Naive Bayes for Multi-Class problem

In this video Jeremy explains in a very intuitive way, how Naive Bayes works. The example that he uses, is a 2-class problem. The review has either positive or negative labels. Let’s assume I have a 3-class problem: positive, negative and neutral.

How do you apply his approach to a 3-class problem? How would the parameters (code) change in this case?