[esip-semanticweb] help with marking up WMS/WCS Capability docs?
Benno Blumenthal
benno at iri.columbia.edu
Thu Jul 16 17:32:23 EDT 2009
On Thu, Jul 16, 2009 at 5:20 PM, Christopher Lynnes
<Chris.Lynnes at nasa.gov>wrote:
>
> On Jul 16, 2009, at 5:08 PM, Benno Blumenthal wrote:
>
>
>>
>> On Thu, Jul 16, 2009 at 4:56 PM, Christopher Lynnes <
>> Chris.Lynnes at nasa.gov> wrote:
>> So you are saying it would be something like:
>>
>> <Keyword
>> xmlns:xlink="http://www.w3.org/1999/xlink"
>> xmlns:role="http://someuri_root/dataset">
>> OMI_A_G
>> </Keyword>
>>
>> No, that is not what I meant. Ignore for the moment that I have not
>> reviewed WMS Schema so that this is not legal there, i.e. still need to find
>> the proper place for this
>>
>> What I meant was that is is either an object, i.e.
>>
>> <Keyword
>> xmlns:xlink="http://www.w3.org/1999/xlink"
>> xlink:role="http://someuri_root#dataset"
>> xlink:href="http://somecontrolled_vocabulary_uri_root#OMI_A_G" />
>>
>
> OK, then this would actually be:
> <Keyword
> xmlns:xlink="http://www.w3.org/1999/xlink"
> xlink:role="http://somecontrolled_vocabulary_uri_root#dataset"
> xlink:href="http://somecontrolled_vocabulary_uri_root#OMI_A_G" />
>
> ...in that "dataset" is also controlled, representing a class, whereas
> OMI_A_G is an instance of the class.
Yes, they are both controlled, but usually the class names are in a
different namespace than the instance names, i.e. the concept dataset is
quite different than an instance dataset and they are likely to be declared
in very different places, e.g. a standards body vs a data provider.
>
>
> I can see how that would work, but it may be counterintuitive to the
> "users", i.e., the data providers who will mark up the Capabilities
> documents. This is because their original idea was to reuse the <Keyword>
> element, yet they are putting no contents in it!
>
> The machine tag alternative, while less rigorous, may be more intuitive to
> them, e.g.:
>
> <Keyword>esip:dataset=OMI_A_G</Keyword>
>
> ("esip" is used as a notional namespace in this case.) This may be good
> enough for their use case in that it is more important to control the
> attribute (i.e., class) side of the equation than the value (i.e., instance)
> side.
Yes, but you need to declare the esip: prefix somewhere. Also if you do it
in an XML standard way, then there is a chance that XML standard software
will understand it. Someday....
>
>
>> where we are representing the controlled vocab as an URI, or
>>
>> <Metadata xmlns:myns="http://someuri_root#">
>> <myns:dataset>OMI_A_G</myns:dataset>
>> </Metadata>
>>
>> where myns:dataset implies a restricted set of possible strings as values,
>> i.e. a controlled vocabulary.
>>
>> Benno
>>
>>
>>
>> On Jul 16, 2009, at 4:40 PM, John Graybeal wrote:
>>
>> To the best of my knowledge, the use of role and arcrole[2] within
>> xlink[1] to specify the purpose of the locator (href) attribute is
>> limited only to requiring URIs. So you can create a vocabulary that
>> describes any set of roles that you want.
>>
>> John
>>
>> [1] xlink semantics: http://www.w3.org/TR/xlink/#att-method
>> [2] Role, Arcrole, and Title: http://www.w3.org/TR/xlink/#link-semantics
>>
>> On Jul 16, 2009, at 1:15 PM, Benno Blumenthal wrote:
>>
>> Hello Chris,
>>
>> I have not chimed in yet because I have not had a chance to review
>> the options for metadata in WMS, and I wanted to say something
>> precise. Essentially if you just want controlled vocabulary, XML
>> and RDF are quite similar (i.e. if Brian's machine tag implies a
>> particular controlled vocabulary for the value, then you have
>> represented the information isomorphically to using RDF objects for
>> the concepts, and you can crosswalk). If you want to relate two
>> objects, the xlink is XML's version of what is native in RDF, but I
>> think xlink is very limited as to the meanings of the connections,
>> whereas RDF lets the connection's meaning be specified explicitly
>> according to a convention.
>>
>> But meanwhile, could you explain your example more fully -- I don't
>> understand what
>>
>> dataset=OMI_A_G
>>
>> means (you called it a relationship).
>>
>> Benno
>>
>> On Thu, Jul 16, 2009 at 3:55 PM, Christopher Lynnes <
>> Chris.Lynnes at nasa.gov
>> wrote:
>> On Jul 15, 2009, at 6:52 PM, John Graybeal wrote:
>>
>> I am not deep in any one of these details, but I am a little familiar
>> with all the possibilities you mention. So consider the following as
>> notions to be verified:
>>
>> 1) RDFa won't fit gracefully into the OGC schema as written, because
>> it is using a different set of elements (that is, it is tuned to the
>> HTML elements). "To date, because XHTML is extensible while HTML is
>> not, RDFa has only been specified for XHTML 1.1." [1] The way it
>> works is to add attributes, which must be in an RDFa-compatible XHTML
>> schema if the document is to validate.
