Fluenta - management of translation memories

[petrht]

Hi guys,
I know this is a question rather for Fluenta staff, but I thougth I might ask here first, maybe someone might have experience with this.
It seems that the major problem of Fluenta add-on is that each DITAMAP you create has a UNIQUE PROJECT/UNIQUE TRANSLATION MEMORY (which is bad). Isn't there any way how to share a single TM for all ditamaps you create/work with?
This unfortunate settings leads to a situation where - for example- you have DITAMAP A, you generate an xliff for your supplier, it gets translated (together with warehouse file or other files you reuse content from). However, when you have DITAMAP B (for example datasheet for a different component), you basically do not have access to the TM from DITAMAP A, as Fluenta automatically creates a separate project/TM. What's worse, even the files you use content from as well as the warehouse file is - from this perspective - untranslated.
Any experience with this?
Thanks
petr

[xephon]

Hello Petr,
we are using Fluenta and Oxygen, but not the Fluenta add-on. To support the development of Fluenta, you should buy a license ($200 per year) here.

So far, we are not using the translation memory functionality of Fluenta at all, mainly because we have an agreement with our LSP, that 100% matches are auto-approved and not paid. This means, it does not hurt, if Fluenta sends already translated content to the LSP again. So we do not really need memories on the clients.

However, we are planning to use the translation memory functionality of Fluenta, because we want a pure AI-based translation for urgent/low amount translation jobs. As you probably know, it takes a lot of time and cost to spin up the big wheel. We have a quite high amount of boilerplate costs for minimum fee and project management costs. So, the Fluenta memory functionality can help us to filter out the new segments and then we could in theory auto translate those XLIFF segments, for example with the Oxygen AI Positron assistant. IMHO this should work with ai:transform-content, but I have not tested this yet.

What you could do in your particular case, is, that you get your TMX files from your LSP frequently and import them to Fluenta, maybe by using the Fluenta CLI: Import TMX File.

I have scripted the entire Fluenta process and blackboxed it for the technical writers. I have created two buttons in Oxygen: Export XLIFF and Import XLIFF. These buttons call scripts, which generate Fluenta projects on-the-fly. Copy and rename files and directories, change xml:lang attributes, and so forth.

[Radu]

Hi,
I also added an internal issue on the Oxygen issues list to see if our Fluenta add-on can be improved in the future to allow sharing the same translation memory for two projects or having a single translation project for multiple DITA Maps.
Pasting the issue ID below:
EXM-54596 Fluenta - use the same translation memory for multiple projects

Maybe a hack to work around this would be to create an extra DITA Map used only for translations which refers to all other main DITA Maps and create a project for that large translation DITA Map.
Other than that you should try to use directly the Fluenta application, maybe it has some support for what you are after or contact the developer of the Fluenta set of tools:
https://maxprograms.com/products/fluenta.html

Regards,
Radu