MarkovModel

Main idea:

name.txt -> name.dat (markov's chain)

Compile and use from command prompt (Windows):
\...>markov.exe (name) (function) (startPos) (length of sentence)

name: name of your txt file (input)

functions: 1 = doSentence (make sentence using startPos and length) -> output.txt (program saves just sentence)
2 = doSentence2 (no forcing random words) -> output.txt (program saves just sentence)
...

startPos: first word in sentence (type -1 to use random word)

length of sentence: amount of words (type 0 to go until the %END% word)

name.txt (your input) structure:

%START%
Your
Text
Is
Here
%END%
%START%
...
%END%

do your input using parsing

name.dat structure:

0xDEADBEEF
(amount of words)
(word1_word)
(word1_freq)
(word2_word)
(word2_freq)
...
(sum of freqs)
(mainword_word)
...repeat

this means program saw
mainword word1, mainword word2, ...
and made a model