Differential Re-Translation POC

Proof of concept: explores how the reasoning capabilities of an LLM (ChatGPT) can support automatic adaption of new translations according to previous manual corrections. This is intended for the following scenario. A text is automatically translated, checked by the user and changed manually. Now a new version of the original comes in. The idea of the differential re-translation is to translate the new version of the original text automatically, while having the LLM keep the manual changes as far as possible. Done for the authomatic translation process for Adobe AEM in the Composum AI.

Enter the specific requirements for the translation, such as tone, formality level, or additional contextual information.

Input the text you wish to translate. This will serve as the base for the initial automatic translation.

View the initial translation of your original text. This text is generated based on the general instructions.

Make adjustments to the automatic translation to better meet your requirements or correct inaccuracies.

Enter a modified version of the original text to see how the application adapts the translation to reflect these changes.

View the result of the translation of the changed original text while trying to maintain the manual corrections.