- From: Danny Ayers <danny.ayers@gmail.com>
- Date: Wed, 21 Dec 2005 20:18:33 +0100
- To: Joshua Allen <joshuaa@microsoft.com>
- Cc: Henry Story <henry.story@bblfish.net>, tim.glover@bt.com, fugu13@mac.com, fmanola@acm.org, semantic-web@w3.org
I'm not sure I've caught all of the thread(s) but this is probably a good point for my €0.02. First off, I like the "emerge" in the subject line. I reckon it's pretty likely that some kind of data web is an inevitable emergent property of a doc-oriented web. Whether it uses Semantic Web technologies is another matter. I personally think that's reasonably likely. A graph-shaped data model is a must-have for a data web; layerable, logic-based schemas/ontologies are probably the only way of sanely managing the kind of diversity that's to be expected. The W3C specs are available... What I'm not so sure about with the original subject field is the implication that the emergence will have a clear, single originating path. There's the SPARQL endpoint side that Henry's talking about - if one or two big players provided such things, that could be a major catalyst. But equally, let's say someone wanted to write a desktop interface using WinFS to Google Base - RDF could well be in the frame, and such a tool could be popular. I would be very surprised if there weren't a few commercial concerns with their eye on PiggyBank. Or what about an enterprise management system using Oracle's new RDF store? Any of these islands /could/ turn viral, or may we'll just see gradual bridge-building. On 12/21/05, Joshua Allen <joshuaa@microsoft.com> wrote: > > worth looking at this. If I were Barnes and Noble of la FNAC I would > > try this out, before Amazon gets there. > > This motivation only works if there is credible evidence that Amazon is preparing to launch SPARQL endpoints, though. And even then, only if there is evidence that such endpoints would see broad adoption. > > > Once more groups get their SPARQL end points out, I forsee that major > > players will wish to standardise on some ontologies to: > > I believe we have enough specific evidence to counter this prediction already. Amazon has already exposed its business data in a variety of flavors: > > 1) Web services with loose contract (POX over HTTP) > 2) Intermediate format -- RSS using some simple, well-known extensions > 3) Web services using tighter schema (SOAP and WSDL) > > Can you guess the relative adoption of each style? This is a pattern we see played out across the industry. The clearest pattern is that when people are willing to expose their data, they're prepared to do so in a variety of ways. I'm not going to hold my breath for a big player to expose a SPARQL endpoint, but I wouldn't rule it out either. All three approaches you list can be pretty straightforward for simple systems, but get increasingly difficult when it comes to more complex systems. None of them go very far into the kind of data integration capability offered by SW tech. To take a specific case, I've been amazed for a long time that no-one's exploited the potential for using an RDF base for a biggish RSS-oriented system, maybe for blog search or whatever. The benefits are pretty obvious were one to emerge. Probable explanations for the absence of such a service are the relative immaturity of RDF/OWL tools and the (not unrelated) general unfamiliarity amongst developers of the techniques. But there have been continuing significant steps forward on both counts. There's enough going on in industry (bits) and academia (*lots*) to suggest that the SW idea has got a strong foothold. Probably more significant, looking around at developments on the Web in the wild (through my tinted lenses ;-), there's a definite tendency towards Semantic Web-like approaches. A good example is the microformats and StructuredBlogging initiatives, both making explicit data available on the web. Ok, a document format is used as a container, and the formatted data is traditional domain-specific, but this is a good way past scraping. It is real data on the web. If the intuition that the SW technologies offer significant advantages is correct, then I think we're long past the tipping point. That was a hard sentence to put together - significant advantages over what? Advantages over existing web techniques..? But virtually all existing web techniques (the best practice kind anyway) appear in the SW stack. It's not either-or. Perhaps major growth it will be triggered by SPARQL endpoints or people trying to integrate diverse microformat data. I'm not sure the environment is a tinderbox yet, and it may be another decade before the web looks remotely semantic. But the subject line question could maybe be flipped over: how will the semantic web *not* emerge? Cheers, Danny. -- http://dannyayers.com
Received on Wednesday, 21 December 2005 19:19:26 UTC