Intelligent Agents

What is an Intelligent Agent?

Intelligent agents go by various names such as software agents, softbots, autonomous agents, or simply, agents (Huhns & Singh, 1998). Russell and Norvig (1995, p. 33) defined an agent as "anything that can be viewed as perceiving its environment through sensors and acting upon that environment through effectors." This definition applies to both humans and artificial agents. One of the key concepts of intelligent agents is autonomous or intelligent behavior. Franklin and Graesser (1996) proposed what they called a mathematical style definition of an autonomous agent: "An autonomous agent is a system situated within and a part of an environment that senses that environment and acts on it, over time, in pursuit of its own agenda and so as to effect what it senses in the future." This definition is broad enough to include most intelligent agents, while allowing further restrictions in more specific types of agents. More recently, Andolsen (1999) suggested a more elaborated definition: "Intelligent agents are software applications that follow instructions given by their creators and which" learn" independently about the information they are programmed to gather. They save time and simplify the complex world of information management. An intelligent agent can anticipate the need for information, automate a complex process, take independent action to improve a process, or communicate with other agents to collaborate on tasks and share information."

Over the last 40 years, and particularly in the last decade, research on intelligent agents has spread over a wide spectrum of disciplines, from computer science, to psychology, management, economics, information systems, and social science. The phrase "intelligent agents" gained popularity in both the research community and the general public in 1994, when a number of important articles on agents were published in the first of several special issues of Communications of the ACM. In an influential paper, Agents that reduce work and information overload, Pattie Maes of MIT explored the potentials of personal and information agents (Maes, 1994). The excitement over intelligent agents research was evidenced in many publications as agent-based computing and was hailed as "the new revolution in software" (Ovum, 1994).

Taxonomy of Intelligent Agents

The terminology of intelligent agents has evolved over the years and there is still no definitive standard. By and large, software agents are named according to their main functions. Examples include news agents, e-mail agents, shopping agents, and search agents. The taxonomy of agents put forth by Franklin and Graesser (1996) provided an initial framework for intelligent agent classification.

Russell and Norvig (1995) assert that agent = architecture + program. The architecture refers to the computing device on which the program will operate. The central task of artificial intelligent research is then to design the program that "implements the agent mapping from percepts to actions." All intelligent agents share some humanlike characteristics. For example, they are autonomous, context sensitive, capable of learning and adapting, goal driven, possessing specialized knowledge, and communicating with people or other agents. It is not necessary, however, for all intelligent agents to have all of these characteristics. Table 1 provides a set of typical characteristics of intelligent agents (Andolsen, 1999; Feldman & Yu, 1999; Franklin & Graesser, 1996).

Table 1: Characteristics of Intelligent Agents.




Being able to exercise control over its own actions. Not only can agents take direct instructions from the user but can also accomplish tasks without intervention from a human user or other agents.


Agents should be able to learn and adapt to their external environment through interaction with information, objects, other agents (includes humans), or the Internet.


Each agent follows its own goal-oriented rules. But, it is often a member of multi-agent systems (MAS). Agents communicate, bargain, collaborate, and compete with other agents on behalf of their masters (users). Some MAS exhibit collective emerging behaviors that cannot be predicted from individual agent behaviors.


Agents are able to migrate themselves from one machine/system to another in a network, such as the World Wide Web in order to accomplish their assigned duties.


Agents do not simply act in response to the environment. They have built-in purposes and act in accordance with those purposes.


Agents continuously work to monitor their environment and update their knowledge base. Many of the tasks agents perform, such as information gathering, filtering, and customizing, require continuous operation.


Agents have the ability to communicate with people or other agents through protocols such as agent communication language (ACL).


Agents do not have feelings, emotions, subjectivity, or bias. Although this is changing, impartial agents may be necessary in certain applications.


Agents exhibit intelligent behavior such as reasoning, generalization, learning, environment awareness, dealing with uncertainty, using heuristics, and natural language processing.

Intelligent Enterprises of the 21st Century
Intelligent Enterprises of the 21st Century
ISBN: 1591401607
EAN: 2147483647
Year: 2003
Pages: 195 © 2008-2017.
If you may any questions please contact us: