How does one efficiently sift through all of this information?
Through the continued development and improvement of intelligent agents, it is likely that the above dilemma faced by most Internet users will be either eliminated or at the least reduced. Agents may be classified in the following manner (Millman, 1998):
Retrievers—This type of agent is also called a "pull agent." These are "keyword" driven programs that are typically dispatched by the end user and programmed to seek out and return specific information.
Watchers—This type of agent is called a "push agent" since it "watches" for a specialized event and notifies the user.
Helpers—This type of agent is simplistic and is used in computer maintenance and diagnostic applications. As such it is intended to reduce the level of human involvement.
Shoppers—This type of agent "learns" user preferences and then searches the Web to link interested buyers with sellers.
Amazon.com has probably gained the most notoriety for its commercial use of agents and the ability of its web site to monitor user preferences. Amazon has, as a result, accumulated a large amount of consumer information as well as the ability to rapidly analyze and predict consumer behavior. It is in the position to market this information to its strategic partners which brings up the issue of privacy and where is the line drawn between privacy and further exploitation of the Internet by sharing its acquired consumer information. As such, Amazon looks at its information base as a revenue source (Hoy, 2003).
The value of an intelligent agent depends on the time it saves the user and the quality of the information that has been obtained. According to Millman (1998), the following represent characteristics of an agent:
Size—Should be between 4Kb and 10Kb to promote mobility.