A Novel Approach for E-mail Classification Using FastText

The upward trend of communication through emails has made the task of handling mails efficiently very vital. As the number of email users is growing considerably each day, the volume of mails to each user is also enlarging. It mostly requires an intelligent classification system for multi-folder categorization of emails to save time and manual labor. Several studies and experiments done till date have been proved to provide a great extent of automation to this classification task. However, multi-folder classification task has always been a little more challenging than others, especially with emails, because of the large number of the possible classes. This paper suggests an approach with machine learning for multi-class categorization of emails in a simple text classification method. We used our own dataset to build a text classifier with fastText, keeping different features of each class into consideration. The results from our proposed method were also compared to the performance of methods like Convolutional Neural Network with the same dataset and fastText was observed to perform better.