External links

Full text: HTML, PDF

Information about this paper

Abstract

The system administrators of large organizations often receive a large number of e-mails from its users and a substantial amount of effort is devoted to reading and responding to these e-mails. The content of these messages can range from trivial technical questions to complex problem reports. Often these queries can be classified into specific categories, for example reports of a file-system that is full or requests to change the toner in a particular printer. In this project we have experimented with text-mining techniques and developed a tool for automatically classifying user e-mail queries in real-time and pseudo-automatically responding to these requests. Our experimental evaluations suggest that one cannot completely rely on a totally automatic tool for sorting and responding to incoming e-mail. However, it can be a resource-saving compliment to an existing toolset that can increase the support efficiency and quality.