In the current information overloaded atmosphere with an explosive growth of textual data, one can find it a challenging task to keep all the ducks in a row. This has resulted in an emergence of many Text Classification algorithms. Text classification is a process of categorizing data in pre-defined categories based on Topics or Genre. It is used - to classify named entities, Twitter and newspaper feeds, medical repository and Email. In this digital era of communication, Electronic mail is an important, and a popular means of communication. An Email inbox is often rife with different messages ranging from High Importance to Low to spam. In order to not lose sight of important emails, it is necessary to organize the emails in proper categories. In my paper, I will be presenting an implementation approach, involving self-learning by creation of pre-built dataset using user preferences and background knowledge for Email categorization.