Email Classification Using Machine Learning Algorithms

Email has become one of the frequently used forms of communication. Everyone has at least one email account. Inflow of spam messages is a major problem faced by email users. Currently there are many spam filtering techniques. As the spam filtering techniques came up, spammers improved their methods of spamming. Thus, an effective spam filtering technique is the timely requirement. In this paper email classification is done using machine learning algorithms. Two of the important algorithms namely, Naïve Bayes and J48 Decision Tree are tested for their efficiency in classifying emails as spam or ham. The experiment focused on classification in combination with pre-processing techniques and concepts of text categorization. The dataset used is Enron Corpus. TF-IDF value is used as the weight score of text. The classifiers are also tested for different feature size. The test results show that J48 is more accurate in classifying emails as spam or ham with a minimum feature size and classification time. Keyword Spam, Machine learning algorithm, Naïve Bayes, J48 Decision Tree, Pre-processing, Enron, TFIDF, Feature size.