NAME "Text::Corpus::CNN" - Make a corpus of CNN documents for research. SYNOPSIS use Cwd; use File::Spec; use Text::Corpus::CNN; use Data::Dump qw(dump); use Log::Log4perl qw(:easy); Log::Log4perl->easy_init ($INFO); my $corpusDirectory = File::Spec->catfile (getcwd(), 'corpus_cnn'); my $corpus = Text::Corpus::CNN->new (corpusDirectory => $corpusDirectory); $corpus->update (verbose => 1); dump $corpus->getTotalDocuments; DESCRIPTION "Text::Corpus::CNN" can be used to create a temporary corpus of CNN news documents for personal research and testing of information processing methods. Read the CNN Interactive Service Agreement to ensure you abide by it when using this module. The categories, description, title, etc... of a specified document are accessed using Text::Corpus::CNN::Document. Also, all errors and warnings are logged using Log::Log4perl, which should be initialized. CONSTRUCTOR "new" The constructor "new" creates an instance of the "Text::Corpus::CNN" class with the following parameters: "corpusDirectory" corpusDirectory => '...' "corpusDirectory" is the directory that documents are cached into using CHI. If "corpusDirectory" is not defined, then the path specified in the environment variable "TEXT_CORPUS_CNN_CORPUSDIRECTORY" is used if it is defined. If the directory defined does not exist, it will be created. A message is logged and an exception is thrown if no directory is specified. METHODS "getDocument" getDocument (index => $index) getDocument (uri => $uri) "getDocument" returns a Text::Corpus::CNN::Document object for the document with index $index or uri $uri. The document indices range from zero to "getTotalDocument()-1"; "getDocument" returns "undef" if any errors occurred and logs them using Log::Log4perl. For example: use Cwd; use File::Spec; use Text::Corpus::CNN; use Data::Dump qw(dump); use Log::Log4perl qw(:easy); Log::Log4perl->easy_init ($INFO); my $corpusDirectory = File::Spec->catfile (getcwd(), 'corpus_cnn'); my $corpus = Text::Corpus::CNN->new (corpusDirectory => $corpusDirectory); $corpus->update (verbose => 1); my $document = $corpus->getDocument (index => 0); dump $document->getBody; dump $document->getCategories; dump $document->getContent; dump $document->getDate; dump $document->getDescription; dump $document->getHighlights; dump $document->getTitle; dump $document->getUri; "getTotalDocuments" getTotalDocuments () "getTotalDocuments" returns the total number of documents in the corpus. The index to the documents in the corpus ranges from zero to "getTotalDocuments() - 1". "getURIsInCorpus" getURIsInCorpus () "getURIsInCorpus" returns an array reference of all the URIs in the corpus. For example: use Cwd; use File::Spec; use Text::Corpus::CNN; use Data::Dump qw(dump); use Log::Log4perl qw(:easy); Log::Log4perl->easy_init ($INFO); my $corpusDirectory = File::Spec->catfile (getcwd(), 'corpus_cnn'); my $corpus = Text::Corpus::CNN->new (corpusDirectory => $corpusDirectory); dump $corpus->getURIsInCorpus; "update" update (verbose => 0) This method updates the set of documents in the corpus by fetching any newly listed documents in the "sitemap_news.xml" file. "verbose" verbose => 0 If "verbose" is positive, then after each new document is fetched a message is logged stating the number of documents remaining to fetch and the approximate time to completion. "update" returns the number of documents fetched. For example: use Cwd; use File::Spec; use Text::Corpus::CNN; use Data::Dump qw(dump); use Log::Log4perl qw(:easy); Log::Log4perl->easy_init ($INFO); my $corpusDirectory = File::Spec->catfile (getcwd(), 'corpus_cnn'); my $corpus = Text::Corpus::CNN->new (corpusDirectory => $corpusDirectory); $corpus->update (verbose => 1); dump $corpus->getTotalDocuments; EXAMPLES The example below will print out all the information for each document in the corpus. use Cwd; use File::Spec; use Text::Corpus::CNN; use Data::Dump qw(dump); use Log::Log4perl qw(:easy); Log::Log4perl->easy_init ($INFO); my $corpusDirectory = File::Spec->catfile (getcwd(), 'corpus_cnn'); my $corpus = Text::Corpus::CNN->new (corpusDirectory => $corpusDirectory); my $totalDocuments = $corpus->getTotalDocuments; for (my $i = 0; $i < $totalDocuments; $i++) { eval { my $document = $corpus->getDocument(index => $i); next unless defined $document; my %documentInfo; $documentInfo{title} = $document->getTitle(); $documentInfo{body} = $document->getBody(); $documentInfo{content} = $document->getContent(); $documentInfo{categories} = $document->getCategories(); $documentInfo{description} = $document->getDescription(); $documentInfo{highlights} = $document->getHighlights(); $documentInfo{uri} = $document->getUri(); dump \%documentInfo; }; } The example below will print some of the most frequent categories of all the articles in the corpus. use Cwd; use File::Spec; use Text::Corpus::CNN; use Data::Dump qw(dump); use Log::Log4perl qw(:easy); Log::Log4perl->easy_init ($INFO); my $corpusDirectory = File::Spec->catfile (getcwd(), 'corpus_cnn'); my $corpus = Text::Corpus::CNN->new (corpusDirectory => $corpusDirectory); my $totalDocuments = $corpus->getTotalDocuments; my %allCategories; for (my $i = 0; $i < $totalDocuments; $i++) { eval { my $document = $corpus->getDocument(index => $i); next unless defined $document; my $categories = $document->getCategories(); foreach my $category (@$categories) { my $categoryNormalized = lc $category; $allCategories{$categoryNormalized} = [0, $category] unless exists $allCategories{$categoryNormalized}; $allCategories{$categoryNormalized}->[0]++; } }; } my @allCategories = sort {$b->[0] <=> $a->[0]} values %allCategories; my $topCategories = 10; $topCategories = @allCategories if (@allCategories < $topCategories); for (my $i = 0; $i < $topCategories; $i++) { print join (' ', @{$allCategories[$i]}) . "\n"; } INSTALLATION To install the module set "TEXT_CORPUS_CNN_FULL_TESTING" to true and run the following commands: perl Makefile.PL make make test make install If you are on a windows box you should use 'nmake' rather than 'make'. The module will install if "TEXT_CORPUS_CNN_FULL_TESTING" is not defined or false, but little testing will be performed. AUTHOR Jeff Kubina COPYRIGHT Copyright (c) 2009 Jeff Kubina. All rights reserved. This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself. The full text of the license can be found in the LICENSE file included with this module. KEYWORDS cnn, cable news network, english corpus, information processing SEE ALSO CHI, Log::Log4perl, Text::Corpus::CNN::Document