I think the regex doesn't match it so it doesn't actually register that but it still replies that. @Undo Can you please check whether there is something in the falsePositives.txt file that is not a pair of a site and a question ID?
@Unihedron, if this was directed at me, I understood that you wanted to create a regex that would be more strict in it's determination of a POSSIBLE phone number, before sending it to the phonenumbers validation check. Correct?
Initially I did. I think you were saying you wanted to be able to narrow down false positives like 0x2e2f6c7236323900 by checking if the "phone number" part of that is surrounded by other chacters
in this case, the preceeding c would eliminate the check
Would that mean that in the case of ELU and its voodoo witchdoctors, a phone number would only need to include an o instead of a zero for it to pass the test?
ELU uses Georgia, with non-lining figures, and 0 and o look very similar.
I'm afraid that as far as I'm concerned, false positives are far preferable to missed positives.