In the second part of this Q&A interview with James Bryce Clark, director of standards development for OASIS, he talks about the future of semantic standards to make more intelligent use of the information organizations rely on for their business applications. Read part one
If security and other standards for SOA plumbing are solved as you've said, where will the focus be in the next year?
Those areas are happily solved and that leaves us free to move upward toward business information. Consider what we've been seeing lately. At OASIS, you see OpenDocument, which is about content organization. The Universal Business Language (UBL) is coming out in a stable 2.0 version. There are also elaborations of it to cover various business cases. We just spent time with the Danish government, which is fulfilling its regulatory requirement for procurement and taxation, to make all invoices of a certain class use UBL. The business cases are there. Also banks involved in money transfers are using the data harmonization model to convert large amounts of electronic transaction traffic into forms that can be read by anybody. There's a tremendous move now to harmonize this information, and more importantly to give it some knowledge representation. What's happening with knowledge representation?
Among the things that are happening now to bring KR, knowledge representation, into our field are all the taxonomy and ontology projects. There are also three business rules projects. Semantic markup people are doing great work. There are several candidates and it's hard to say who will win out because they have different models. But when our people on OASIS committees want to add semantic information to their models there are ways to do it. And it's very important to add that functionality. Also, a lot of the action in adding semantic content, in making business documents smarter, isn't necessarily happening through large complex academic exercises. There are a lot of fairly light methods for adding information that also seem to be gathering a lot of steam. Because not everybody who wants to make their business information smarter wants to go down the road of re-writing everything they have, and hiring four PhDs, and doing ontological research. Sometimes they just want to get a little more metadata or a little more organization. There are a number of really cool methods for doing that that are becoming mature now. What are those methods?
One of them that is happily at OASIS is Darwin Information Typing Architecture (DITA). Also there's BPEL, which is widely used. A lot of complex transaction engines and business rules engines are proprietary and have their own models. Nevertheless, they use BPEL for their exchange format. Vendors of systems that orchestrate and run transactions with their own engine that's not necessarily standardized, most of them support BPEL now. That's something I didn't expect to see. It's become an important exchange format. A customer who commits to the Foo Company's method of running their transactions, knows that if they ever want to switch out of Foo Company, the use of BPEL will help them move their data out in a standard format. Where will the first applications of semantic technology come?
With semantic content, there's more smart information out there. I'm particularly thinking of bodies of information like medical diagnostic classifications where there's a huge amount of data and it doesn't do you any good if it sits in files or in paper piles because you need to be able to query it and get at it and think about it and use it and dynamically grow it over time. Every year medicine learns more so we have to change and improve our classifications. This is an example where the world is rife not just for electronic information, but information expressed in electronic forms that is readily index-able, reusable, organize-able, so information can be interpolated. We need to go about that in a smart semantic way. What's the next step?
Some of the experiments we're seeing now are using DITA, XML Topic Maps, RDF (Research Description Framework). There are ways of using those methodologies to make data live, smart and useful. It's a whole generation ahead of having static electronic files. What are standards bodies like OASIS doing in the semantic area?
There are a couple of projects that directly create semantic methods, as well as existing standards like Open Document and ebXML that are using them. You can use these, for example, where you want to consume ontological information because you're validating a transaction on the fly. We want an ontological statement of which transactions we're willing to do and not willing in do. There's a great hunger to use these methodologies. In OASIS, there's the semantic execution environment (SEE) technical committee working out of Austria and Asia to create a simple, lightweight sub-ontological way to organize semantic Web services. In other words, we want to do more than just offer services. We want to actually go out there into WSDL space. There is such a thing as WSDL space. If you imagine yourself as a consumer running around the Internet, hopping from node to node looking for service descriptions you're in WSDL space.
But WSDL space isn't too well organized. The SEE people are describing a framework for how to make sense out of that and use it, and what kind of information should be reliably available. There's also a lot of hope in the management of Web services, led by our Web Services Distributed Management (WSDM) committee. Also, the DITA group is doing semantic content. They are offering a method to markup what we think of as documents in electronic form to provide a self-organizing framework for their meaning that doesn't require a lot from the user. This is an instance of lightweight layers of semantic action. There are also all the rules projects happening at a variety of places including some academic institutions, government research, along with W3C, OMG. There are a lot of business rules projects that are going to pop into something concrete fairly soon. That's also about semantic content of the message because rules just reduce down the operands to a symbolic set that you work on instead of the raw content.
We have started to see some OASIS standards leverage RDF. Of course, RDF is from W3C. It's early days for RDF. It seems to work for some people in some cases and not others. But it's significant that they're seeing technologists who are not trying to solve semantic problems voluntarily, looking at RDF and saying "I can see stuff I can do with this" and including it in their architecture. For me that's a turning point. I've been waiting to see how it comes out. In 2006, we started to see some important experimentation with it.