"To win the battle for public opinion in America, you have to present yourself as a victim; that is how you win the heart of Americans. He who apologizes for using force loses.."
With supply chains stretching to China, customers as likely in Paris as Peoria, outsourcing operations in India and development in Russia, the reality of today's economy is that we must be global to succeed. But how do we succeed at being global? Here are three areas where most American companies need to get better, if they are going to succeed globally.
"Exploring the New World a thousand years ago, a Viking woman gave birth to what is likely the first European-American baby. The discovery of the house the family built upon their return to Iceland has scholars rethinking the Norse sagas."
The EU struck at the commercial backbone of Microsoft’s global software empire today (12/22/04) – forcing the company for the first time to strip its Windows platform of a popular feature to give competitors a better shot at success.
An EU court ruled Microsoft must immediately divulge some trade secrets to competitors and produce a version of its flagship Windows operating system without the program that plays music and video.
Dean Allemang:(Mr. Allemang is a senior consultant for TopQuadrant.) The big vision of semantic interoperability is that you have all sorts of different things talking to each other. The obvious things would be different databases. We'd also like to have integration with perhaps some office devices or mobile devices—a little radio or a telephone, and some of the more zany pictures of semantic integration, like toasters and refrigerators. Today we'll talk about enterprise solutions.
Semantic integration is here today. Yahoo has a portal called Yahoo Finance, where you can say, Here are all my bank accounts, credit cards, credit union. They have a list of financial institutions that goes on for pages. You pluck out the ones that you subscribe to, you give it your login information for all these things, and it provides you an integrated portal of your financial picture. It will chart your net changes over time. It will do analyses; it's integrating all these things together, and this is available today. And they promise you your privacy.
mySimon crawls websites and finds all the people who are providing [what you're shopping for] and gives you a comparative rating for suppliers and a bunch of information so you can comparison shop. This is a federated search over sources from all over the web.
How does interoperability work? I'm going to say there are three approaches: Hard-coding—what Yahoo and mySimon are doing—you do once for each system. Data standards: You get a group of people together, a standards organization. STEP is a good example, ebXML, RosettaNet, these sorts of things. Get everyone to agree. People say, I am RosettaNet compliant, and now you can all communicate. Then the extreme is the semantic web, and I'm going to stick my neck out and say that the semantic web is really a different way of doing this.
-- ADVERTISEMENT --
Some examples: You go to a website to find the nearest store. They send you off to another website. This is a hard-coded way of doing things in a certain format. Integration standards, where we have things like RosettaNet, SUMO, ebXML—why is it that standards aren't a good enough way? The famous quote, "Standards are great; you have so many to choose from." If you're doing one standard and I'm doing another, we don't gain anything from having joined into the standardization effort. I asked a fellow from STEP, "I want to add a new field to something inside of STEP; what do I have to do? How long will it be?" If I'm really, really lucky, I might get it within two years. Chances are, I probably will never get it because of stakeholder issues. There're a lot of issues with standards.
With semantic integration you make your dictionary explicit. A dictionary has connections—a human dictionary—so if two people have different messages, they each have a dictionary they can point to. That's what standards like OWL, RDF, and RDFS are all about. The idea of doing this without RDF and RDFS is that one or the other or even both could actually be machines that are reading the explicit semantics so that they understand what their messages mean. And now we can do integration by, "If you don't understand something, look it up in the dictionary."
With these ideas in mind, I've asked the panelists to address what role standards play in successful interoperability. What about things like SUMO, RDF, and OWL? What are barriers to adoption—in the enterprise, but also in the world at large?
Dave Hollander:(Mr. Hollander is CTO for Contivo.) When you start thinking about how to apply things to systems—the more complicated, the more intricate the behavior, the more you have to think about whether you have a sufficient insight into the semantics to try to make them interoperable.
As a system architect. I see that magic traction point between when you change the data does the behavior change, or if the behavior changes what should be the reflection in the data. I don't have to worry about the real cause; I can simply say, Oh, I now have a cause-effect relationship I can manage.
Semantic interoperability is semantically aware systems. It's the high-end goal where you really want to go over the long term, where society is going to force us systems architects to get to. It certainly is driving a lot of interest in the conversation here today.
But there's a step before that of semantic integration, for which I make a very strong distinction. When you get to the point where you are describing your systems with sufficient semantic detail that you can go out and work with another system, I would argue you are no longer doing the kinds of things that you are normally doing in integration; you're now in a whole different world—so dramatically different that it deserves to be thought of as a different thing. In my mind dealing with semantic integration is a good stepping stone to getting your information organized, beginning to understand your information well enough that you can go through this top-down ontology question.
Why integration? Integration is a huge business problem. I've seen numbers ranging from 50% to 95% of the spin on putting in applications and getting applications to work with each other on simple integration problems. It's not the plumbing; we've solved plumbing. Web services are the end point of solving the plumbing problem. The real problem is around the semantics, and when you look at how to integrate real-world business systems with very sophisticated semantics where the relationship between the data and the behavior is not really well known, you need an intermediate step. That's what I've been calling semantic integration.
