Spam Filtering based on Naive Bayes Classification
Tianhao Sun · 2009
This project discusses about the popular statistical spam filtering process: naive Bayes classification. A fairly famous way of implementing the naive Bayes method in spam filtering by Paul Graham is explored and a adjustment of this method from Tim Peter is evaluated based on applications on real data. Two solutions to the problem of unknown tokens are also tested on the sample emails. The last part of the project shows how the Bayesian noise reduction algorithm can improve the accuracy of the naive Bayes classification.