Classifying Reuters news articles using Naive Bayes Classifier.
Class Reuters gets a location argument that specifies *.sgm files of dataset, then you can use read() function to read files into 5 output lists corresponding:
train_documents, train_labels, test_documents, test_labels, label2topic (list that maps labels to topics string).
Class NaiveBayes is used for initializing the classifier. then you train classifier with train data, finally you can test the trained model using test() function.
This classifier first preprocess data by removing punctuation marks and numbers, removing extra white spaces, and making all letters lower form.
Then tokenize the documents for training the model.