Hey, when we waked through this on the phone. it looks like you weren't getting the analysis and tokenization that you expected.
I would recommend taking a look at the various analyzers to see if there is one that matches your data better. Sometimes that might be a pattern analyzer, although that might be more intensive than some of the default ones. Also, take a look at in Inqusitor, which allows you to experiment with the different analyzers and tokenizers to get exactly what you want.
Cheers!