Versioning the dataset is an excellent idea.
Would it be possible to create a VERSION file within the archive instead of using the filename to describe its contents? Filenames can get munged in transit and, more importantly, the archive filename won’t be present once it has been extracted.