I’ve recently (2016) started working on some code that generates readable texts, with the ultimate goal of algorithmic generation of a whole novel. Realistically, I’d love to generate a page or two of readable text…
Inspired by NaNoGenMo (though I’m not actually participating - maybe one day?).
Below are some samples of what I’ve produced so far, combining bi-directional Markov chains, snippets from Wikipedia and chunks of Project Gutenberg texts. I’ve also put in a few notes about methods I’ve explored.
Update: looking back at this in early 2020, and the code methods have been surpassed by the current generation of neural models, such as BERT and GPT-2, that produce far richer text than I have. But I learned a lot doing this, so here it stays!