While training a Naive Bayes classifier, zero probabilities in the likelihood term for any class will cause the overall posterior probability of that class to be zero, regardless of other evidence. A common solution is Laplace (add-one) smoothing, which is defined as:

$$P(w_i|c) = \frac{count(w_i,c)+1}{\sum_{w\in V}{(count(w,c)+1)}} = \frac{count(w_i,c)+1}{\sum_{w\in V}{count(w,c)}+|V|}$$

University of Michigan - Ann Arbor

Google

To train a Naive Bayes classifier using maximum likelihood estimation, we first concatenate all documents belonging to category $$c$$ into a single text representation. Then, we use the word frequencies in this concatenated document to compute the maximum likelihood estimate of the probability of each word $$w_i$$ given the class $$c$$:

$$P(w_i|c) = \frac{count(w_i,c)}{\sum_{w\in V}count(w,c)}$$

where $$V$$ represents the vocabulary containing all unique words in the training dataset.

Training the Naive Bayes Classifier

An on-going but a helpful book resource about NLP
https://web.stanford.edu/~jurafsky/slp3/

Speech and Language Processing (3rd ed. draft) 

For words that occur in our test data but are not in our vocabulary (because they did not occur in any training documents),  ignore them—remove them from the test document and not include any probability for them at all.

Unknown words

Some systems choose to completely ignore another class of words: stop words, very frequent words like the and a. But using a stop word list doesn’t improve performance, so in most cases, the entire vocabulary is used.

Learn Before

Related