Sentiment Analysis

text_mining

Hi,

I'm trying to find sentiment of a sentence as positive, negative mediocre, etc.. This is what I'm doing - (I have a test file with labeled data)..
Read CSV --> Process Documents (transform case --> tokenize-->stopword-->stemming-->N-gram) --> Cross Validation (Naive Bayes --> Apply Model --> Performance classification).

I'm getting accuracy and recall, but I'm getting infinite cross entropy and logistic loss.. I need to find the probability of sentence, given the category => p(sentence/category) which in fact is used to calculate cross entropy and log loss.. Any idea of what I must do to get the probability distribution, cross entropy and log loss??

Thanks, any help is greatly appreciated...