Please log your contribution to the panel discussion

[Olga Caprotti]

Hi,

for future reference it would be good to keep a trace of what was discussed during our panel session,
while it is fresh in our minds.


Please enter your position statement or your remark if you made one during the panel discussion.

https://sites.google.com/site/gfschool2011/about-us/news/paneldiscussiononthefutureofgf

I think comments are enabled on the news item I just entered above (if not shout :) ).

Thanks!
olga

[Olga Caprotti]

Hi again.

Rogan just spotted a bug of the sites.google page settings for the Announcements template in that it does not
allow comments (to anonymous) even when the option is checked.
Please enter your discussion log here below.

Thank you.
--olga

[John J. Camilleri]

What I said:

I believe there is need for better development tools to aid grammar development, potentially including:

1. General test suite tools for any grammar
2. Test suites for the RGL
3. Scripts for creating empty RGs
4. Syntax highlighting, code folding, etc
5. Inline documentation for RGL, code completion, jump to function definitions, etc
6. Refactoring tools

The two main approaches would be writing a plugin for some popular desktop IDE such as Eclipse, or work towards building these features into a web/cloud-based GF IDE. Both approaches of course have pros and cons.

I myself am willing to devote some time towards developing such tools.

John

[Rogan Creswick]

On Thu, Aug 25, 2011 at 5:41 AM, John J. Camilleri
<jo...@johnjcamilleri.com> wrote:
> The two main approaches would be writing a plugin for some popular desktop
> IDE such as Eclipse,

Here is a link to the project I mentioned for (relatively) quickly
adding language support to Eclipse:

http://www.eclipse.org/Xtext/

--Rogan

[Rogan Creswick]

I believe my comments can be summarize in four areas of improvement
that would bring the most gains to commercial application grammar
developers and users:

1. Increase the robustness to non-conforming input.

My experience has been that GF applications appear to work "magically"
when I, or another developer, are providing example inputs for
demonstration. However, we have never seen a user succeed without
substantial guidance when they need to provide natural language input.
Guided editors are one solution, but that does not fit well in all
use cases (for example, if the text is already written, such as is the
case with the Patent work Adam et. al are doing).

2. Anticipating run-time parse failures or translation failures is
very difficult.

I believe this arises even more frequently when developing a RGL; it
would be fantastic to have some measure of the progress towards a
complete mapping to an abstract syntax, and/or estimates of language
coverage. I believe this is both a relatively simple interface issue
(in the case of an application grammar with an abstract syntax) as
well as a new research area (in the case of guaging language coverage
of a resource grammar.)

3. Run-time adaptation for new syntax / structures.

This is closely related to the issues we have with robust input.
Providing users with the ability to augment an existing (compiled)
grammar with new terms and--potentially--new grammatical structures
would help enormously when deploying GF-based applications to
end-users who are neither linguists nor programmers.

4. Tool support / tutorials / templates

Many others have also suggested these areas, so I won't belabor the
points other than to mention that I believe language tools are very
closely tied to the need for an introductory GF book ("GF for
dummies"). It seems unwise to develop both things (the next book and
the tools/templates) separately.

Thanks for organizing the panel! I found this to be very interesting
and enlightening!

--Rogan

[Normunds Grūzītis]

In addition to the great thoughts that were risen during the panel
discussion, I would like to rise attention to one more aspect of GF
that has been theoretically developed by Aarne (1994?), but hasn't
been implemented yet. Its about handling anaphoric references.

I suppose it would be a huge work to implement support for this in GF,
but I think in the long term it is a fundamental feature that would be
crucial for certain classes of application grammars.

Normunds

2011/8/25 Olga Caprotti <olga.c...@gmail.com>:

[Jordi]

• Better documentation:
• GF for dummies
  
• Linguists: Functional programming, Context-free grammars
• Domain experts: Functional programming, Context-free grammars, Linguistic notions
• Helping the gf developer
• A hoogle-like thing for gf
• Better error messages
• Accurate file/line position
• Clearer
• Allow reloading of modules from the shell
• Deep integration with NLTK
• Through c runtime (in progress)
• Grammar development using examples
• API to access GF ASTs
  

[Michal Boleslav Měchura]

I think GF will do well if it markets itself more actively to people
who might be interested in writing application grammars for specific
domains. An annual or bi-annual competition for best application
grammar would be nice.

