Personalized Email Prioritization Using Multi-attribute and Multiclassification Data Mining Techniques

Abstract

Abstract The increase in the volume of electronic email communication that is received daily by an individual is becoming alarming and it threatens to cause a state of “email-overload” where the volume of messages exceeds individual capacity to process. With email being one of the most efficient and effective mode of communication that is widely used among business personnel and organizations, there is need to pay apt attention to the serious problem of email information overload that pose serious productivity challenges for busy professionals and executives. This necessitated the adoption of Data mining techniques to develop a model for prioritizing email using a multiattribute and multi-classification algorithm for the automatic classification of mails into predefined categories while eliminating the problem of manual labeling or annotation from users (an approach that is tedious and time consuming for users) in previous research work [1]. This study was introduced to automatically classify and prioritize email messages into folder structures, in a declining order of importance according to the priorities of each user’s email inbox content, without manual labeling or annotation of email categories from users. This model extends the application of K-means, Hierarchical Clustering and SVM classifier to the domain of email prioritization. The model developed, when used, eliminates the traditional manual labeling/annotation and method of triaging through a large volume of incoming email in no particular order and introduces a well-structured and organized hierarchy and priority level in each user personalized email categories.