Brought to you by EarthWeb
IT Library Logo

Click Here!
Click Here!

Search the site:
 
EXPERT SEARCH -----
Programming Languages
Databases
Security
Web Services
Network Services
Middleware
Components
Operating Systems
User Interfaces
Groupware & Collaboration
Content Management
Productivity Applications
Hardware
Fun & Games

EarthWeb Direct EarthWeb Direct Fatbrain Auctions Support Source Answers

EarthWeb sites
Crossnodes
Datamation
Developer.com
DICE
EarthWeb.com
EarthWeb Direct
ERP Hub
Gamelan
GoCertify.com
HTMLGoodies
Intranet Journal
IT Knowledge
IT Library
JavaGoodies
JARS
JavaScripts.com
open source IT
RoadCoders
Y2K Info

Previous Table of Contents Next


2. HISTORICAL ACCOUNT/BACKGROUND

The agent metaphor came to the fore in the early 1990s. In Artificial Intelligence there was a recognition that an alternative to working on complex problems such as factory scheduling and medical diagnosis was to focus on simpler tasks such as personal calendar management and automatic filtering of electronic mail and online information sources. These simpler tasks were typically those that a user would consider delegating to a personal assistant. Since these tasks were intrinsically simpler, they were more tractable to design and implement. These applications were termed "agents" because they undertook to perform routine tasks for users.

The origins of intelligent agents come primarily from artificial intelligence research work done in the 1970s and 1980s in areas such as planning, robotics, knowledge-based systems, and learning techniques. The significant applied R&D work done on autonomous vehicle navigation for the DoD in the 1980s also explored many of the concepts now appearing in intelligent agents. Probably the key transition from the AI research mentioned above and intelligent agents involved the concept of a softbot, i.e., a transition to the concept of autonomously functioning software robots rather than physical industrial robots or robot vehicles that could roam a battlefield.

Gradually, the agent metaphor has extended to applications outside Artificial Intelligence. Today, the concept of an agent has become somewhat independent of its roots in Artificial Intelligence. If an application performs an intelligent task automatically that is of the sort that one could imagine a person performing manually, then the application can be called an agent. So, an agent is now a label for an application belonging to a certain class -- the class of applications that automate intelligent tasks people could perform manually.

There is no one way to build agents. A multitude of technologies are employed. There is no one domain where agents are deployed. Agents can be found in virtually all domains. The exact nature of the computation performed by an agent application is determined by the task the agent is meant to perform. The task specification also determines what underlying technology is emphasized in the design of an agent application for that task.

The pervasive nature of the Internet has helped fuel the growth of agent applications. Vast amounts of information were becoming available online, and users could simply not keep up with the information explosion. Therefore, Internet agents were devised to help users to benefit from the available information. Now that the Internet is moving into the realm of electronic commerce, agents to automatically negotiate and obtain services from electronic storefronts are being developed.

The success of database technology has led to explosive growth in the number of databases deployed to support businesses, and huge quantities of data are accumulating in these databases. It is impossible to analyze manually such large quantities of data. Therefore, data mining agents are emerging to automatically examine large data sets to derive value for businesses.

The entertainment software industry is one of the fastest growing segments of the commercial software industry, with revenues in excess of $1 billion annually. One way to design more engaging games is to inject intelligent decision-making into the software. This gives rise to agents embedded in software games. These agents exhibit interesting behaviors and sometimes adapt in response to interactions with users.

Business process automation is proving to be a profitable focus for agent technology. Many steps in a business process can be automated by using agents. For example, consider an order filling process whose goal is to order replenishments for a certain product at the best possible price. The first step is to monitor the inventory level for the product and generate a reordering signal when appropriate. Then, all the possible sourcing alternatives need to be queried to determine the best pricing and availability for the product. An agent can be set up for each of the steps of the order filling process. This way, an entire business process can be automated. Such process automation is a powerful means to reduce the operating cost structure of a business. In addition to reducing cost, automation increases process efficiency.

It is well known that maintenance costs constitute the bulk of the expenditure incurred on computer systems. Intelligent agent technology has provided a means to reduce these costs.

Two types of agents have been deployed in the customer support area. First, self-help agents have been deployed that carry out an intelligent dialog with users to provide solutions to support problems with computer hardware and software systems. Second, self-repair agents have been packaged with computer systems to identify various types of equipment failures and either rectify the errors through software actions or log calls for onsite service or remote diagnosis.

In the next section, we discuss each of the primary types of agent applications that have been successfully deployed.


Previous Table of Contents Next

footer nav
Use of this site is subject certain Terms & Conditions.
Copyright (c) 1996-1999 EarthWeb, Inc.. All rights reserved. Reproduction in whole or in part in any form or medium without express written permission of EarthWeb is prohibited. Please read our privacy policy for details.