Michal

[Carla Parra]

Hi,

I explained how from my "linguist" perspective the book was not always very clear and I was missing more guidance to understand some points and that therefore I agreed with the "GF for Dummies" proposal and that I could give feedback of the more technical parts from the linguist perspective and can also help trying to explain linguistic notions.

And I also supported the idea of simplifying the editing environment to make it more accessible for linguists as they may think GF could be interesting for them but may get scared by the technical part. I mentioned the survey we did in Spain for the EU-Project CLARIN and how we found out many researchers do things manually just to avoid having to install a new tool and learn how a program is used/updates, etc. If it serves as an example, this is the "virtual lab" that was developed in Spain so that they can execute programs on the cloud and can forget about the "technical stuff": http://clarin-es-lab.org/ (it is in Spanish). 

And well, as for the rest I agree with what was mentioned as far as lexicon enlargement is concerned. Maybe this is also a piece of work boring for programmers but in which linguists and terminologist/lexicographers can contribute without needing broad programming skills.

Carla

[Aarne Ranta]

Hello,

The mission of GF is: 

  Formalize the grammars of the world and make them available and useful. 

I feel GF is going to the right direction and we should go on like that: extend it with new languages and new applications, and make it available for more people. We should make it more stable, correct, efficient, and easy to use - and we should in particular care enough about the users so that we keep the compiler and the library API backward compatible, even when this would cost us the developers some extra work.

The "GF for dummies" website and book is a great idea. But I should keep my fingers out of it - I have written enough GF books, and new perspectives are needed for any new books. I'm confident many of you in the summer school have such perspectives!

Regards

  Aarne.

[Muhammad Humayoun]

Hello,

Here is my piece:

Application grammars are part of large systems usually. The method of adding semantic checks to application grammars in such host systems is already there. In my opinion, it could be very important to keep supporting this feature and further extend it to all GF utilities such as word completion, web services, etc.


/Humayoun

[virk]

Here is mine.
Sooner or later we have to think beyond borders. South Asia is the
home of several hundred languages with a lot of different taste and
variety. I think we should have a bit more collaboration, which might
open doors to many interesting problems. In my view we need to involve
as many people as we can to take GF from research community to real
world problems (even though GF already has started this journey). I
can be the person who can act as a kind of bridge to link to South
Asia through GF. I have already talked about GF and its applications
in my home university, and I see a lot of interest there. A little bit
more struggle and resources can strengthen this link.
Best,

> > On Thu, Aug 25, 2011 at 5:33 PM, Carla Parra <carlaparraescar...@gmail.com

> > > wrote:
>
> >> Hi,
>
> >> I explained how from my "linguist" perspective the book was not always
> >> very clear and I was missing more guidance to understand some points and
> >> that therefore I agreed with the "GF for Dummies" proposal and that I could
> >> give feedback of the more technical parts from the linguist perspective and
> >> can also help trying to explain linguistic notions.
>
> >> And I also supported the idea of simplifying the editing environment to
> >> make it more accessible for linguists as they may think GF could be
> >> interesting for them but may get scared by the technical part. I mentioned
> >> the survey we did in Spain for the EU-Project CLARIN and how we found out
> >> many researchers do things manually just to avoid having to install a new
> >> tool and learn how a program is used/updates, etc. If it serves as an
> >> example, this is the "virtual lab" that was developed in Spain so that they
> >> can execute programs on the cloud and can forget about the "technical

> >> stuff":http://clarin-es-lab.org/(it is in Spanish).

[Laurette]

The future of GF as seen from the southern tip of the African continent ...

The MOLTO project developments and contributions to GF are most exciting and from my perspective it is now strategically important for us to develop resource grammars for the Southern African languages in order for them to share in future GF developments. To achieve this we would have to (1) provide training in GF to local collaborators; (2) host GF experts and post docs to assist in developments; (3) acquire funding to support these initiatives. I hope to work with Olga and Wanyiku (and other interested parties) to develop a funding proposal that might go some way towards facilitate points (1)-(3).  I also support the development of basic material for beginners and would like to be part of this process in some way.

Laurette (sr.)

[Pretorius, Laurette]

This message (and attachments) is subject to restrictions and a disclaimer. Please refer to http://www.unisa.ac.za/disclaimer for full details.

---

Please read “facilitating” for “facilitate” …