The system provides an efficient solution for matching of orders/enquiries with products and/or services available for sale over the Internet or an internal network. It automates the process of searching for required goods and negotiating of the purchase.

It enables the development of customer-specific e-commerce applications relating to either a single virtual shop or a virtual department store containing partitions for a number of sellers.

MagentA Home

e-Commerce System has two key subsystems: one for trading and the other for managing dialogues with users. An additional subsystem is available, known as the News Agency, which can also play an important part in certain business situations.

A result of matching,                                       negotiated by agents

Trading

The subsystem for trading contains agents assigned to buyers, sellers, orders and goods, which dynamically interact with each other in accordance with a real market situation. For example, a negotiation can be conducted with a buyer about the price of goods and delivery using a range of possible criteria such as: the buyer is a preferred customer; express delivery is requested; the warehouse is overstocked; a specified volume of sales; and desirable profit levels . The system also supports pro-active sales promotion. Several buyer agents (potential competitors) can combine their orders to obtain a bigger discount, substituting co-operation for competition. A similar situation can occur in cross selling, when additional, related goods and/or services are offered to a buyer for a discount. Receiving a commission for each item sold, agents make money and offer goods at reduced prices, adapting their business practices to the market situation.

Dialogue Management

The subsystem for dialogue management generates HTML pages displaying the results of agent negotiations. Each part of an HTML page, including a picture of the day, news reports and perhaps an advertising banner, is assigned its own agent, to respond to requests from site visitors. If a buyer, who usually buys a particular product, logs on the system, the banner for the preferred product proposes to all other banners to temporarily leave the screen accepting compensation. If the likelihood of selling their products is low, other banners would leave the screen allowing the prime contender to show, for instance, attractive screen-size pictures of its product, information about product options and available discounts. If the customer buys the product, then the profit is shared among agents; if the deal does not go through, the agent that failed to sell pays a penalty. Next time the agent will have less money to make a similar offer to other agents. Thus banners of poorly sold products gradually lose their money, stop competing actively and finally 'leave' the system. This mechanism enables dynamic site evolution aimed at supporting and selling more popular and profitable product lines.

At any time, system users can enter new products into the application ontology without interrupting the system operation.