Ironically, naive Bayes, as standardly presented, is not Bayesian. At least not in the sense that we estimate a posterior distribution over parameters given data and use it for predictive inference for new .

Nice intro to Dirichlet and why “Naive Bayes” is not truly Bayesian.