Adaptive Spam Filtering Using Only Naive Bayes Text Classifiers
Aris Kosmopoulos, Γεώργιος Παλιούρας, Ion Androutsopoulos · 2008
In the past few years, machine learning and in particular simple Naive Bayes classifiers have proven their value in filtering spam emails. We hereby put Naive Bayes filters to the test, against potentially more elaborate spam filters that will participate in the ceas 2008 challenge. For this purpose, we use the variants of Naive Bayes that have proven more eective in our earlier studies. Furthermore, we propose a simple active learning method for adapting the filter, under partial online supervision.