Starting today, 2-JUL-2020, I am using a new technique for the antipatterns to fix false positives:
- Do a check against 200 000 sentences (“before.txt”);
- Do another check just by placing the false positive tokens in the antipattern (“after.txt”);
- Place the exceptions found in the previous step and do another check.
I will do a diff using TortoiseSVN to validate the results.
This technique will remove more false positives than before because sometimes I would spend hours trying to fix false positives and the fix would only solve very little matches (wasting hours just to fix three or four false positives in some cases).