Markov Chain Sentence Generator

Overview

This is an implentation of the Markov Chain algorithm for generating sentences. The Markov Chain algorithm is simple: It parses a string and determines the probability that one word will follow another. There are four main functions that are used:

• feed(string) - Feeds in a string that is used to create the data for generating new sentences. The idea is that the sentence generator will generate sentences similar to the sentences fed into it.
• feed(XmlDocument) - Feeds in an XML document that was previously exported.
• save(string) - Saves the data stored from inputs into an XML document at the specified path, which can then be imported later using feed(XmlDocument).
• generateSentence() - Generates a sentence using the data stored.

Classes included

• TextMarkovChain - A simple implementation of a single Markov Chain.
• DeepMarkovChain - An implementation of the Markov Chain that takes into consideration the two previous words when deciding what word to generate next. (Note: This implementation does not currently include import and export to XML)
• MultiDeepMarkovChain - An implementation that will consider the previous X words when deciding what word to generate next, where X is any integer greater than 2.

GUI overview

Included in this code is a WPF application that can be used to test out the package. The default Markov implementation used is one that used a word storage depth of four. The GUI allows you to:

• Enter in sample strings.
• Generate one sentence at a time.
• Export the data entered into an XML document.
• Import the data from an XML document.

Note about Depth

The depth entered into the MultiDeepMarkovChain class makes a large difference in the storage size required and the sentences that will be generated. Here is a quick overview:

• Low depth - Sentences will be versatile but are more likely to not make sense. Data storage required is minimum.
• Medium depth - Sentences will be more consistent, but more data storage will be required.
• High depth - Sentences will be very consistent, but will be less flexible. Oftentimes the exact sentences will be generated as were fed into the program. Data storage required is much greater.
• Very high depth - Sentences will generally be exact repeats of sentences that are fed into the program. Data storage will be very high. Essentially a very high depth just stores sentences is a very inefficient manner.