Calculating B and |V| in naive bayes text classification

Question

I found a link about multinomial naive bayes classifier

How we could calculate the B' or |V|?

The page said that it is the number of terms in the vocabulary. In its example, how we could get 6 for B? Is it the counting of all term?

"chinese", "beijing", "shanghai", "meacao", "tokyo", "japan"

One more question, what if new term appear in testing document? example, in doc 6 appears "bangkok" or any new word that never appear before. how to count the probability of new term ?

score 1 · Accepted Answer · edited Apr 20 '13 at 08:33

1

You are right. It's the total number of words in the vocabulary, since there can be only one entry for a term in the vocabulary.

edited Apr 20 '13 at 08:33

quetzalcoatl

32,194
8
68
107

answered Apr 20 '13 at 08:29

passion

44
1

One more question, what if new term appear in testing document? example, in doc 6 appears "bangkok" or any new word that never appear before. how to count the probability of new term ? thank you passion :) – mhabibi Apr 20 '13 at 09:38

Calculating B and |V| in naive bayes text classification

1 Answers1