« Internet Zero | Main | Orlowski slams Wikipedia »

Semantic Web

Sir Tim Berners-Lee explains the Semantic Web in an interview published in the latest Technology Review. The interviewer is Mark Frauenfelder of boingboing.

The Semantic Web technology tackles the problem in two stages. The more mundane is a common data format. You can take a database or a calendar or an address book or a bank statement or a weather reading—basically anything with hard data in it—and make the machine write it in the basic Semantic Web language, instead of some proprietary or application-specific format. This solves the “syntactic” problem.

It still doesn’t solve the “semantic” one, though. For that, the Semantic Web first gives names to the basic concepts involved in the data: date and time, an event, a check, a transaction, temperature and pressure, and location. These are all defined just to mean whatever they mean in the system which produces the data—for example, “Transaction date as I get on a bank statement,” and so on. This set of concepts is called an ontology. Then, where there are connections between ontologies, such as when the date and time on a photograph is the same concept as the time on a weather report, we write rules to take advantage of these connections. This allows one to query the Semantic Web agent for photos taken on sunny days, for example. Bit by bit, link by link, the data becomes connected, interwoven. The exciting thing is serendipitous reuse of data: one person puts data up there for one thing, and another person uses it another way.

jon posted this at 4:07 PM
Share on Facebook| email to a friend

 Subscribe to RSS feed for Weblogsky