I'm posting this message from Facebook, since it might be relevant for users of CafeTran Espresso too:
FYI: If you translate #Studio files in memoQ and do a QA in Studio afterwards, you'll get "Non-edited fuzzy match" warnings for high fuzzy matches, regardless if you edited them them in memoQ or not.
This is because when you edit a fuzzy in Studio, the program stores additional information abut that edit in the SDLXLIFF file, including hash of the original match, hash of the edited segment, date of the edit and name of the editing user. Because of the content hashes, I don't think it would be possible to add a "dummy" content externally (e.g. with a script executed on export from memoQ).
A possible workaround (suggested by Kevin Lossner) is to filter such fuzzies in Studio and run Find and Replace to replace, for example, the first word of a segment with itself (Find: ^(\S+); Replace: $1) and confirming in bulk.