Is there a way to prevent duplicate transactions in Transaction Matching?

We are looking for a solution where we could load all "new" transactions except those that are marked as a duplicate within a file.

For example, using an identifier field such as ID #, we could load transactions with new ID #'s while ignoring transactions with ID #'s that have been previously loaded.

The only workaround we are aware of is marking a field as a key attribute. But it seems that in this case, the entire file would get kicked out if any transaction contains a value from a field marked as a key attribute.

I know this question has been asked before, but looks like it had been a while so figured it was worth re-asking.