dbacl (GPL 3 or higher)

dbacl is a fast Bayesian text and email classifier. It builds a variety of language models using maximum entropy (minimum divergence) principles, and these can then be used to categorize input data automatically among multiple categories.

You can find out more about dbacl by reading any (or all) of these documents:

• Language models, classification and dbacl (tutorial)
• Email classification with dbacl (tutorial)
• Is it working? (tutorial)
• Comparing dbacl with other classifiers (tutorial)
• The dbacl classifier (technical, ps.gz)
• dbacl at the TREC 2005 (technical, ps.gz)
• Can a Bayesian filter play chess? (essay)
• Online manual pages

dbacl also has its own sourceforge project page.

dbacl-1.14.tar.gz (MD5)

dbacl-1.13.tar.gz

dbacl-1.12.tar.gz

dbacl-1.11.tar.gz

dbacl-1.10.tar.gz

dbacl-1.9.tar.gz

dbacl-1.8.1.tar.gz

dbacl-1.8.tar.gz

dbacl-1.7.tar.gz

dbacl-1.6.tar.gz

dbacl-1.5.1.tar.gz

dbacl-1.5.tar.gz

dbacl-1.4.tar.gz

dbacl-1.3.1.tar.gz

dbacl-1.3.tar.gz

dbacl-1.2.1.tar.gz

dbacl-1.2.tar.gz

dbacl-1.1.tar.gz

dbacl-1.0.tar.gz

dbacl-0.9.tar.gz

dbacl is a digramic Bayesian classifier. It can distill text documents into categories, and then compare other text documents to the learned categories. Both dbacl and its companion programs are written in C and run on UNIX/POSIX (screenshot).