Interface Agents that Learn: An investigation of Learning Issues in a Mail Agent Interface

Abstract

In recent years, interface agents have been developed to assist users with various applications. Some researchers have embedded machine learning techniques within agents to assist the user in performing various tasks, and to adapt to the user's changing requirements. With the increase in the volume of data on the Internet, agents to assist in handling this data have emerged. These agents monitor and learn from their users to identify topics of interest. One such agent, described here, has been developed to filter mail messages. The work described here examines the issues involved in constructing an autonomous interface agent which employs a learning component, and explores the use of two different learning techniques in this context.