Electronic discussion group moderators' experiences with flame messages and implications for the design of a multi-modal adaptive content-sensitive filtering tool

• Authors: Richard H. C. Seabrook
• Venue: CHI '01 Extended Abstracts on Human Factors in Computing Systems
• Year: 2001

Flame messages pose a persistent and growing problem for discussion groups who rely on the Internet for their primary means of communication. Online communities maintaining public bulletin boards are especially susceptible to undesirable posts which can have a considerable effect on the membership and continuing discussion. Most such groups rely on a moderator or host to take appropriate action when flame messages appear. Filters applied to email headers have been partially successful identifying unwanted messages in electronic mailing lists. While message content provides a consistent determiner of what a message is about, not being able to predict the content of unwanted messages makes filtering on content difficult. The current study interviews list moderators to gain an understanding of current practices with respect to handling flame messages and their effect on continuing discussion. A filtering tool is proposed using a multi-modal filter technique and a prototype tested in the field. A panel of experts is used to evaluate the tool's interface.