Markov Wordsearch Generator

This was a little experiment to play with Markov chains. The wordsearch is generated using a 2nd order Markov-chain. That is, the probability of the next letter is determined by the previous two letters.

The Markov model was trained from the complete works of Shakespeare, and used to generate a 2D grid of letters, randomly creating a wordsearch, which likely contains words. This also makes it difficult, because everything has the appearance of English words.

Limitations:

• Biased towards common words (like “the”, “and”, “with”, etc.), because these weren’t filtered out of the training set.
• Biased toward shorter words, because it is only 2nd order.