In my first Thinking XML column, I introduced the idea of semantic transparency and its importance to XML-related developments. Because semantic transparency is so important, there has been a flurry of activity in the area lately -- more than I could cover in one installment. In this installment, I introduce some of the emerging players in XML and semantics. But first, I'll cover an interesting play by the old guard which I omitted from the first installment.
The Implementation Guide Mark Up (IgML) working group is an effort by a group of electronic-data interchange (EDI) vendors to represent EDI implementation guidelines and standards in XML format. They are developing a DTD (document type description) for this representation, with the goal of providing a high degree of structure to the normative text and directing the implementation path to EDI for maximum interoperability.
While IgML does not itself provide a framework for semantic transparency, it will provide a useful tool for those implementing XML business-to-business systems that either work with EDI or just take advantage of the semantic infrastructure provided by EDI standards.
One of the pioneering players in the SGML and the XML games has been the Organization for the Advancement of Structured Information Standards (OASIS). You might recognize this organization as the host of Robin Cover's excellent compendium of XML news and notes, and more recently, as the host of a variety of technical committees for open investigation into XML technology. In 1999, OASIS teamed up with the United Nations Centre for Trade Facilitation and Electronic Business (UN/CEFACT), a key organization in the development of EDI, to form the electronic business XML (ebXML) initiative.
The goal of ebXML is to develop infrastructure and practices that use XML to facilitate trade between organizations regardless of their size, chosen tools, or geographical location. The effort has operated on an aggressive timetable, perhaps mindful of how long it took EDI to develop. The first artifacts, including an overall architecture document, began emerging late last year.
The portion of ebXML most related to semantic transparency of XML exchange is the Registry Service, described in the draft Registry Information Model and Registry Services specifications, which deal with the management and exchange of object-oriented entities. These registries aim at a higher architectural level than the International Organization for Standards Basic Semantic Registry (ISO BSR), which I covered in my first column. They provide an object model for business information that uses the Uniform Modeling Language (UML), and add metadata and design features not unlike the Object Management Group's (OMG's) meta-object specifications.
They also define an XML vocabulary for querying objects by metadata, and of most interest, an XML representation for instances of compliant object models. Documents following this representation can be exchanged as trade transactions between partners. The Simple Object Access Protocol (SOAP) was recently adopted as the transport protocol for such transactions.
The ebXML technical architecture provides for three phases of ebXML systems development: the implementation phase, the discovery and retrieval phase, and the run-time phase. Of the discovery and retrieval phase, the specification says: "This is the phase where Trading Partners discover the meaning of business information being requested by other Trading Partners." The intention is that this "meaning" is provided by the metadata models set up in ebXML registries.
One complaint against ebXML's well-meaning efforts to elucidate business transactions is that it reinvents many wheels; it's unclear whether this is for political or technical reasons. After many cries of political maneuvering behind its earlier intentions to develop a messaging transport protocol specifically for ebXML, ebXML.org recently made the decision to adopt SOAP in ebXML. However, ebXML still seems to ignore many facilities that are already available using RDF, XMI, and other XML-based metadata specifications. And there is little evidence that ebXML takes advantage of the ISO BSR work.
Another complaint is that the registry's approach comes uncomfortably close to a common, and erroneous, tendency in second-generation XML initiatives: to import concepts wholesale from object-oriented systems into XML generics. This tendency has the potential to do severe damage to XML's extensibility and expressiveness, especially where semi-structured data is important. Perhaps the most dangerous manifestation of this aberration is the data-typing and inheritance features of W3C's XML Schema specification. Initiatives such as ebXML and W3C Schema would do best to focus on core XML needs and leave object-oriented architecture to the OMG and similar bodies.
The CommerceNet organization is working to coordinate industry, government, and academic efforts to advance practices and technologies that enable electronic commerce through activities such as the eCo Framework, a project to develop open registries for interchange of semantics. The ambition is that such registries will allow a much greater degree of automation in electronic commerce that ultimately will lead to procurement processes completely driven by autonomous agents.
eCo registries would provide access to formal data models and the element-type and attribute definitions provided by a variety of XML business transaction specifications. The eCo working group boasts involvement by key figures in the development of such XML format specifications as Open Buying on the Internet (OBI), RosettaNet, and XML/EDI. Some of these other efforts already are in courtship with ebXML, and it's possible that a strong relationship will develop between the eCo and ebXML work.
