I've been playing around with extending the named entities via the sample code from issue #187. Nearly everything is working fantastically. However, if I try to train a sentence with an "im" (or "Im" for that matter) I get an list index out of range error from GoldParse. For example, attempting to train the following sentence and it's associated BILOU format will throw the exception:

What is Jon Im Chan phone number
['O', 'O', 'B-XPERSON', 'I-XPERSON', 'L-XPERSON', 'O', 'O']

My only guess is that internally it's expanding Im to I'm and not adding in the extra tags, but that's just a wild guess.