Information technology managers are familiar with the concept of the Semantic Web, even if they haven't heard the name before. The Semantic Web is about giving users the ability to manipulate, connect and associate Web resources in new and powerful ways. It's a capability similar to that of the corporate workhorse, the relational database.
The Semantic Web is about taking the relational database and "webbing it," as Tim Berners-Lee, the inventor of the World Wide Web, described it at a recent forum of the National Science Foundation in the US. The Semantic Web will allow businesses to manipulate external, heterogeneous Web data in much the same way they do internally, he says. But its most immediate use may be as a tool to solve data integration problems.
The Semantic Web is intended to infuse meaning into the Web and make data "machine-understandable." It relies on the development of standards and "ontologies," the vocabularies that systems need in order to associate and connect data across multiple databases. These ontologies would be linked to documents and other Web resources, giving computers the ability to infer relationships among these information sources.
The Semantic Web would allow manipulation across multiple, heterogeneous databases. This capability could, for instance, allow an electronic airline reservation service to automatically interact with a personal calendar program to arrange a flight that fits a user's schedule, even if there was no pre-established interface between the two pieces of software.
But accurately forecasting the kinds of services that the Semantic Web will deliver is as difficult as it was to anticipate at its inception what the Web might become. "In the beginning, I think it was very unclear what the dominant business model would be," says Ora Lassila, a research fellow at the Nokia Research Center. Similarly, the kinds of services that may emerge from the Semantic Web remain a "big question," he says.
But for corporations, the immediate value of the Semantic Web may not rest in using it for new services, but rather as a means of integrating data in heterogeneous environments. And the concepts and standards are sufficiently developed to allow that now.
Eric Miller, who heads the World Wide Web Consortium's (W3C) Semantic Web effort, believes it's the Holy Grail of enterprise integration. "We're talking about really reducing the turnaround time of data integration," he said.
The W3C, headed by Berners-Lee, is leading the effort to develop Semantic Web standards, such as the Resource Description Framework (RDF), the standard used for expressing ontologies, and XML, the standard for creating customized tags for the exchange of data between heterogeneous applications.
A more difficult aspect of building the Semantic Web is the creation of ontologies. This process requires efforts by diverse communities, such as the medical, insurance and finance industries, to develop common vocabularies that systems will use to recognize what's in a Web document.
Fortunately, creating ontologies doesn't require a global coordinated effort. If words are used differently, such as "title" in insurance vs. the "title" of a book, services will be able to map those differences to allow interoperability.
"The Semantic Web is very much a future-looking vision," says Ronald Schmelzer, an analyst at ZapThink in the US. "Trying to get computer systems to not only be able to communicate with each other, but also able to understand each other is traditionally a very difficult problem."
Indeed, developing the Semantic Web will take years, says Jas Dhillon, president and CEO of Celcorp, a US start-up, that has used RDF and ontologies to integrate applications.
"The Web is kind of stuck in terms of where it is right now, and if it is going to take the next step and enhance productivity and use, it has to become intelligent," Dhillon says. "We are betting that Tim Berners-Lee will succeed."
And if Berners-Lee does succeed in convincing software vendors to invest in Semantic Web applications, businesses will have a new way of integrating databases and applications, instead of using the traditional method of building interfaces one by one.
Celcorp uses ontologies built on top of RDF in its "semantic learner," which can create new "meta-applications" by extracting features from multiple existing applications. It analyzes what users are doing in the existing programs, how they navigate, what actions are taken and what data is used. It then puts that information in a knowledge base.
In the next step, a "reasoning engine" builds the new application. It does that by allowing users to specify goals or results they want to accomplish. Using RDF, the reasoning engine looks across the knowledge base to see if all the pieces are there to assemble the process.
In the third step, code is generated and executed based on the functions stored in the knowledge base.
Semantic Web proponents predict that more companies like Celcorp will emerge in the near term, offering similar services and new tools for applying semantic concepts to existing business processes.
Berners-Lee says technology managers can prepare for this future by asking their application vendors whether they're using RDF in their software development.