Hello Vincent Ciganik,
Welcome to the Microsoft Q&A and thank you for posting your questions here.
I understand that you are having issues with the Incorrect language detection and translation.
By implementing multi-layered strategy below, you can address both the initial language detection flaw and the more critical issue of translation inaccuracy for short phrases in low-resource languages. Use the associated links for more steps and details:
- Avoid Auto-detection for Short Texts: Prefer specifying the
fromparameter whenever possible. - Translate Method -fromparameter- https://free.blessedness.top/en-us/azure/ai-services/translator/reference/v3-0-translate#optional-parameters - Boost Detection with Context: When using auto-detection, provide a longer text sample. - Detect Method- https://free.blessedness.top/en-us/azure/ai-services/translator/reference/v3-0-detect
- Fix Bad Translations Permanently: For recurring translation errors, use Azure Custom Translator with a dynamic dictionary to enforce the correct output. - Custom Translator - Dictionary- https://free.blessedness.top/en-us/azure/ai-services/translator/custom-translator/what-is-dictionary
- Implement a Client-Side Fallback: As a last resort, create a heuristic post-processing layer to correct known translation failures.
For a production system, your integrated workflow should look like below:
I hope this is helpful! Do not hesitate to let me know if you have any other questions or clarifications.
Please don't forget to close up the thread here by upvoting and accept it as an answer if it is helpful.