A naive Bayes classifier with Laplace smoothing for arbitrary features written in C#.
src
contains the library's source codetest
contains the unit testssamples
contains the test application
The SmsSpam demo application uses the SMS Spam Collection v.1 in order to train the classifier with Spam and Ham SMS text data and then match individual SMS against that. Cleaned, but otherwise raw words are used (i.e. stripped from whitespace, numbers, punctuation, ...) as tokens.
This code is, if not otherwise stated, available under MIT License. See LICENSE
file for more information.