Raw

I just added links on WWW2007 Dev Track Resources to the handful of slides (PDF) for the Open Data panel and Tim Berners-Lee's slides (HTML) on Linked Data and the Tabulator. Good prompt to mention the subject here.

In short, Linked Data is what makes the (Semantic) Web as a whole a navigable distributed database, encouraging maximal reuse of the data. Not rocket science, Tim's four rules should seem trivially sensible to anyone that's ever put material on the web:

1. Use URIs as names for things

2. Use HTTP URIs so that people can look up those names.

3. When someone looks up a URI, provide useful information.

4. Include links to other URIs. so that they can discover more things.

The poster child on Linked Data right now is DBpedia.org, which contains a machine-friendly representation of the Wikipedia's data, but also (more to the point) is full of links to other data sources - e.g. places as described in Wikipedia are semantically linked to their equivalents in Geonames, e.g. :

<
http://DBpedia.org/resource/Berlin>

owl:sameAs
<
http://sws.geonames.org/2950159>
 .

Linked Data was a real honeypot topic at WWW2007, both for people already using Semantic Web technologies and those with problems which the technologies should be able to help solve (e.g. Peter Murray-Rust's chemistry). There was a scheduled session on it, but also a couple other less formal discussions. It seemed very much the case of an idea who's time has come (...or - a nice label for an existing practice who's time has come).

One thing that really impressed me was the relaxed confidence of the speakers on the topic, underneath there was something like " we know what we're doing - and it works!". For example, Chris Bizer had a slide showing DBpedia linked to half a dozen datasets. Another slide included an additional half dozen of so linked datasets, what DBpedia will be linked with in 2 months time ( PS. here are the slides - in comments, Georgi: "Well, I can say that 2-3 month was a quite cautious estimate. We almost got there :)") . No messin'. Plenty of per-case technical issues for sure, but overall straightforward, the hardest problems of doing knowledge representation in a global space having been solved to a useable extent.

Ok, when it comes to user interface, Semantic Web applications built from the ground up currently tend to lag behind those of the desktop or shiny in-browser stuff ( noting that existing apps can often be connected to the Semantic Web without much difficulty). But on the one hand Tabulator (and conceptually similar tools such as the OpenLink Data Web Browser) demonstrate that generic data browsing is entirely possible. On the other hand applications like Revyu.com show that Semantic Web/Linked Data applications can have very much the same look & feel as more traditional silo-based Web 2.0 applications. It's like the web is waiting, get on with it...

Also related, in his keynote ( slides, HTML) Tim talked of big picture stuff, the "philosophical engineering" of web science. But the iterative design process (with obligatory node & arc diagrams) led it directly into the realm of Linked Data. (Paul has some notes on the keynote).

Loads more can be said on this topic - Paul Miller has more, Dan Connolly has more.

See also: Linking Open Data community project. Not sure if it's back online yet, but Browsing Musicbrainz's dataset via URI dereferencing is another nice example of Linked Data in action.

A final thought - I can't remember exactly where this came up, but recently there was a bit of discussion around the use of HTTP URIs for XML namespaces and around schema. Anyhow it used to be only the newbie (or the poorly-optimised validator) that tried to resolve these. But everywhere you have a HTTP URI there's the potential for the client to follow its nose and get more information. Cute Easter Egg.

@en