When reading a story posted in the JMP User Community, you may have seen (or regularly use) the Choose Language feature at the top of a story. This feature is provided to translate the Community pages into seven languages and dialects.
Starting Jan. 18, Community translation will move away from using the Lingotek system, which is a cost savings for JMP, and start utilizing Google Translate along with a solution developed by our JMP Localization team.
What does this mean for you and the JMP Community? “Immediate translations will be unavailable for several weeks, but eventually they will be translated and available,” User Community Manager Ryan Gilmore said. “Any posts before Jan. 18 will have translations. Otherwise from a UI perspective, users will not see a change.”
A little more about our JMP Localization team translations
Last year, the JMP localization team trained custom Machine Translation (MT) engines using existing JMP User Interface and Help translations. The team accumulated this translation data over the years for nine languages they translate JMP into: German, French, Italian, Spanish, Simplified Chinese, Japanese, and Korean. Those translations have been performed by professional and experienced human translators and were validated by in-country experts and users.
“By using existing translations for training custom MT models, we get translation output that is far more consistent with JMP’s linguistic style and vocabulary than the output from off-the-shelf, generic MT models like Google Translate,” Globalization Architect Krzysztof Jozefowicz said.
Anyone at JMP can already use the custom models interactively.
Together with JMP Marketing, the Localization team will use the MT models for on-the-fly and batch translation of discussions on Community pages. Right now, they are being translated using generic MT models.
After the switch to JMP MT engines, translations will be far more precise and consistent with JMP’s internal terminology, especially for text strings that refer to JMP user interface elements and statistical and analytical vocabulary.
“Our MT engines are continuously re-trained with new human translation data that we build with every new release,” Jozefowicz said. “Hence, the output is getting better and better with time.”
As of January 30, 2024, translations are now available again in the Community!
Great work @Ryan_Gilmore and team.
I know our community of users will appreciate the higher quality of translations that this change has enabled.
I tested several things.
Thanks on behalf of Korean users, Great !,
