Masterarbeit Johannes Roith

(Diese Arbeit wurde extern in Stanford verfasst, daher keine Kickoff-Folien verfügbar /Marin Zec)

Abstract

Email has made it possible to initiate many parallel but independently progressing email conversations with a large number of recipients. Examples include sending newsletters, marketing mails and surveys or replying to customer support requests. Private users also use email to organize events, to apply for jobs, find apartments or to request feedback from friends or colleagues. Performing these tasks manually with conventional email clients is difficult: Requesting and responding to feedback, for example, can result in hundreds of similar conversations that need to be organized. Existing email services can deliver large amounts of personalized emails, but offer limited support for organizing responses and are not well integrated into existing webmail applications. As a result, users may avoid specialized tools altogether or rely on impersonal but automated surveys.

This thesis explores how to integrate a scalable mail merge system that is capable of dealing with personalized conversations into the existing workflow of webmail clients. By responding to individual emails with generalized responses, users build a repository of reusable templates, which are organized automatically and can by recycled in new conversations later. Valet crowdsourcing techniques are used to reliably extract information from incoming emails. By selecting recipients based on the extracted metadata, users implicitly create rules which can be used to deliver the messages to conversations in a similar state in the future.

The system is evaluated in a within-subjects experiment and compared to conventional systems for sending large amounts of emails. The results suggest that new users can organize even small events more efficiently than using conventional tools.