On Tue, Aug 14, 2012 at 2:14 PM, Francis Tyers <[email protected]> wrote:
> El dt 14 de 08 de 2012 a les 10:35 +0200, en/na Per Tunedal va escriure:
> > Hi,
> > offering non-released language pairs to non-expert users may accelerate
> > the development of new language pairs if we collect information from the
> > users. Víctor Manuel Sánchez Cartagena described un interestering
> > project in a previous post:
> >
> > "very common sequences in a monolingual corpus in SL are provided to
> > non-expert users, and they are asked to translate them. They are
> > encouraged to use the translation of each word according to the
> > bilingual dictionary and the context of the segment to translate in the
> > SL monolingual corpus is also provided."
> >
> > My idea is to offer users to try unreleased pairs and simultaneously
> > offer them the the opportunity to contribute in various ways, cf Google
> > translate: "Suggest a better translation". We need a built in system for
> > reporting and collecting errors and suggestions, as an input for the
> > developers. It should be simple to suggest a new or better translation
> > for a word or expression.
> >
> > Reactions from real users would also be stimulating for the developers.
> > It's more satisfying to sit with tedious work if you know that some one
> > cares about the result of your efforts. Thus, an implementation of a
> > system for user contributions would be of double use: accelerating the
> > development from two sides.
> >
> > Yours,
> > Per Tunedal
>
> This would be interesting. We used to have a page set up where people
> could see unstable pairs. The problem was that it made the server go
> down quite often, so we turned it off.
>
It would also be possible to create packages for unreleased pairs so that
the programs that are based on lttoolbox-java (the Android app, Apertium
Caffeine...) would support them. From my point of view, we shouldn't. If a
language pair happens to be unusable in practice, I think that there is no
point on offering it to final users. And, if somebody really wants to play
with it, it's always possible to
go to SVN and compile it. I guess that the typical user that just wants to
translate a text would never need to do something like that. And for people
more familiar with Apertium, it wouldn't be hard to do it. But, well,
that's only my opinion. If you think that it would be useful, I will create
packages for unreleased pairs(GSoC is about to finish, but I could work on
that later too).
As for the idea of having something similar to Google's "Suggest a better
translation", I guess that we could also implement that in Apertium
Caffeine. I think that it would make sense not only for unreleased pairs as
you propose, but also for released ones (in fact, I think that user
feedback would actually make more sense for released pairs, since
unreleased pairs probably contain many obvious mistakes that developers are
already aware of). The work in the client side would probably be easy. The
question is what to do in the other side (i.e. how/where to send the user
reports).
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Apertium-stuff mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/apertium-stuff
