Raw

While compiling " This Week's Semantic Web" I ran across a couple of things that aren't particularly new but deserve calling out. These things fit into the categories of being RDF based, having an app-centric (i.e. domain-specific, vertical) user-friendly front end, covering a domain that isn't exclusive to experts in that field, and having enough API goodness open to third parties to make them available to non-human agents (HTTP with RDF and/or SPARQL) and hence part of the semweb at large.

• Bitzi's Bitpedia digital media encyclopedia - metadata about binary files on the Internet, particularly those available through file-sharing services
• ITerating - Wiki-based Software Guide, directory & reviews of open source and commercial software
• Revyu.com - review anything
• Semantic MediaWiki - what it says on the box

Any more?

In a meeting the other day, David Errington ( Talis CEO) caught me on the hop somewhat by asking about existing real-world applications/services. My handwaving answer at the time was to point to the LinkingOpenData material. It's a tricky question in general because one of the key benefits of the Web of Data is that it will support a " small pieces, loosely joined" style of application building unlike anything we've seen before (in addition to more traditional apps). The nearest we currently have is probably the Web 2.0 mashup as catalogued over on ProgrammableWeb, though the vast majority of these apps are built without using the shared language of Semantic Web technologies ( RDF). Without this mashups have effectively to be built as one-offs, filling in points in the API x API matrix. Atom/RSS does approach being a lingua franca for metadata-enhanced human-readable content, but doesn't go anywhere near the problem of integrating data in general.

Even aside from the " small pieces..." angle, the space is wider than its ever been. One edge there are knowledge domain expert-oriented apps, like some of the things around Healthcare and the Life Sciences. There are at least two major apps ( Joost and Garlik) that use Semantic Web technologies internally, but don't as yet reveal much to the outside world in a machine-readable form (I'd be delighted to be corrected on that, btw). There are systems that use RDF-like models internally (e.g. Metaweb/ Freebase) that are on their way to becoming fully interlinked members of the semweb at large. There is a pretty large number of applications that can easily (even transparently) bridged to the semweb - e.g. Dopplr exposes GRDDLable microformat data (complete with @profile URIs - bravo!), and many of the mashup APIs can be mapped as semweb services (leading to more general data " meshups" as Kingsley calls them).



Of the "small pieces" in the LinkingOpenData cluster, I'm not sure but I think Revyu.com is the only part with a shiny app-centric (i.e. domain-specific) front end, as well as using RDF throughout the back end. The others involve some kind of bridging from non-RDF data.



David's question wasn't actually so general, he was asking in the context of a couple of a couple of apps in the pipeline at Talis (built internally on the Platform) that would definitely fit the categories above. Although it's rather a cheesy marketing trick, I think we should still be able to claim one or two "firsts" (without too much narrowing of category) despite the apps mentioned.

@en