MarkVSharp is a .Net library to generate random text that looks grammatically correct.
It is inspired by the original Mark V Shaney algorithm:

http://en.wikipedia.org/wiki/Mark_V_Shaney

This is done by by parsing a large input text document to create base word snippets that are related.
These base word snippets are randomly selected to generate correct looking text while attempting to preserve
the original relationship.

All consecutive words represent states in a Markov chain.  The decision of what word to generate next is
made by the prior two words.


Requirements:

  Mono version 2.8 or greater
  or
  .Net version 3.5

Building:

  On Windows open and build MarkVSharp.sln using VS 2008 or SharpDevelop
  On Linux open and build Mono.MarkVSharp.sln using MonoDevelop 2.4 or greater
  All unit tests should pass after building

Using MarkVSharp:

  Mark V Sharp has been tested with text novels hundreds of pages in size as input.

  Add a reference to MarkVSharp.dll
  The only namespace needed is MarkVSharp 
  
  A text file with only plain text data is needed.  Good files that can be used are books 
  found at Project Gutenberg: 
  http://www.gutenberg.org/wiki/Main_Page

  Let us assume we have a book in plain text called book.txt.

  Create Generator:
  GeneratorFacade gen = new GeneratorFacade(new MarkovGenerator(System.IO.File.ReadAllText("book.txt")));


  We can generate the following:

  Words -- Lower case words
  gen.GenerateWords(6) ; //generate 6 words

  Sentence -- One sentence
  gen.GenerateSentence(3) ; //generate a sentence with a minimum size of 3

  Title -- Upper case words
  gen.GenerateTitle(6) ; //generate 6 words in title format

  Paragraphs -- Multiple paragraphs with appropriate indentation
  gen.GenerateParagraphs(4); //generate 4 paragraphs with random number of sentences

TODO:

  The following is planned for the future:
  -- REST service hoster for MarkVSharp in order to use it from other languages
  -- More fine grained control over the various generation methods
  -- Additional convenience methods to better filter input text
   
LICENSE:

  BSD 2-clause license
  See LICENSE file