skwzrd / markov_chain

Markov chain for sentence generation.

Markov Chain

Set Up

Add a text file called ramblings.txt to the same directory as markov.py. Have this file contain a reasonable number of sentences.

Basic Principles

A dictionary is going to be generated from the sentences in ramblings.txt with the structure {word_A: {succeeding_word_A1: count_A1, succeeding_word_A2: count_A2, ...}, word_B:{...}...}. An example of this dictionary is:

ramblings = "The dog chased the dog around the yard."
# forms the dicitonary:
word_dict = {
    'the': {
        'dog': 2,
        'yard': 1,
    },
    'dog': {
        'chased': 1,
        'around': 1
    },
    'chased': {
        'the': 1
    },
    'around': {
        'the': 1
    }
}

With this dicitonary, when we're given a word, we can calculate a probability of another word succeeding it, and then use that probabilistic word to build a string of words (a sentence).

Sentences begin by choosing a random word from word_dict's keys, and they end if (1) the set sentence length is reached, or (2) if a word has no successors.

Examples

>>> py markov.py
"Someone insists on the importance of southeast china and studies."

>>> py markov.py
"Only real programming language for the best language for days."

>>> py markov.py
"Snapshot of the place was earning really my main chat."

Obviously these sentences are not very coherent, but they're fun to read nonetheless.