The point—and the basis of the Contiva product—is to take what's current in a very ad hoc fashion in terms of describing the information sets that have to be exchanged and doing it in a much more structured way, to do it in an explicit, well-managed metadata—centrally managed data, a repository. It gives you huge cost reductions, both in initial design and in maintenance, which may be the more important.
One question was what are the things that we're concerned about as we move into more semantic development. To me one of the big ones is how do you maintain these semantic relationships? The earlier panel talked about RDF graphs where you don't prune the branches; you just keep adding more branches. You are basically creating an environment where we're going to have as much semantic trash as we have information trash on the web.
From a network viewpoint, though, one of the beauties of looking at integration is it significantly reduces the amount of information and behavior that you have to observe, because you're no longer worried about the stimulus, what goes on inside of the system, how the operators have to be trained. All you're worried about is the interface between the systems.
[Having this] leverage is huge, because any very simple business process is going to have hundreds of touchpoints that need to be integrated; whether they're business processes, systems, internal, external, across platforms, there's a tremendous amount of information to pull together to solve the kind of business problems that IT is facing today. Sarbanes-Oxley is a good example. The auto industry—within seven days of ordering your car, it should be delivered the way you ordered. Today the average delivery time is well over two months, and the accuracy rate is under 5%. Solving that problem is immensely huge, because that's a seven-tier supply chain that has to be working in real time. Imagine the number of touchpoints to address that one thing!
Important is the description logic. Two examples of description logics: the web was a significantly simplified way of representing documents in HTML and transporting them, HTTP. Those are two of the simplest protocols ever invented, and that was its beauty: simplicity.
Remember also one of the things about the web is I could always take any HTML page, look at the source, and figure out what it was trying to tell me even if what it was telling me was wrong. This kind of visibility is going to have to exist. And I think as a community, bringing the visibility to the description logics and really trying to address that is going to be one of our huge challenges.
So semantic integration is kind of this bridge. It's a way to address the standards that Dean mentioned earlier. But all of the standards are always going to run into this problem of variability. Semantic integration allows you to use the ontology to build these associations, but then go in at a point of bottom-up building capability, and identify instances that do not follow the pattern.
And veracity: Think about the quality of the relationships and how much you can trust that those relationships are true; in an enterprise environment, you have to have 100% reliable relationships, and you have to understand why they were made and where they were put. Otherwise you're going to be accepting orders that have no legal context and committing resources that cost the company.
Tell us what you think of this post using our On or Off rating system. Only your most recent vote will count.
Member Comments
Dave makes an interesting and very valid point about the quality of relationships. In fact, certain situations call for high quality ontology (i.e., populating ontology with facts/assertions based on more reliable, more trustworthy, higher quality knowledge sources). These include situations such as:
-- when you want automated reasoning (albeit not necessarily based on logic based reasoners) and computations, or minimize human involvement
-- Enterprise semantic applications used to make business decisions (compared to say panWeb applications, where because of broader scope and diversity it is hard to get consistency and quality on which automated decisions can be based)
-- semantic applications involving analytics, knowledge discovery and integration (compared to semantic search and browsing where we expect that human intellect and knowledge will be applied over the results a software system offer)
I would like to offer a comment regarding the use of and the importance of description logics. Formal representation, esp. one that affords unambiguous interpretation of meaning (to agents/programs or humans), is crucial and critical. In this sense, I feel RDFS and OWL are valuable, especially when there seems to be a fair degree of acceptance. With DoD and Intelligence practically making use of these standards as requirements, I would guess vendors are unlikely to look for alternatives. However, in my view, this importance of description logics as the basis of broadly accepted reperesentation language/framewwork does not extend to the reasoning and computation. Even for representation, there are situation where expressiveness of RDFS/OWL is not sufficient. But more on this at some other time (or see: Sematics: implicit, formal, powerful: http://www.google.com/search?q=Semantics%3A+Implicit%2C+Formal%2C+Powerful)
TG, not sure what you mean, but send me whatever you have. You know my real e-mail address. BTW, I can't seem to get you on LinkedIn. I received your request, but when I click on the link, it goes nowhere.
BTW, since I have your real e-mail address, I'll zap you another address which is better to use for sending documents, etc.
"The real problem is around the semantics, and when you look at how to integrate real-world business systems with very sophisticated semantics where the relationship between the data and the behavior is not really well known, you need an intermediate step. That's what I've been calling semantic integration. "
DS
Be sure to ask for a demo/whitepapers, or if you like I can give you a bit of feedback as well. We are starting to move in this direction, changing human behaviour and processes in IT and business departments takes time.
This is an *extremely* important topic. It will be interesting to see how many views it gets ... and how many comments, as well. In the grand scheme of things, to most AO members, the Semantic Web will have a greater impact on their careers than anything to do with China. How about a "Letter from the Semantic Web"?
CBS MarketWatch
Michael Powell Read This: FCC Needs Broadband Trust Fund
E-NEWSLETTER SERIES