llwyd / mBot

Markov chain twitter bot engine

markovMagpie

A terrible markov chain engine for generating tweets

How it works

Data Harvesting

• Assuming a sentence has n words.

• Two Arrays (Dictionaries)

  1. Contains word[0] with all possible word[1].
  2. Contains word[1:n] with all possible word[x+1].

• Python allows arrays within arrays

  • {'Word[0]': ['Word[1]', 'Word[1]', 'Word[1]']}
  • {'Word[x]': ['Word[x+1]', 'Word[x+1]']}
    • Array can be accessed recursively

Generating Tweets

• Randomly select word[0].

  • Known as the ‘key’
  • Key is chosen at random to start the chain

• Randomly select word[1] based on word[0].

• Randomly select word[x] until:

  • word[x] has no word[x+1].
    • Typically a word with a full stop.

Further Reading/References

• [1] zirak (2013) http://blog.javascriptroom.com/2013/01/21/markov-chains/

• [2] Vivin Paliath (2010) https://stackoverflow.com/questions/4081662/explain-markov-chain-algorithm-in-laymans-terms