Autumn 2002

Linne is a program that can classify news by topic. The Linne architecture was designed so that different classifiers and learners can be added with ease. In this particular implementation, three classifying and learning techniques were implemented: naive Bayes, decision tree, and bagging.

I compared the different techniques and found that the bagging learners that use naive Bayes learners bettered the other techniques.

Click here for the diagram of the Linne architecture.

Click here for the API documentation of Linne.

Interesting tidbit: the name Linne came from Carl von Linne, better known as Carl Linnaeus, the "Father of Taxonomy."