We’ve already seen how to use Markov chains to generate random words that are based on the essence of a previously analyzed corpus. Well the exact same algorithm can be applied to text. The base entities become words instead of letters. I make punctuation be part of the entities, this way, sentence flow becomes part of the extracted statistical essence.
Feel free to send me ideas of cool corpora to analyze.
You can play with it here: