Let Google suggest translations

I think this is a good idea, but:

- let me prioritize the reworked UI before this :)
- we'd better make this a per-group feature, so groups can turn it off, or some people will scream on us to allow automated translations where people will not follow the guidelines for teams (many teams are very quality sensitive)

Let Google suggest translations is really the most worst idea I have ever heard... I would never use Google as translation service. Have you ever read the damn ugly sh** that came out? I believe not... better translate by hand than using Google translation service.

I remember many approvers complained that many suggestions have "Google translation service" quality and are therefore not approved. It was often used as a "name" for bad suggestions :-)

This should be an extra and optional submodule what could also allow us to enable other (much more better) services.

Additional - in reference to claudiu.cristea's comment - Google translation service results are not a suggestion. This "suggestions" are simply wrong in most if not all cases. I cannot remember a day where I've got a suggestion by google translation service that was usable.

CNW for abstraction layer.

I'm strongly against Google translation service. RIP!

In an ideal world, we'd have an abstraction layer for suggestions on the UI at least and to some degree on the backend as well. Plugging in a translation memory based CAT tool would provide way better results. So having optional support for that would be great. Google translate can be one example where the text will indeed not be of highest quality :) See http://en.wikipedia.org/wiki/Computer_assisted_translation

subscribe

I'm with hass. Translation is a job for humans; machines cannot do it. And I wouldn't use Google's suggestion as a basis to start from, because I'd probably leave cruddy bits in the text.

I'm still at the position that this could be a good boost in some cases (it works on the UI and will not mass-submit suggestions for you). However, it would clearly need to be configurable per group I'd say. Also, it might be best to implement this as separate submodule maybe. Anyway, this is not currently in the critical path of the module as far as I'm seeing, since we are focusing on solving workflow issues with suggestion submission and moderation first before we add tools to submit even more suggestions :)

+1 to submodule with abstraction layer. The "suggested staff" should never pasted directly to input element of translation interface.
My experience tolls me that first we need a tool to cleanup machine-translated strings that already take exists because lazy translators

to #5
@claudiu.cristea

I can understand you prefer manual translations.

Form a professional point of view: If the translation can be done faster with the same quality (or maybe better), than I prefer this way.

I don't use google alone, but also bing, open-tran.eu and tausdata.org

on http://www.tausdata.org/index.php/starters-guide you can read:
‚Check out TDA member use cases to learn how others are increasing productivity and producing better machine translation. The benchmarks so far show 50% improvements on average.‘

and i can confirm the value of 50% for machine aided translation.

The most important tools i use are virtaal http://translate.sourceforge.net/wiki/virtaal/index
and omegaT http://omegaT.org

So, maybe you and others are interested to test these tools.

Google Translation is now a paid service, so I'm going to postpone that for a while on branch 7.x