SchedMail: Sender-Assisted Message Delivery Scheduling to Reduce Time-Fragmentation
Although early efforts aimed at dealing with large amounts of emails focused on filtering out spam, there is growing interest in prioritizing non-spam emails, with the objective of reducing information overload and time fragmentation experienced by recipients. However, most existing approaches place the burden of classifying emails exclusively on the recipients' side, either directly or through recipients' email service mechanisms. This disregards the fact that senders typically know more about the nature of the contents of outgoing messages before the messages are read by recipients. This thesis presents mechanisms collectively called SchedMail which can be added to popular email clients, to shift a part of the user efforts and computational resources required for email prioritization to the senders' side. Particularly, senders declare the urgency of their messages, and recipients specify policies about when different types of messages should be delivered. Recipients also judge the accuracy of sender-side urgency, which becomes the basis for learned reputations of senders; these reputations are then used to interpret urgency declarations from the recipients' perspectives. In order to experimentally evaluate the proposed mechanisms, a proof-of-concept prototype was implemented based on a popular open source email client K-9 Mail. By comparing the amount of email interruptions experienced by recipients, with and without SchedMail, the thesis concludes that SchedMail can effectively reduce recipients' time fragmentation, without placing demands on email protocols or adding significant computational overhead.
SchedMail, Sender-Assisted Email Prioritization, Email Delivery Scheduling
Master of Science (M.Sc.)