Share the Ontology in XML-Based Trading Architectures

Recent e-commerce application activity involving the extensible markup language (XML) has led to a proliferation of XML-based standards and markup language proposals. Among them are several designed to support site-to-site Web automation that lean naturally toward the agent paradigm of distributed computation.

Although XML represents a major step forward in e-commerce technology, business-to-business trading partners should also recognize XML’s limitations. XML is not a cure-all for system interoperability, but a widely accepted foundation layer on which to build. Moreover, there are differing views on how to extend or complement XML to support agent-based e-commerce (see Glushko et al.’s "An XML Framework for Agent-based E-commerce" in this issue). This challenge is further complicated by debate over some fundamental questions: How should XML be extended to support the representation of business information? Should XML be enriched with tags reflecting higher-level concepts, especially business domains, such as standard business processes? How should foundation ontologies (from which higher-level content is composed) be defined? How can the numerous heterogeneous e-commerce frameworks (such as ICE, OBI, OTP, and XML/EDI) be unified to enable the expected low-friction market of the future? And will the future electronic marketplace be dominated by a series of commerce islands with trading groups isolated by the proprietary protocols and domain models with which their commerce agents interact?

Answers involve not only solving the related technology and intellectual challenges, but how to bring together the various communities of industrial standards developers. Each holds the essential elements of the overall solution. These communities, including EDI, Internet, knowledge engineering, and SGML, bring to the table subtly differing angles on the problem, including representation approaches associated with rich documents, publish/subscribe protocols, transactions, content syndication, and business semantics. To survive in this market, e-commerce component providers will have to support a number of different content formats and transaction frameworks, translating among them to achieve significant penetration. It appears that the main barrier to e-commerce lies in the need for applications to share information, not in the Internet’s reliability and security.

Due to the wide range of enterprise and e-commerce systems being deployed by businesses and the way these systems are variously configured, the problem is particularly acute among large electronic trading groups. E-commerce will increasingly focus on trans-enterprise communication, while the number of trading partners and sophistication of e-commerce applications also increase. The need to unite business models, processes, and representation formats is greater than ever, while expectations run ever higher. Although many companies have already begun to organize, standardize, and stabilize their digital services in order to create and maintain sustainable network relationships with their trading partners, they are doing so only in conjunction with their immediate trading partners. This relatively narrow focus can limit the return on investment possible from each of these initiatives.

A global environment. There is now a need for e-commerce participants to create a global environment providing significant interoperability between the systems used by all engaged. Such an environment can be achieved through improved semantics within Internet transactions and in networked service definitions. It will facilitate consistent behavior among participants in large trading networks or within complex virtual organizations. Many of the foundation concepts needed to achieve this consistent behavior have already been established through work on distributed problem solving, intelligent agents, and knowledge sharing, yet to date these technologies have had little effect on Internet-based commerce.

Agent-based systems to support the next generation of Internet commerce must adopt common ontologies if they are to interact without misunderstanding. For example, content can be defined to enable application interoperation as well as information synthesis. An e-commerce standard being developed by major PC vendors, resellers, and distributors has shown by practical example in the PC distribution chain that quite sophisticated representation issues can complicate even straightforward commerce scenarios. For example, the required catalog model includes the need to represent the topology of the parts comprising a PC product.

But to bring semantic order to the world of XML, we have to be clear about what we mean by "ontology." The term is often used to refer to a vocabulary, yet even the terms within a simple vocabulary can be prone to misinterpretation, particularly in combination, unless they have been chosen carefully. Consider some of the problems already apparent in the plethora of e-commerce standards that have emerged during the past few years. As new online trading environments are developed, the potential protocol mismatches between participants’ commerce platforms can become major inhibitors to achieving industrywide e-commerce solutions and delivering supply-chain and market-efficiency benefits. Realizing Web automation in such complex environments reopens many of the problems and issues the knowledge-sharing and intelligent-agent communities have been wrestling with in such initiatives as the shared design environment, or SHADE, and the advanced technology operations system, or ATOS, using ontologies to enable agents working on different problems to interoperate over networks.

XML as a representation is just too forgiving at the document type definition (DTD) stage at the expense of the information processing stage. However, steps are being taken in the right direction; an example is the definition of schema languages to enable consistent schema semantics in the definition of objects in XML (such as by the World-Wide Web Consortium reflecting proposals from a number of organizations).

Consistent schema semantics will certainly enable efficient e-commerce using predefined DTDs between fixed networks of trading partners. But to enable the full benefits of agent-based e-commerce—where agents act in an autonomous or semiautonomous way, comparing and contrasting products or suppliers and negotiating with other agents—participating agents have to communicate in terms of a detailed ontology of the business domain.

The challenge for technology vendors, e-commerce participants, and standards bodies is to capitalize on the experience available in the knowledge representation and distributed agent communities.

Veo Systems is pursuing a pragmatic approach to solving some of these issues through the Common Business Library, an extensible, public collection of business interface definitions and document templates. This library is being rationalized and further developed by the CommerceNet eCo Framework Working Group established last year and should provide a foundation for addressing many of the unanswered questions in agent-based e-commerce. Ontologies will play a key role.

Back to Top