In the world of Internet today, huge amount of data is transferred between computers in the form of emails. Consequently, it is getting difficult to sort the important emails manually from the unimportant ones. Email classification has been extensively studied and researched in the past but most of the research has been in the field of spam detection and filtering. This paper focuses on classifying emails into custom folders that are relevant to the user. We have used two different approaches here—Naïve Bayes classifier and k-nearest neighbors algorithm. The Naïve Bayes classifier is based on a probabilistic model, while the k-nearest neighbors algorithm is based on a similarity measure with the training emails. We propose the method of using these two approaches in email classification, analyze the performance of these algorithms, and compare their results. Then, we propose some future work for further optimization and better efficiency.
A technical paper of this project was published in Advances in Intelligent Systems and Computing (AISC, volume 564) series of Springer 2017