Here at Talis, we’re pretty big on the idea of Linked Data. It’s one of the key features of the Semantic Web vision, the idea that data are interconnected so man and machine can “follow their noses” to find new or relevant information. It’s also key to the idea of the “Web of Data” in which all published, online information can become part of a massive, usable database. An important element in seeing this become a reality is the notion of discoverability.

I remember, maybe ten years ago, someone explained that the World Wide Web contains so much information, that I’d be unlikely to find the “best” answer to my question. Of course, search engines have changed this, to such an extent that we “google” for information with a fair certainty that we’ll find what we’re after—or something closely related. If I want to find out about plane tickets to Boston, Massachusetts, i’d be unlikely to want to read about someone complaining about theatre tickets in Boston, Lincolnshire. Search engines made it possible to discover the right information that we’re after. Search engines, however, can only go so far in their ability to find information, because they’re looking at keywords within the documents they scan. They can’t identify many kinds of data, nor can they make use of them: they just point.

Fast-forward now to surfing the Semantic Web. Part of what makes the whole thing great is that the information (data) we’re after on the Semantic Web comes with its own information about what it is (metadata). So, I can know with an even greater degree of certainty, that the info I’ve found is what I’m after. It’s like finding a book in the Library. If we’re looking to borrow a particular book about the Battle of Hastings, for example, we probably don’t settle for the first one we come across in the catalogue. We look for the book we’re after, and know when we find it because it’s record identifies it to us. When we know what data is (and our software can be told what kind of data we’re looking at, and what’s in it), we can start to make use of these vast published datasets.

The authors of the voiD vocabulary are looking to make this metadata more relevant by providing data publishers with tools that describe their particular collection of data (dataset). It gives each dataset the ability to describe itself saying what topic it’s covering, what kind of data it comprises, numerical information about the data et cetera. voiD co-author Keith Alexander told me earlier: “It’s a way of saying “I have this dataset, this is what it’s all about, come and use it!”

You can find out more about the voiD vocabulary over on its site: http://rdfs.org/ns/void. The authors have gone a step further, however, and written a manual to make voiD that much easier to implement: http://rdfs.org/ns/void-guide. They’d appreciate feedback, too, so if you can think of a way to improve voiD or the guide, give them a ping at: void-rdfs-internals@googlegroups.com

Related articles by Zemanta

• Enabling the Linked Data Ecosystem (blogs.talis.com)
• Semantic Web and the Cloud have much in common (blogs.talis.com)
• Doing more with your data (blogs.talis.com)