What is the role of log-likelihood ratios in Naïve Bayes classification decisions?

Log-likelihood ratios compare evidence strength between classes for decision-making.

Naïve Bayes often uses log(P(C1|X)/P(C2|X)), which simplifies to difference in log-likelihoods plus prior log-ratio. This provides a stable and interpretable decision metric. It also connects NB to logistic regression structure in log-space.