Unfortunately, so far the only resource available to the developer with itchy fingers is a handful of somewhat lightweight demonstrations offered at the eCo site.
XML might just be the foremost generator of TLAs (uh, that's three-letter acronyms) and business consortia. One consortium, RosettaNet, set out to define processes for standardized supply chain management.
RosettaNet plans to deliver a "master dictionary" that defines terms used in supply chain interactions, as well as business processes and exchange protocols that make up the framework for such interactions. Naturally, the dictionary aspect interests me the most from a knowledge-management perspective. But the combination of dictionary and protocol is the basis for what RosettaNet calls Partner Interface Process(PIP). Each PIP is a specification for aligning a particular process between business partners, using XML messages. A complete RosettaNet-enabled supply chain can consist of dozens or even hundreds of PIPs.
RosettaNet is probably the farthest along of the XML generation of semantic registry and interchange initiatives, and already products such as BEA WebLogic Collaborate include implementations. In February 2001 a gaggle of speakers gathered at RosettaNet's EConcert Readiness Day to claim successful use of PIPs in their particular industries.
This maturity equates to goodies for developers. The RosettaNet site already offers three dictionaries: RosettaNet Business Dictionary, RosettaNet IT Technical Dictionary, and RosettaNet EC Technical Dictionary. These provide definitions, data types, and representations restrictions for hundreds of terms (for example "DigitalCertificateSignature") in HTML tables and XML+DTD. This is a treasure trove, especially valuable for its polish, which far exceeds that of the experimental ISO BSR listings mentioned in the last column.
Rounding out this month's discussion of semantics is RDF, perhaps the most important tool for semantic reference and other tasks related to managing the knowledge latent in XML documents. You'll hear a lot about RDF as this column evolves, so I'll just mention here that RDF's ability to efficiently manage XML-related metadata makes it an important tool for modeling relationships between terms, and relationships between the abstract concepts that underlie the terms. In fact, RDF has been called an "assembly language for knowledge" because of the flexibility it allows in modeling relationships, definitions, and properties.
I've worked on several projects where semantic transparency was important, including one in which a client used RDF to implement a framework for synonyms and basic translations to accommodate free-form customer queries. I have always been quite impressed at how easy it is to set up such systems using RDF, and by the resulting performance.
Therefore, I was quite disappointed to see so many of the XML process initiatives reinvent the RDF wheel. But as I noted in a developerWorks article the Web Services Description Language (WSDL), all too many XML applications overlook RDF's strengths. Hopefully, with the steady growth of RDF tools -- and success stories such as RDF Site Summary (RSS) -- this situation will change.
Now that I've outlined the importance of semantic frameworks as a layer above XML, future columns will move on to examining practical ways to manage the knowledge represented by these high-level frameworks. The next article will discuss the use of RDF to develop inexpensive search and reference systems for XML data repositories.
- Participate in the discussion forum.
meets semantics: The reality, the first installment of this column, defines semantic transparency and metadata, and introduces some of the players.
- ebXML is an international effort driven
by UN/CEFACT and OASIS to provide trade facilitation through open XML standards
for information exchange.
- UN/CEFACT is an agency of the United
Nations dealing with process and technology for trade facilitation, particularly
with computer technology.
- OASIS is an international consortium
promoting electronic commerce solutions that allow different information
systems to operate with each other.
- The XML Cover Pages section
on ebXML is the definitive source for news and resources related to
the initiative. Also see the RosettaNet section.
- RosettaNet is a consortium of organizations
interested in using XML to improve the automation of supply chain processes.
The RosettaNet home page has free downloads of several dictionaries
of supply chain terms in HTML and XML format.
introduction to RDF, by Uche Ogbuji describes Resource Description Framework (RDF), developed by the W3C for Web-based metadata, using XML as an interchange syntax.
WSDL with RDF, by Uche Ogbuji, illustrates the use of RDF to represent
descriptions of Web-based services.
- James Lewin's An introduction to RSS news feeds describes an application of RDF Site Summaries, with sample code that uses the Perl XML::RSS module.
- Check out Thinking XML's previous columns.
Uche Ogbuji is a consultant and co-founder of Fourthought Inc., a software vendor and consultancy specializing in XML solutions for enterprise knowledge management applications. Fourthought develops 4Suite, an open-source platform for XML middleware. Mr. Ogbuji is a computer engineer and writer born in Nigeria, living and working in Boulder, Colorado, USA.