From HomoExcelsior

Markov chains refer to several different computational techniques. All are based on utilizing the transitional probability between states to derive solutions to various problems.

A common markov technique would be to examine a large text corpus and analyze the frequency of word pairs (or triplets, etc). The analysis could be used to generate random similiar texts, or to compare the document to other documents for common chains.

See also: semiotics, Bayes Law, conditional probability, Bayesian Modeling, Monte Carlo, Simulated Annealing, Genetic Algorithms

Examples:

Here's a fun programming exercise in text generation: Read in a text, for every pair of words in the text record the word following the pair. Use this to create a list of all the words (follow-on words) which might follow a pair of words. Generate a random text by starting with any two words and randomly choosing one of the follow-on words.

---