You're right but I don't think staying accurate is all that important in this case. Judging by the "Features" section this seems to be mostly concerned with recognizing names in strings which means you can discard some extra information as long as you can still accurately match names without too many false positives. I can't say anything about exhaustiveness.