EIEME : email information extraction made easy : an email summarization and categorization system using semantic technologies

Abstract

People around the world use email on a day to day basis to communicate and send out information to colleagues, friends and family. The popularity of email has given rise to various problems involving email congestion and information overload people cannot retrieve the required information efficiently because inboxes are being jam-packed with email messages, There has been a lot of research and investigations regarding automatic email classification into folders. Nevertheless few systems try to look at the problem from a user's perspective, where the user is mostly concerned with the retrieval of the requested information from an email message as efficiently as possible rather on how email is categorized internally. This project presents ELE.ME (Email Information Extraction Made Easy), a system which looks at the email classification problem from a user's point of view It retrieves and processes email messages from a specific corpus and extracts certain particular topics from each email. The system creates a semantic network using all the topics retrieved and therefore provides a semantic relationship amongst the email messages. It then provides the user with a sophisticated search system, where these topics can be searched for and the network can be thoroughly queried. The user is allowed to parse through the network of topics until the desired topic (statement) is found. From every particular statement, the email message(s) from which that statement was extracted can be retrieved. The system also offers summarization functionality in order to help the user retrieve the desired information faster. Evaluation and testing are conducted on the Enron Corpus using seven specifically chosen users who possess a large number of messages categorized in folders. The system's precision, recall and accuracy are established and compared to other classic, email classifiers.