Editing Portal:Artificial Intelligence (section)

=== Classification, machine learning, and search ===

Perl has had CPAN modules for traditional machine-learning tasks, especially classification and optimization. `AI::DecisionTree` is a module for learning decision trees from training data.<ref>MetaCPAN, “AI::DecisionTree,” https://metacpan.org/pod/AI::DecisionTree, accessed July 3, 2026.</ref> `Algorithm::NaiveBayes` provides Naive Bayes classification tools.<ref>MetaCPAN, “Algorithm::NaiveBayes,” https://metacpan.org/release/Algorithm-NaiveBayes, accessed July 3, 2026.</ref> `Algorithm::SVMLight` provides a Perl interface to the SVMlight support-vector-machine package.<ref>MetaCPAN, “Algorithm::SVMLight,” https://metacpan.org/pod/Algorithm::SVMLight, accessed July 3, 2026.</ref>

Perl also has modules for evolutionary computation. `AI::Genetic` is described as a pure-Perl genetic algorithm implementation.<ref>MetaCPAN, “AI::Genetic,” https://metacpan.org/pod/AI::Genetic, accessed July 3, 2026.</ref> `Algorithm::Evolutionary` provides classes for evolutionary computation in Perl.<ref>MetaCPAN, “Algorithm::Evolutionary,” https://metacpan.org/pod/Algorithm::Evolutionary, accessed July 3, 2026.</ref>

Text categorization has been especially natural for Perl because Perl is strong at tokenizing, normalizing, filtering, and counting text. `AI::Categorizer` is described by Fedora packaging metadata as a framework for automatic text categorization, with flexible choices for algorithms, features, and document formats.<ref>Fedora Packages, “perl-AI-Categorizer,” https://packages.fedoraproject.org/pkgs/perl-AI-Categorizer/perl-AI-Categorizer/, accessed July 3, 2026.</ref>

These tools do not represent the cutting edge of modern deep learning, but they matter historically. They show that Perl was useful for machine learning before the current AI era, especially when the problem involved text, documents, feature extraction, classification, or integration with existing systems.