As far as I understand, this routine triggers only when word1 is within the regexp 2, but not in regexp 1 (and vice versa)
I would rather see it triggers always for both of them unless it is word 1 in context 1, or word 2 in context 2.
That way, it would be triggered more. Which would improve the rule (in my opinion).
A sentence is just too short to determine the context of a word.