Markovian discrimination

The Markov filters (after Andrei Markov ) is a spam filter based on a Hidden Markov Model and is a further development of the Bayesian filter dar. The Markov filters calculated includes the likelihood of the match the word chains of the revised text to word chains typical spam texts. While a Bayesian filter the likelihood of individual words is calculated moves of Markov chains word filter approach to determine the probability and weighted the individual possible combinations. Resembling the word chains of the revised text which typical spam texts, the revised text is spam.

Example of weighting of the possible combinations

Using the example of the sentence " The quick brown fox jumps ... " you can illustrate the possible combinations and weightings 22N in Markov filters:

Formal representation of the probability computation in Bayesian and Markov filter

While the probability on the basis of the Bayesian filter by

Is specified, for the Markov filters