This project is about experimenting with Markov Chains to generate News text having a writing style similiar to that of Fiction or Non-Fiction texts .
For more detail kindly follow report.pdf
Steps to generate the text
Class:- Markov(text_corpus)
show_summary()
generate(n,start,length,alpha)
mixed_n_generate(start,length,alpha)
show_summary() is a function to print the summary of the text
generate() is the function to generate text with the following arguments:-
-
n: - previous n words to consider
-
start:- seed words to generate the text
-
length:- number of words to generate in the text
-
alpha :- Hyperparameter, perferably 0 for better results
mixed_n_generate() is the function to generate text by using a combination of 1-word, 2-word, 3-word markov chains. The arguments are same as that of generate()
To generate using a single n-word markov chains:-
open the terminal in the file location and run python markov.py path_to_text_corpus n seed length alpha 1
example: run python markov.py path_to_text_corpus 2 "He is" 50 0 1
To generate using a mixed markov chain model:-
open the terminal in the file location and run python markov.py path_to_text_corpus seed length alpha 2
example: run python markov.py path_to_text_corpus "He is at" 50 0 2