The BBC creates news stories in more than 40 languages. So earlier this year we built a prototype to help keep track of them. https://bbcnewslabs.co.uk/projects/mat/
Multilingual Article Tracker (Mat)

Matching translated news articles with the original English-language text.

@BBC_News_Labs This is probably a foolish question, but here goes: would it be possible to assign a unique article ID to the original English language article, then include that original article ID as additional (provenance?) metadata to the translated versions when they're created? That way, you don't have to infer origin by translating back to ENG — it will be explicitly declared in the metadata.
@mdy Not foolish, it's one we asked ourselves. It was just a case of working within limitations of not being able to change an old system, when the new system will eventually make this redundant
@BBC_News_Labs Ah, that totally makes sense. Thank you; appreciate the extra context.
@mdy @BBC_News_Labs I had the exact same thought 👍
It would have been a waste to build such a tool for the future instead of an integrated way to match the articles at the time of creation.