Introduction

Abstract

It is evident that cyberspace will become more and more commercialized in near future¹. Reasonably, trading information agents have to be equiped with sophisticated methods for making economically rational decisions [666, 80]. A personalized rational information agent may make purchases for his customer up to a preauthorized limit, filter information and solicitation from vendors, dynamically trades commodities and components within business-to-business digital markets, and increase the level of trust in his actions gradually over time. Most current tools for developing agents for electronic commerce (EC) in the Web fall into the categories conversational interfaces and profiling, collaborative filtering, comparison-shopping, and agent-based marketplaces [304, 264].

According to the recent Forrester report, 3.2 trillion USD e-commerce sales are expected worldwide by 2003. As one example, currently the highest earning Web site in the UK in terms of advertisement revenue is the site of the newspaper The Financial Times, bringing in more than 1.7 million USD in the six months up to August 1998. The site has 1.2 million registered users spread across 230 countries [305].