Final Thoughts - Ending Spam: Bayesian Content Filtering and the Art of Statistical Language Classification [Electronic resources] نسخه متنی

Final Thoughts

Markovian classification performs primitive, conceptual, and lexical analysis on a text sample, providing much higher levels of precision than the standard primitive tokenizers that usually accompany Bayesian filters. It is significantly more resource intensive than standard Bayesian analysis and therefore requires special attention to training mode and storage implementation. For small- to medium-sized systems, Markovian classification can provide very high levels of accuracy.