Your requirements aren't quite clear (at least to me). You say lines don't have to match 100%, but then, how exactly do they have to match? Is it enough for the first columns to match, or does the second column have to match partly in some way? In the example you give (repeated below), the first column values are identical, but in the second column, one is a substring of the other, and your attempt to tokenize makes me think you might be trying to match more than just the first column. Can you clarify?

# these should match, but why exactly?
#FileA
irriterend    vervelend
#FileB
irriterend    irritant//vervelend
[download]