Problem

In the first step of TQE in translate5 all prompts assigned to a LLM ressource where used for either translation or TQE (except the system prompt which was used according to the context in which the LLM resource has been used: translation prompt for translation, TQE prompt for TQE). But users may want to keep the number of resources low and use them for both usage scenarios or even more to come.

Solution

Further improvements to use quality estimation in translate5.

Introduce "Area of usage" to LLM language resource

To be able to use one resource for multiple purposes, we need to define the allowed areas of usage in the language resource properties.

Add additional dropdown for "Area of usage" on the same hierarchical level then "use for client". Currently 2 values are offered: "AI translation" and "AI quality estimation".

In the future "Translation proposals" and "Length shortening" will be added (when those features will be implemented).

When only TQE is selected, do not show the translation specific dropdowns and vice versa. Show them all, when both is selected, TQE and translation.

This Area of usage needs its own column in the language resource overview and all usage areas selected should be displayed there. This should be implemented for the language resource grid in the client tab, tpoo. Inn the project wizard, each are of usage will be displayed in a new window (as already the case with TQE).

Prompt window of the language resource

In the prompt building window of the language resource (not the prompt management in the preferences) introduce a new tab "TQE" and move the existing window to another tab "AI translation". Only do so, when for the resource both areas of usage are selected. 

For "AI translation" it must be ensured, that only prompts are offered for usage, that have "AI translation" set as value for usage.

And in the "TQE" tab only prompts of the prompt management can be used, that have the value "AI quality estimation" set. If a prompt should be used in both contexts, the user can go to the prompt management and give both usage prooerties to the prompt.

The test area is changed to one that (in contrast to the current one that will further be used for translation) contains fields for source and target, that will be sent to the AI and as a result we will show:

• the unparsed answer as json
• the parsed answer
• the actual result that would be saved as estimation value