[georss] Discovery of georss and other geographic information?

Tue Sep 19 11:18:16 EDT 2006

Stefan -

Thanks for the quick and considered response.

Would you mind if I share some of your observations with the OGC Catalogue WG? As you may know, there is considerable debate in the OGC regarding the way forward for the CSW spec.

With regard to GML, I was not suggesting that you use GML - just that WFS supports multiple versions of GML. Also, the GML Simple Features Profile is the result of considerable work and consensus by the OGC members - as well as public input. It is an adopted OGC standard. While not as parsimonious as GeoRSS GML, it is a heavily restricted subset of GML this profile designed to handle the encoding of the vast majority of non-topologically structured geographic content.

And yes, I agree that discovery services access catalogues. We have violent agreement there.

Carl

  ----- Original Message ----- 
  From: Stefan F. Keller 
  To: Carl Reed OGC Account 
  Cc: georss at lists.eogeo.org ; geodata at geodata.osgeo.org 
  Sent: Tuesday, September 19, 2006 12:47 AM
  Subject: Re: [georss] Discovery of georss and other geographic information?

  Carl,

  Thank you for the hints. I'm aware of most of theses documents except for the GML simple features profile (yet another one?). 

  The background of these activities is that there is a growing malaise about how metadata is being approached by the specifications you mentioned. So, there grew up a felt need for a really simple metadata exchange protocol which is low barrier and lean to implement. See here http://wiki.osgeo.org/index.php/Simple_Catalog_Interface and below the reasons. These finally let me propose a harvesting protocol like OAI-PMH 2.0, together with a slightly specialized Dublin Core metadata model (still tbd). 

  2006/9/18, Carl Reed OGC Account <creed at opengeospatial.org>: 
    Stefan -

    I browsed the wiki sites mentioned in your email. A couple of questions/observations:

    1. I believe that WFS 1.1 supports both GML 2.1.2 and GML 3.1.1 (see outputFormat) and by restriction, the new GML simple features profile. The new GML app schema is only 61 pages long with 20 pages of examples.

  Why do we need GML machinery just for the encoding of a bounding box? Dublin Core's dct:spatial or GeoRSS seem to be much more parsimounious. 

    2. I was wondering why OAI-PMH page compares OAI-PMH with WFS? WFS is not designed for harvesting or for being a Catalogue service. WFS is designed so that once a content resource has been discovered (and perhaps registered in a Catalogue/Registry) that source can then be queried and asked to return a set of features (as a GML payload). 

  CAT/CSW is based on a distributed query architecture and there's redundant spec. (and therefore code) to WFS. Why another protocol when there is WFS? 

  As explained in http://www.gis.hsr.ch/wiki/OAI-PMH distributed query like in CAT/CSW means that search is at best limited to the slowest server and to a least denominator of implemented specs. That is because each server needs to implement exactly the same query functionality. OAI-PMH does harvesting and indexing before hand. 

    3. Was wondering if the new OGC Catalogue 2.0.1 19119/19115 Application Profile has been looked at.? Seems to me that there could be some real synergy between this Cat App Profile and OAI-PMH. https://portal.opengeospatial.org/files/?artifact_id=8305 . There are already quite a few implementations of this Application Profile.

  You mentioned CSW implementations. To me it's like somebody who owns a truck saying "why don't all take trucks to move metadata around?" when bicycles would make the job. OAI-PMH is there since five years and even Google accepts it as alternative to Sitemaps. 

  From my experience, what you can expect at most from a data owner is nothing more than what's in Word/Visio (metadata) properties or what's mentioned in WMS GetCapabilities. So to me nothing more than Dublin Core is needed like the common returnable properties of Catalogue Services Specification 2.0.1, OGC 04-021r3, p.22 (as Josh mentioned) but extended by additional semantics which includes more URI for automation.

  I take ISO 19115, ISO 19119, ebRIM and all those forthcoming profiles as templates for the specification of information models internal to an organization. When talking about geodata discovery we are not talking about catalog services but search services on top of catalogs; see http://www.gis.hsr.ch/wiki/OSGeodata_Discovery and http://www.foss4g2006.org/contributionDisplay.py?contribId=234&sessionId=70&confId=1  .

  Stefan 

    ----- Original Message ----- 
    From: Stefan F. Keller 
    To: Raj Singh 
    Cc: georss at lists.eogeo.org 
    Sent: Tuesday, September 05, 2006 2:58 AM
    Subject: Re: [georss] Discovery of georss and other geographic information?

    Raj,

    I agree that encodings can differ as long there is a common understanding on the semantic level and as long 'best encoding practices' are fulfilled (what currently is the case in GeoRSS).

    But newly published encodings should consider established ones, which seems to not the case actually (in W3C draft?). I know that standardization is slow but programmers often don't care...

    But my initial question was not only targeted to adjust GeoRSS to be accepted as a Microformat. What I'm thinking about currently is (auto-)discovery of geodata and services as documented here http://www.gis.hsr.ch/wiki/OSGeodata_Discovery .

    I'd like to discuss if GeoRSS icons are means to guide webcrawlers to xml-encoded content or if there is a need for a 'friend' attribute in a (yet to be re-defined ISO 19115) metadata as described in http://www.openarchives.org/OAI/2.0/guidelines-static-repository.htm 

    -- Stefan

    2006/9/4, Raj Singh <raj at rajsingh.org>: 
      If you have a hammmer, everything looks like a nail.

      I say that because the hammer many people on this list have is information 
      architecture. We try hard to make everything elegant in the information
      model. I think interoperability occurs best at the programmer level, and
      therefore we shouldn't try to hard to make all encodings of geography 
      interoperable from an encoding standpoint. If it takes less time for
      programmers to code with less elegant encodings, then that's the "right" way
      to do it.

      This is a long way to say that I agree that GeoRSS should stop with support 
      for Atom and some other RSSes. If it's simpler to diverge from this encoding
      to do microformats and/or XHTML, so be it.

      My 2 cents,
      Raj

      On 8/30/06 10:38 AM, "Andrew Turner" < georss at highearthorbit.com> wrote:

      > Stefan F. Keller <sfkeller at gmail.com> wrote:
      >> 
      >> P.S. Still: Anyone who knows the status of GeoRSS (simple) as microformat? 
      >>
      >
      > What is the purpose of a GeoRSS microformat? Isn't Geo*RSS* targeted
      > and meant for RSS/Atom? There already is a 'geo' and 'adr' Microformat 
      > with widespread support. If you're just looking at adding line, 
      > polygon, etc. to a Microformat then expand on geo, but it doesn't seem
      > worth it, or a good idea, to try and force GeoRSS into XHTML. 
      >
      > There was a howto on possibilities of mixing RDF and GeoRSS: 
      > http://www.geospatialsemanticweb.com/2006/06/08/mixing-rdfa-with-georss

----------------------------------------------------------------------------

    _______________________________________________
    georss mailing list
    georss at lists.eogeo.org
    http://lists.eogeo.org/mailman/listinfo/georss

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/geodata/attachments/20060919/bd4f3a3d/attachment.html