium-stuff
>>>
>>
>> ------------------------------------------------------------------------------
>> Check out the vibrant tech community on one of the world's most
>> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
>> _______________________________________________
>> Apertium-stuff mailing list
>> Apertium-stuff@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/apertium-stuff
>>
>
> ------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> _______________________________________________
> Apertium-stuff mailing list
> Apertium-stuff@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/apertium-stuff
>
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Apertium-stuff mailing list
Apertium-stuff@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/apertium-stuff
