NAME

AI::CRM114 - Wrapper for the statistical data classifier CRM114

SYNOPSIS

  use AI::CRM114;
  my $crm = AI::CRM114->new(cmd => '/path/to/crm');

  # Learn new text
  $crm->learn(['osb'], 'spam.css', 'MAKE MONEY FAST');

  # Classify some text
  my $class = $crm->classify(['osb'], ['a.css', 'b.css'], $text);

DESCRIPTION

The CRM114 Discriminator, is a collection of tools to classify data, e.g. for use in spam filters. This module is a simple wrapper around the command line executable. Feedback is very welcome, the interface is unstable. Use with caution.

METHODS

AI::CRM114->new(%options)

Creates a new instance of this class. The following options are available:

cmd => '/path/to/crm': Specifies the path to the crm executable.

$crm->learn(\@flags, $file, $text)

Learn that the text belongs to the file using the specified flags. Permissable flags are specified in the QUICKREF.txt file that comes with CRM114. Examples include winnow, microgroom, and osbf.

classify(\@flags, \@files, $text)

Attempt to correlate the text to one of the files using the specified flags. Permissable flags are specified in the QUICKREF.txt file that comes with CRM114. Examples include unique, fscm, and svm.

In scalar context, returns the path of the best matching file. In list context, returns a list containing the path of the best file, and the probability and pR values as reported in (:stats:).

AUTHOR / COPYRIGHT / LICENSE

  Copyright (c) 2009 Bjoern Hoehrmann <bjoern@hoehrmann.de>.
  This module is licensed under the same terms as Perl itself.