Magazine article Computers in Libraries

In Which 'Linked Data' Means 'A Better Web'

Magazine article Computers in Libraries

In Which 'Linked Data' Means 'A Better Web'

Article excerpt

In this year's columns, I've written off and on about linked data. I'm still convinced that it's an important topic, but now I think I've been pitching it from the wrong angle. If our gracious editors will do me the favor, I'd like to give it one more shot because I can't get it out of my head yet. Similarly, linked data seems to be engaging more of my colleagues in the course of their day-to-day work. I've traveled often this year and have discussed linked data with diverse groups of librarians, hackers, digital humanists, web developers, and administrators. I think I've come around to a new way of talking about it that might make more sense to more people. But to set the stage, here's what I think I've been doing wrong.

First of all the name has its issues. "Linked data" sounds like the kind of lofty objective that requires capital letters. That's a turnoff right there. It's also a passive formulation, though "linky data" and "linkin' data" are probably worse. The next issue is that talking about linked data makes it sound like something fundamentally new, but it's not new at all. It's a way to do what we've already been doing but even better. And that's the third issue--that what we're already doing is building a web of linked stuff; we've been doing that for a long time. Somehow adding "data" to the equation misdirects attention from the "linked" part. "Data-y links" doesn't sound any better either, even though it would at least make a noun out of the part I've come to think is most important--the links. Tim Berners-Lee called it linked data, though, so that's what it is.

Instead of fussing about the name so much, though, it's more helpful to focus on what I now see as the main point of linked data: building a better web.

We Are Still Building a Web

We've been building a web for a long time. The web's original specs are 20 years old, and most of us saw something like the web on our own computers 15 years ago. Since that time, those of us who've built various parts of the web have gone from hand-coding static HTML to using server side includes and CGI, from Perl scripts to database-backed sites and XML. Even library standards have followed along in their own way, with Z39.50 moving to the web with SRU, OpenURL, and the OAI specifications. Through those years we've all been putting things up and onto the web or, more to the point, into the web. Few things live on the web in isolation. And when it's up to search engines to find our stuff by following links around, we adopt approaches, such as search engine optimization, that focus how we structure links and HTML to draw attention to our stuff. These days, if it can't be found online easily, it doesn't really matter; it's as if it didn't even exist--or at least that's what more and more people tell me.

So even though how we build the web has changed steadily over the years (and keeps changing, as we programmers switch toolkits and frameworks every few years, disposing of older languages and tools when newer, better ones come around), the importance of links in the web has continually increased--after all, without links, there's no web. The first killer site on the web was the National Center for Supercomputing Applications' What's New page, which consisted solely of links to other new sites. Search engines mine bibliometric aspects of the relationships among links in pages to decide which sites to bump up in search results. Academics filling out their curricula vitae (CV) add as many active links to their papers as possible. Librarians like us scramble to get copies of these papers into their institutional repositories so they can offer at least one semireliably stable link to scholars for their CV. All that web building reduces to all that linking.

Subtle Shades of Linking

In April I wrote about Berners-Lee's "Linked Data--Design Issues" (http:// Data.html) in which he spells out the principles of Linked Data: using URIs (uniform resource identifiers) as names, using HTTP URIs, providing useful information at those HTTP URIs, and linking to other useful information. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.