>>
>> 2) I wholeheartedly endorse the intent to use controlled vocabularies
>> in a way that is compatible with the semantic web. All that should be
>> necessary to do this is to find a suitable place where a URI can be
>> placed. Then you can create controlled vocabularies whose terms
>> correspond to URIs. There is extensive guidance on this topic at the
>> MMI site [2]. I do not know of any reason these approaches would be
>> incompatible with the ESIP ontologies.
>>
>> 3) The way OOSTethys [1] chose to add more specific descriptions/
>> references to SensorML/O&M was through the use of xlink, which is
>> supported in the OGC schema. Examples are on the OOSTethys site. As
>> far as I know, this is the most used OGC practice to meet this need to
>> date. (Because I don't know of any other particular recommendation.)
>> Note that if a standard allows a name to be specified as a URI (which
>> most of the SWE standards do, yes?), that is another place where the
>> sensor web.
>>
>>
>> Can you point me to any specific examples (i.e., URLs to files)?
>>
>> I'm having some difficulty seeing how a *relationship*, like
>> 'dataset=OMI_A_G' can be expressed with the xlink or URI...
>>
>> Thx,
>>
>>
>>
>> On Jul 15, 2009, at 12:53 PM, Christopher Lynnes wrote:
>>
>> Greetings!
>>
>> Over in the Air Quality Cluster, we are experimenting with using
>> some kind of structured markup / tagging of OGC WMS and WCS
>> capabilities documents (inside <Keyword> elements) to allow us to do
>> structured searches on the documents. An example might be, "give me
>> the layers where Dataset = 'OMI_AI_G'". Seehttp://
>> wiki.esipfed.org/index.php/WMS_GetCapabilities#WMS_GetCapabilities_Layer_Description
>>
>> Thing is, we figure if we are going to try to implement this kind of
>> markup with a quasi-controlled vocabulary, we should do it in such a
>> way that it is compatible with or even leverages the semantic web.
>> We have pondered a machine tags approach, e.g.,
>> <Keyword>esip:dataset=OMI_AI_G</Keyword>. (A link to an initial
>> attempt of a WMS that includes the current keyword encoding:
>> http://webapps.datafed.net/AIRNOW.ogc?service=wms&request=getCapabilities&version=1.1.1
>> )
>> .
>>
>> Alternatively, we have heard RDFa mentioned for microformats, though
>> mostly in the context of XHTML. Can this be applied to OGC's XML
>> and if so, how?
>>
>> Can the ESIP Semantic Web cluster provide a recommendation or
>> suggestion in how to move forward that would be:
>> (a) flexible and extensible,
>> (b) compatible with the evolving ESIP datatype and services ontology
>> and
>> (c) lightweight and easy to use?
>> --
>> Christopher Lynnes NASA/GSFC, Code 610.2
>> 301-614-5185
>>
>> _______________________________________________
>> esip-semanticweb mailing list
>> esip-semanticweb at rtpnet.org
>> http://lists.deltaforce.net/mailman/listinfo/esip-semanticweb
>>
>>
>> John
>>
>> --------------
>> John Graybeal <mailto:graybeal at mbari.org> -- 831-775-1956
>> Monterey Bay Aquarium Research Institute
>> Marine Metadata Interoperability Project: http://marinemetadata.org
>>
>>
>> --
>> Christopher Lynnes NASA/GSFC, Code 610.2
>> 301-614-5185
>>
>> _______________________________________________
>> esip-semanticweb mailing list
>> esip-semanticweb at rtpnet.org
>> http://lists.deltaforce.net/mailman/listinfo/esip-semanticweb
>>
>>
>>
>> --
>> Dr. M. Benno Blumenthal benno at iri.columbia.edu
>> International Research Institute for climate and society
>> The Earth Institute at Columbia University
>> Lamont Campus, Palisades NY 10964-8000 (845) 680-4450
>> _______________________________________________
>> esip-semanticweb mailing list
>> esip-semanticweb at rtpnet.org
>> http://lists.deltaforce.net/mailman/listinfo/esip-semanticweb
>>
>>
>> John
>>
>> --------------
>> John Graybeal <mailto:graybeal at mbari.org> -- 831-775-1956
>> Monterey Bay Aquarium Research Institute
>> Marine Metadata Interoperability Project: http://marinemetadata.org
>>
>>
>> --
>> Christopher Lynnes NASA/GSFC, Code 610.2 301-614-5185
>>
>>
>>
>>
>> --
>> Dr. M. Benno Blumenthal benno at iri.columbia.edu
>> International Research Institute for climate and society
>> The Earth Institute at Columbia University
>> Lamont Campus, Palisades NY 10964-8000 (845) 680-4450
>>
>
> --
> Christopher Lynnes NASA/GSFC, Code 610.2 301-614-5185
>
>
--
Dr. M. Benno Blumenthal benno at iri.columbia.edu
International Research Institute for climate and society
The Earth Institute at Columbia University
Lamont Campus, Palisades NY 10964-8000 (845) 680-4450
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rtpnet.org/pipermail/esip-semanticweb/attachments/20090716/03cc472c/attachment.htm>
More information about the esip-semanticweb
mailing list