Implement google backend for MT
Description
Activity
Former user 3 August 2017 at 22:51
@Benson Muite Thanks for the info. I will see what I can do to make it easier to integrate with new systems.
Benson Muite 3 August 2017 at 08:00
May also consider presenting what you are doing to academic community to get feedback. For example http://ee.dlut.edu.cn/CWMT2017/index_en.html and http://www.statmt.org/wmt17/
Of particular interest might be the workshop on machine translation, they usually organize a yearly competition to compare quality of different machine translation systems. Usually news tasks are considered, but it may be of interest to add additional categories in future.
Benson Muite 2 August 2017 at 22:46Edited
Ok. There are a variety of other translation engines, e.g.
https://www.apertium.org/index.eng.html?dir=nld-afr#translation
http://anusaaraka.iiit.ac.in/drupal/
Typically have pairwise training. Not sure how hard it would be for users to define their own integrations.
Former user 2 August 2017 at 21:27
@Benson Muite We are still evaluating the translations engines. MS and Google are the ones for initial implementation. Thanks for the suggestion.
Benson Muite 2 August 2017 at 14:25
Might also allow for Yandex, Baidu and Bing translators. An api that would allow use of other translation engines would also be welcome.
Details
Details
Assignee
Reporter
Story Points
Sprint
Priority

Implement new Google backend for MT service
Register account for Google translators service
DTOs for request/response
Requestor should be able to select different type of backend
Default backend if not present in request (env variables)
Refactor data model to support it
Must not mix translations between different providers (because of attribution requirements)
New API to query available translation engine
If no credentials are provided at start up, fall back to dev mode
CP QA notes:
Ensure UI uses correct attribution based on the provider