Markov Chain based text generator. Reads a Twitter user's tweets and produces similar tweets.
The prod branch is for the code which will run on the server, posting every hour to Twitter.
The master branch is for the current up-to-date code. It's separate from prod because things like absolute file paths inevitably make it through and must be weeded out before pushing to prod.
- Implement burn-from-both-sides algorithm for sentence generation
- Add support for YML/JSON config files
- More graphing
- Tidy the pre-processor even more
- Tests...
- More tests...
- Try a neural approach
There is a bot currently running on Twitter