(C)NB classifier scores

[Grant Ingersoll]

What's the interpretation of scores for the output from the new (complementary) 
naive bayes classifiers?  Larger is better, right?

Thanks,
Grant