[OSGeodata] OSGeodata] discovery requirements

[Kralidis,Tom [Burlington]]

Comments prefixed with TK: 


________________________________

	From: Stefan F. Keller [mailto:sfkeller at gmail.com] 
	Sent: 04 August, 2006 7:16 PM
	To: geodata at geodata.osgeo.org
	Subject: [OSGeodata] OSGeodata] discovery requirements
	
	

	Tom Kralidis wrote Friday, August 04, 2006 3:23 PM 
	> Following the discussions on cataloguing, it may be a good
idea outline
	> what we would want a catalogue to do.
	> 
	> 1./ discover the existence of a web service 
	> 2./ discover the existence of a resource which is available
via web
	>     services (i.e WMS layer, WFS feature type)
	> 3./ discover the existence of a static resource (dataset,
document,
	>     etc.) 
	> 4./ ability to performs operations 1 - 3 with spatial,
aspatial and
	>     temporal predicates
	> 5./ provide a RESTful request API
	> 6./ provide responses in XML

	To me it's all this except perhaps 4. because the operations
should concentrate on (incremental) 'harvesting' rather than distributed
online querying. Security, as Norman Barker suggested, is solved
otherwise and has been discussed in debates about RESTful services and
security. 

TK: what part of req 4 are you concerned with exactly?

	Having said that: I speak about 'search service' and to me such
a web service has to important 'interfaces' one to the user and a
programmatic one to the internet (besides internal persistence layers
etc.).

	Paul and others strongly emphasized to start with the focus on
what users are probably looking for. To me they look for decisions and
want to have access to information which is spatially related (sic:
'Open Access' and OAI). They are not looking for data per se and also
not for services but for the map information behind them. 

TK: right.  The 19115 metadata would be available through a service
interface (i.e. OGC:WMS Capabilities), no?

	On the other hand we have spatial data providers as well as a
community, like us (not actually the same as the data providers) who
calls for open and better access to this geodata. What seems now
crucial, is what have the geodata providers to do such that their
information can be 'discovered' and hopefully bound automatically to
users seeking geospatial decisions. 

	So to me the big challenge is now, to define what minimal
information data providers should document as metadata and how they
could publish it, so that spiders and registries can find, register and
publish it as catalogues, specialized search engines for the deep web or
build in geodata discovery components (as p.ex. in uDig).

TK: sounds to me like we need a lean and mean metadata profile of 19115.
In Canada, for ISO19115, we are awaiting a North American profile
(December 2006, I believe).  The search interface (underpinning) would
merely map to those elements.


	-- Stefan

	BTW: I could transfer my stuff here
	http://www.geometa.info/rappiinfo/wiki/index.php/OSGeodata
	to
http://wiki.osgeo.org/index.php/Geodata_Metadata_Requirements#Discovery
(but I would open an extra page there)?

TK: how do you create a new page (I'm not used to this wiki)?  If you
can create a new page (Geodata_Discovery), then we can go from there,
and I will move the items from the metadata page (requirements) to a
part of the new page.