Email is one of the most prevalent communication tools today, and solving the email overload problem is pressingly urgent. A good way to alleviate email overload is to automatically prioritize received messages f1ording to the priorities of each user. However, research on statistical learning methods for fully personalized email prioritization has been sparse due to privacy issues, since people are reluctant to share personal messages and priority judgments with the research community. It is therefore important to develop and evaluate personalized email prioritization methods under the assumption that only limited training examples can be available, and that the system can only have the personal email data of each user during the training and testing of the model for that user.
We focus on three aspects: 1) we investigate how to express the ordinal relations among the priority levels through classiﬁcation and regression. 2) we analyze personal social networks to capture user groups and to obtain rich features that represent the social roles from the viewpoint of a particular user. 3) We also developed a semi-supervised (transductive) learning algorithm that propagates importance labels from training examples to test examples through messages and user nodes in a personal email network. These methods together enable us to obtain both a better modeling priority and an enriched vector representation of each new email message.
Our contribution is as follows. First, we have successfully collected multiple users’ private email data with their ﬁne grained personal priority labels. Second, we apply and propose learning approaches from multi-type information such as text, and sender / recipients information. Third, to supplement additional information to sparse training data, we identify the importance of a contact and similar contacts from social networks. Fourth, we exploit a semi-supervised learning on the personal email networks. Finally, we conducted and completed systematic evaluations with respect to email prioritization, targeting the discovery of better modeling of email priorities. Through our suggested approaches, email prioritization alleviates email glut and should help our daily productivity.