Automatic classification of E-mail messages by message type
A D May · Journal of the American Society for Information Science · 1997
This article describes a system that automatically classifies E-mail messages in the HUMANIST electronic discussion group into one of four classes, Question, Response, Announcement, or Administrative. A total of 1,372 messages were analyzed. The automatic classification of a message was based on string matching between a message text and predefined string sets for each of the message types. The system's automated ability to accurately classify a message was compared against manually assigned codes. The Cohen's Kappa of .55 suggested that there was a statistical agreement between the automatic and manually assigned codes.