Building a Naive Bayes Text Classifier and Accounting for Document Length (English Edition)

Classification is the process of determining the class to which an object belongs. An example of such classification is text classification: assigning a document to a category. When the number of documents is large, classification requires automation. One statistical method for performing text classification on a computer is the naive Bayes classifier. This report presents the theory behind such a classifier and explains the relationship between document length and the Bayesian equations for classification. This report also presents the implementation of various naive Bayes classifiers and examines their performance. Among these classifiers, the multinomial naive Bayes classifier performs best.