The Structured Web | Hacker News Reader

The Structured Web

groups.google.com

43 pointsnslater9 comments

Why the Semantic Web didn't work, won't work, and what we can do about it.

The Semantic Web utopia suffered from pretty bad chicken-or-egg problem: why would anybody publish semantic data when nobody used it, and why would anybody consume it when none was available. Instead of the utopia, I'm seeing technologies of the Semantic Web stack pop up in different places... CMSs publish RDFa for SEO purposes (see Google's Rich Snippets), and for example MeeGo's Tracker data handling layer is actually an RDF triplestore, with SparQL provided as the main API for app developers.

Apart from these, I see the big potential for these technologies in integration: if you publish your content with RDFa annotations, your site has an API 'for free'.

It might not seem like it, but publishers are coming around (at least in some domains). RDFa is getting traction as a source for metadata in communities like education and open content (disclaimer: it's part of my job to convince people in this area why they should also publish RDFa metadata).

I'm convinced that this will be a gradual thing that takes hold in certain areas where it's immediately useful, and from there extends to other areas, sort of like the beginnings of the internet.

Yep, I'm also seeing positive signals. Both the SEO and the integration ("your website is your API") angles seem to work pretty well with publishers.

In the IKS Project we're trying to implement some semantic features to many popular Open Source CMSs. When that is built-in, there will be lots more RDFa out there... http://www.iks-project.eu/

I got the impression that something was going to be proposed, but it seemed vague . . . is there something being proposed?

On the 303 stuff: I don't understand the problem. RDF can talk about various things, not just web pages: RDF uses general URIs, which could be URLs, URNs, or whatever else is invented. If you don't want to make statements about a web page, you don't use an http URL. What is the need for using 303 responses?

And I don't understand the JSON comparison. Sure, RDF in XML is dismal, but RDF in Turtle is OK. The core, real, RDF is just URI triples -- subject, predicate, object -- it is already very simple, elegant, and powerful. How does JSON help?

I do like the idea of RDF/data-web/GGG uses and applications, but I can't figure out what this article is really trying to say.

Nice article. I would argue that JSON is suited for a structured web though. Imagine even something like RSS turned into a JSON structure — if everyone would simply publish open JSON feeds we could have a lot of interesting client-side applications...

Sure, JSON seems to have lots of uses.

I know a few people who've been working on a JSON Schema for the more formal datasets. I worked on an Internet Draft for this for a while, before the effort crumbled. Like most of the Web though, JSON really tends to shine when it is semi-structured — with any additional structure applied to it, post-hoc, by the consumer. CouchDB is a good example of a system that embraces this property.

You can even publish semantic data with http://json-ld.org/

What's the "old Atom wiki"?

I wasn't actively involved, but as far as I know, the old Atom wiki was the birthing place of Atom. Sam Ruby hosted it on his site, and got a lot of discussion around the idea of a new syndication format. I believe it was called Echo for a long while.

http://intertwingly.net/wiki/pie/FrontPage

Reply on news.ycombinator.com