On Agent-Mediated Electronic Commerce

1 INTRODUCTION

ELECTRONIC commerce (e-commerce) is increasingly assuming

a pivotal role in many organizations. It offers

opportunities to significantly improve (make faster, cheaper,

more personalized, and/or more agile) the way that

businesses interact with both their customers and their

suppliers. However, in order to harness the full potential of

this new mode of commerce, a broad range of social, legal,

and technical issues need to be addressed. These issues

relate to things such as security, trust, payment mechanisms,

advertising, logistics, and back office management

[152], [45], [138], [169].

Even more fundamental than these issues, however, is

the very nature of the various actors that are involved in ecommerce

transactions. In most current (first generation) ecommerce

applications, the buyers are generally humans

who typically browse through a catalog of well-defined

commodities (e.g., flights, books, compact discs, computer

components) and make (fixed price) purchases (often by

means of a credit card transaction). However, this modus

operandi is only scratching the surface of what is possible.

By increasing the degree and the sophistication of the

automation, on both the buyer's and the seller's side,

commerce becomes much more dynamic, personalized, and

context sensitive. These changes can be of benefit to both

the buyers and the sellers. From the buyer's perspective, it

is desirable to have software that could crawl all the

available outlets to find the most suitable one for purchasing

the chosen good (e.g., the one that offers the cheapest

price, the highest quality, or the fastest delivery time) and

that could then go through the process of actually

purchasing the good, paying for it, and arranging delivery

at an appropriate time. From a seller's perspective, it is

desirable to have software that could vary its offering (in

terms of price, quality, warranty, and so on) depending on:

the customer it is dealing with (e.g., offering discounts or

special offers to particular target groups), what its competitors

are doing (e.g., continuously monitoring their prices

and making sure its own price is competitive), and the

current state of its business (e.g., if it has plenty of a

particular item in stock, it may be appropriate to reduce the

price in order to try and increase demand).

To achieve this degree of automation and move to

second generation e-commerce1 applications, we believe

that a new model of software is needed. This model is based

upon the notion of interacting agents [64] (hence, the term

"agent-mediated electronic commerce"). Against this background,

the aim of this paper is to motivate the use of

agents in e-commerce, to highlight the roles that agents can

and are fulfilling in this domain, and to review some of the

key technologies that underpin this vision. First, however,

we define exactly what we mean by the terms "agent" and

"electronic commerce."

1.1 Interacting Agents

An agent is a software program that acts flexibly on behalf of

its owner to achieve particular objectives [64].2 To do this, the

software must exhibit the following properties [162]:

. it needs to be autonomous: capable of making

decisions about what actions to take without

constantly referring back to its user;

. it needs to be reactive: able to respond appropriately

to the prevailing circumstances in dynamic and

unpredictable environments;

IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING, VOL. 15, NO. 4, JULY/AUGUST 2003 985

. M. He and N.R. Jennings are with the Department of Electronics and

Computer Science, University of Southampton, Southampton SO17 1BJ,

United Kingdom. E-mail: {mh00r, nrj}@ecs.soton.ac.uk.

. H.-F. Leung is with the Department of Computer Science and Engineering,

The Chinese University of Hong Kong, Shatin, Hong Kong SAR, P.R.

China. E-mail: lhf@cse.cuhk.edu.hk.

Manuscript received 7 Feb. 2002; revised 28 Aug. 2002; accepted 25 Nov.

2002.

For information on obtaining reprints of this article, please send e-mail to:

tkde@computer.org, and reference IEEECS Log Number 115866.

1. Second generation systems are characterized here as having a greater

degree