According to Nova Spivack, the next generation of web technology, known variously as the semantic web and Web 3.0, is closer than you think. He has started a company that aims to exploit the next generation of web technology, and he expects to launch his first product this year. Radar Networks, the San Francisco startup Spivack, 37, heads, is in stealth mode, and he won't say exactly what it has planned except that it won't compete directly with Google. "We're working on a different strategy," he promises. "It's a game-changing one."
The idea behind the semantic web is to make data on web pages and in online databases better able to be read and understood by computers and used and shared by different software applications. "The semantic web will do for data what the web did for documents," Spivack says. "It will make it universally searchable and sharable." The standard way to organize and present data on the semantic web is described by the Resource Description Framework, or RDF, which plays a role on the semantic web similar to the one HTML plays on the original web.
The Web 3.0 approach to accomplishing at least part of this grand vision is to take Web 2.0-style tagging and formalize and expand it so documents and other web data that now must be interpreted by humans can be read and understood by computers. "It's about the machine doing more work on your behalf," says Oren Etzioni, a computer science professor at the University of Washington in Seattle.
Now, for instance, a human web user might tag a digital photo with vacation and Alaska to let himself and others know what it contains. But another user might tag the same photo with Juneau and summer. In neither case would the tags help a computer figure out what to do with the photo. And most web data does not even have tags, Etzioni notes, which makes it impossible for computers to do a great deal of the work that they could be doing for us.
If, say, all addresses on web pages were tagged using a consistent system, an entrepreneur's address book software could scan the web to keep it constantly updated with fresh addresses for customers, suppliers and others. If your calendar software could look into your online bank statement, it could automatically let you know when a check will clear. In the extreme version of this vision, all data everywhere could be read and exchanged by any computer program, to the benefit of users.
Doing the same for even a significant fraction of the billions of pages and trillions of bits of data on the web is obviously a tall order. The tools for generating RDF are, at present, costly and hard to use, which is slowing things down and frustrating semantic web visionaries. "A lot of work on the semantic web assumes it's already there, saying, 'If we had millions of pages in RDF, what would we do with them?'" Etzioni says. "People can tag, but they haven't shown the ability or propensity to write reams of RDF on their own."
Some of the most immediate and compelling applications for the shareable data of the semantic web and Web 3.0 lie in search. While Google does a good job of general web searches, it's not as good at searching corporate websites, says James Hendler, co-author of Spinning the Semantic Web: Bringing the World Wide Web to Its Full Potential. If data in corporate websites were better organized, Web 3.0-enabled search tools could make that data much more useful. "A search that does that better is the Holy Grail now," says Hendler.
Vertical search opportunities also lie in areas such as travel, where a vacationer could use a natural language search instead of keywords to find, say, beach destinations suitable for a family on a limited budget. Other promising applications lie in searching images, which at present are generally cataloged only with inconsistent Web 2.0 tags.
Perhaps the most tantalizing mystery about all this is where the first eye-catching success will come from. "What we're waiting for with Web 3.0 is a killer app," says Etzioni. HTML and HTTP, the technical underpinnings of the first generation of the web, were around for years before the web started taking off, he notes. "It was the presence of the browser that made it visual and appealing, and that's when it started snowballing. We're still waiting for that killer app, but once that's out there, we'll see a lot more activity around it."
It will be two to three years before the first big commercial successes occur in the semantic web space, Hendler predicts. But others say it could happen sooner, depending on where you put your efforts. "For people who are very technical, there's a big opportunity now around developing tools," says Spivack. "The next opportunity, from 2008 on, will be starting to build applications that use semantics. Then the third phase, which will be a few more years after, is when it will be baked into every browser."