Remember that for classification, we want to find:

To classify, we just pick the class with the maximum posterior probability:

The naive assumption assumes all features are conditionally independent given the class.

To make it less naive, we have a smarter Bayesian Decision Theory Framework with Costs.