Many agent-based systems are available in the research literature to support one or more activities such as auctions, automated negotiations, catalog search, supply chain management, etc. (Blake, 2002; Chari, 2003; Chavez & Maes, 1996; Faratin et al., 1998; Fox et al., 1993; Guttman et al., 1999; Shen et al., 2000; Zeng & Sycara, 1996). However, these research initiatives have not focused on a comprehensive framework for B2B commerce using agents-technology.
Research on using agent-based technologies for building e-exchanges and B2B systems is limited. A notable research initiative that is partly supported by the industry is the CASBA project (Kraft et al., 2000). This project provides a multi-agent framework to support e-commerce. Various functions such as need identification, product brokering, merchant brokering, negotiation and payment/delivery are supported. On the commercial front, the ViniSyndicate Supplier integration solution from ViniMaya (www.vinimaya.com) is an agent-based software that provides information on product, price and availability in real-time from supplier websites, e-commerce systems and Web-accessible sources. Friction Commerce (www.frictionless.com) has the Frictionless Sourcing 3.0 platform to support multiple activities in e-commerce. Details of the underlying framework and architecture of commercial systems are not available in the open literature.
The distinct contributions of the multi-agent architecture presented in this chapter with respect to the CASBA project are as follows:
A more comprehensive architecture that covers all the participants in an e-exchange such as buyers, sellers, financing, shipping, and warranty companies.
A service oriented architecture that enables the use of Web services and agents technology to implement some services.
The notion of object models to enable flexible configuration of services.
Support for multiple enabling technologies by which exchange participants could interact with the e-exchange.
There are few research issues pertaining to the proposed architecture that need to be addressed in the future. First, in order to support a flexible architecture, a representation scheme for various object models of services is needed that allows services to be configured easily. Similarly, a representation scheme is required to represent the data, logic, and workflow rules of an agent. This is needed to enable the participants to customize their agents in the e-exchange in real-time. In the general area of B2B commerce and agents technology, research is needed to enable agents to infer patterns from seemingly unrelated events that occur in the market place. In this regard, data mining techniques hold lot of promise (Fayyad et al., 1996). Research is also needed to create a shared universal ontology for creating a semantic Web (Berners-Lee et al., 2001). This would lead agents not only to gather product, service, buyer, seller, and service provider information from the Web, but would also enable agents to infer the semantics necessary to do intelligent processing.
In the future, with increased globalization, efficient agent-based B2B infrastructure is likely to be the norm. Value will be derived more from timely information, autonomous processing of information and instantaneous decisions. This will enable enterprises to create more creative business models and thrive.