https://github.com/greg-kennedy/markovchain
A Markov Chain class for text manipulation, in Perl.
https://github.com/greg-kennedy/markovchain
generative-text markov markov-chain perl perl-module perl5
Last synced: 11 months ago
JSON representation
A Markov Chain class for text manipulation, in Perl.
- Host: GitHub
- URL: https://github.com/greg-kennedy/markovchain
- Owner: greg-kennedy
- License: other
- Created: 2015-12-04T19:59:56.000Z (over 10 years ago)
- Default Branch: master
- Last Pushed: 2020-08-01T03:44:55.000Z (almost 6 years ago)
- Last Synced: 2025-02-05T16:47:48.347Z (over 1 year ago)
- Topics: generative-text, markov, markov-chain, perl, perl-module, perl5
- Language: Perl
- Homepage:
- Size: 9.77 KB
- Stars: 0
- Watchers: 3
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.pod
- License: LICENSE.md
Awesome Lists containing this project
README
=pod
=head1 NAME
MarkovChain - A module using a non-weighted Markov Chain for text manipulation.
=head1 SYNOPSIS
use MarkovChain;
my $chain = MarkovChain->new($order, $limit);
$chain->add("some string", @list_of_more_strings);
print $chain->spew;
=head1 DESCRIPTION
This module creates new sentences from a list of input strings. After the
module creation, you should seed the generator with one or more sentences.
Calling the spew method will print a new sentence composed of words from
the previous.
=head2 Methods
=item C
Returns a new MarkovChain object.
The first parameter to this object sets the "order", or
length of sub-phrases to use for generation. Higher order
leads to more of the input text being preserved, but
also more coherent output. Lower is more chaotic.
Defaults to 3.
The second parameter, "limit", sets a maximum length
of the output chain. When the limit is reached, the sentence
is terminated. This is primarily used to avoid infinite loops.
Defaults to 50.
=item C
Adds one or more strings to the class. This has two effects:
the first words of the string are added to the "sentence
beginnings" list. These are randomly chosen to use later.
Second, the sentence is parsed into phrases. Each phrase
is stored with a follow-up word, and are used to form the output.
Input strings are somewhat normalized for consistency, by removing
all non-word non-space characters, before usage as lookup keys.
This helps tokens find more possible matches. Outputs are not
normalized, however, so users may wish to remove problematic characters
such as quotes or parentheses first.
Returns the number of sentences successfully added.
=item C
Returns a new string, composed randomly from the phrases
currently added to the class.
=head1 LICENSE
This is released under the Artistic License. See L.
=head1 AUTHOR
Greg Kennedy - L
=head1 SEE ALSO
L
=cut