[Geowanking] [Fwd: [Ann] LinkedGeoData.org]
chris at platial.com
Wed Jul 15 15:42:55 PDT 2009
> Many many programmers, some from a web background, others GIS, make perfectly awesome use of OSM. It also happens to have geographic entities identified by URIs, and a RESTful UI. That doesn't mean there isn't anything to improve, but I'm doubtful RDF is it
k i'll stick up for RDF.
RDF is quite simple in concept, and its core ideas are at last being widely applied even if not in the form originally envisaged.
The aim of the RDF enterprise is of course data integration: we want to pour structured data from all over the web and from diverse domains into a common pot where uniform tools for presentation and analysis apply.
The means are straightforward: 1) use an object-property-value graph as the basic data model (rather than eg tables), and provide a uniform syntax for such, and 2) give things globally-scoped names (URIs), and work towards using the same name for the same thing whenever possible. This principle applies to properties as well as real world objects.
Unfortunately, these sensible core ideas came wrapped in a way that many found unpalatable (and rightly so).
First, a lot of the propaganda for the semantic web, and for RDF, mixed in notions from AI and philosophy that are not intrinsic to the RDF idea. The word "ontology" has a pretentious sound even when used for down-to-earth purposes, and TimBL's famous semweb diagram http://www.w3.org/2000/Talks/1206-xml2k-tbl/slide10-0.html tops out with "Logic", "Proof" and "Trust".
Second, the most widely known syntax for RDF (RDF/XML) is way too complicated (though coevals such as N3 were not). And of the two well known type systems, RDF Schema and OWL, the first is inexpressive, and the second while elegant and powerful is very far removed from the type formalisms familiar to programmers.
But other incarnations of the core ideas have come along, and are now getting substantial traction, eg Microformats,RDFa, and Freebase. And now Google has announced that it will utilize RDFa and Microformats in web pages that it crawls in filling its "big semantic pot"; Google Squared is Google's initial presentation mechanism for this structured data. (See http://radar.oreilly.com/2009/05/google-adds-microformat-parsin.html - in which RV Guha a co-creator of RDF and Googler is interviewed)
This has big implications for the GeoWEB, which is fundamentally a data aggregation enterprise.
RDF traditionalists such as the Linked Data and DBPedia projects are getting somewhere too. Luckily all these wings of the RDF world are interoperable, so each path contributes to the grand aggregation.
IMHO OSMers should be happy to have their data traveling this route, however excellent its own APIs for current purposes.
----- Original Message -----
From: Mikel Maron
To: Bill Thoen ; geowanking at geowanking.org
Sent: Monday, July 13, 2009 11:34 AM
Subject: Re: [Geowanking] [Fwd: [Ann] LinkedGeoData.org]
> Sean Gillies wrote:
> > It's curious how alien the concept of geographic entities identified by URIs is to GIS folks. See Adena Schutzberg (not a programmer, but no dummy) scramble to make sense of it:
> > http://apb.directionsmag.com/archives/6086-Linked-Geodata-OSM-Gets-Linkable.html. For all the talk of "GeoWeb" in GIS, we still don't quite get the web.
> From: Bill Thoen <bthoen at gisnet.com>
> Well, when you pack 80 lbs of baggage into a 20lb suitcase, it makes that suitcase surprisingly hard to lift. With sentences like this one:
> "LinkedGeoData currently comprises RDF dumps, Linked Data and REST interfaces, links to DBpedia as well as a prototypical user interface for linked-geo-data browsing and authoring."
> Adena concludes "that project is about making OSM more useful for programmers and to provide tools to browse it and author in it."
Adena is smart, but gets this one wrong. Can anyone explain how an RDF representation of OSM is going to help any programmer?
Many many programmers, some from a web background, others GIS, make perfectly awesome use of OSM. It also happens to have geographic entities identified by URIs, and a RESTful UI. That doesn't mean there isn't anything to improve, but I'm doubtful RDF is it
you're not going to reach a very wide audience. The unprepared mind chokes on these gobbets of concepts, packed so tightly in one sentence. Adena's right to point out that most of us have to look up about every third word in that announcement and even then it's still hard to make sense of it. Of course, that announcement isn't meant for everyone, but it reminds knowlessmen like me that GIS on the web can get complicated. And as long as it remains portrayed as that complicated, we're not likely to get anywhere until Google simplifies it and shows us how it's done, like they did with web mapping a couple of years ago.
In fact, that's happening already. Look around and you'll see a growing number of interesting mash-ups with a Google map component being developed by people with little or no GIS experience. People who don't know the difference between large and small-scale maps, or which one's the longitude in a coordinate, or what "geocoding" means, are coming up with some pretty clever and innovative mash-ups where location plays a part. If GISers (or geowankers) focus too closely on the minutia of standards and protocols instead of what they're for, we'll get caught flat-footed again. (But it was amusing seeing the dominant GIS companies come up off their beds of laurels and scramble to come up with a plausible response as to why they didn't get there first!)
- Bill Thoen
Geowanking mailing list
Geowanking at geowanking.org
Geowanking mailing list
Geowanking at geowanking.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Geowanking