Inexperienced users typically obtain one of three possi-interfaces (for example, Mosaic and Netscape) has led ble outcomes when they search for online information:
to an explosion in the amount of information being added.
They are buried under an information avalanche, they
In particular, the World Wide Web (WWW) provides an are unable to locate any useful information at all, or they exponentially growing amount and range of information find what they need in roughly the amount they need. through which people can browse. A few dozen biblio-Unfortunately, the latter outcome is the most rare. Unfamiliarity with search tactics creates difficulties for many graphic databases have mushroomed into several thouusers of online retrieval systems. When faced with poor sand databases of everything from full-text documents to results, even experienced searchers may use vocabulary movie clips. To give a feel for the magnitude of the incorrectly and often fail to reformulate their queries. Far growth, consider that 100,000 artifacts were Web-accessifrom being the answer to everyone's information ble in 1994. By 1995, 1.5 million artifacts were Webdreams, distributed sources of online information, i.e., the World Wide Web (WWW), compound the problem accessible (Eichmann, 1995), and today there are over and may often turn into an information nightmare. To 30,000,000 Web pages on 225,000 Web servers (Alta address this problem, intelligent online search assis- Vista, 1996). Demand for this information has also extants, or agents, are being developed for information ploded. For example, the WWW Worm search system retrieval applications. There are many approaches, both theoretical and implemented, to using intelligent softreceived 2,000,000 search requests per month in 1994 ware agents for information retrieval purposes. These (McBryan, 1994). However, the Alta Vista search system approaches range from desktop agents specialized for now received 10,000,000 search requests daily (Alta a single user to networks of agents used to collect data Vista, 1996).
from distributed information sources, including Web sites. This article presents an overview of intelligent soft-
In the early days of online information retrieval sysware agents in information retrieval, including an explatems, individuals met with search intermediaries who nation of agents and agent architectures, and presents were trained to use the online systems and were often several agent systems. We distinguish between agents knowledgeable about the information seeker's area of inas individual entities, whose properties and characteristerest. Through an interview process, the search intermetics we describe separately, and agent systems as collections of agents utilized for information retrieval tasks, diary would determine the individual's information need, which we discuss in terms of individual implementations. perform the actual searches, and send the results to the information seeker. Technology, in the form of personal computers and networks, now provides many people with * To whom all correspondence should be addressed. make effective use of this information from users' homes and desks. The research community could make a sig-