Bayesian filtering for spam

Question

I was wondering if there is any good and clean oo implementation of bayesian filtering for spam and text classification? For learning purposes.

Answer 1

maybe https://ci-bayes.dev.java.net/ or what about http://www.cs.cmu.edu/~javabayes/Home/node2.html?

I never played with either.

Answer 2

Here is an implementation of Bayesian filtering in C#: A Naive Bayesian Spam Filter for C# (hosted on CodeProject).

Answer 3

In French, but you should be able to find the download link :) PHP Naive Bayesian Filter

Answer 4

Check out Chapter 6 of Programming Collective Intelligence

Answer 5

I definitely recommend Weka which is an Open Source Data Mining Software written in Java:

Weka is a collection of machine learning algorithms for data mining tasks. The algorithms can either be applied directly to a dataset or called from your own Java code. Weka contains tools for data pre-processing, classification, regression, clustering, association rules, and visualization. It is also well-suited for developing new machine learning schemes.

As mentioned above, it ships with a bunch of different classifiers like SVM, Winnow, C4.5, Naive Bayes (of course) and many more (see the API doc). Note that a lot of classifiers are known to have much better perfomance than Naive Bayes in the field of spam detection or text classification.

Furthermore Weka brings you a very powerful GUI…

Answer 6

Don't waste your time on SPAM filtering usages. Spammers easily bypass Bayesian filtering by adding random text to their spam emails.

FIX:
OK,OK, Bayesian filtering can be useful when trained personally. However, at a corporate level or above, its probably useless.

Answer 7

nBayes - another C# implementation hosted on CodePlex