[Geodata] Minimal metadata for modern geodata

[Pat Cappelaere]

Jo,

I think you hit that nail on the head and would encourage you to even pound
on it harder.
More distance between the 2 stools is a potential approach.  How much
further can we push?
The ebXML group is doing fine.  They target a specific audience.
The  GeoRSS group targets another audience (Mass Market).  The only catalog
they need might be Wikipedia-like.  This would be simple enough!
Pat.


From: "Stefan F. Keller" <sfkeller at gmail.com>
Date: Fri, 16 Mar 2007 00:54:39 +0100
To: Jo Walsh <jo at frot.org>, <geodatacommons at lists.gsdi.org>,
<geodata at lists.osgeo.org>, <ddanko at esri.com>
Subject: Re: [Geodata] Minimal metadata for modern geodata

Hi Jo,

Thank you for giving feedback, especially also to INSPIRE: Delicious
reading. 
 
These are some of my insights so far "towards a minimal metadata information
model" - just to backup your "Reading the INSPIRE Metadata Draft":
* Attributes in general: cardinality? consider embedding URL.
* How to deal with data set, supersets and series?
* dc:subject: Prefer open taxonomies over hierarchical and/or fixed keyword
lists.
* dc:relation: What to put into this important attribute? which
relationships are allowed?
* dc:coverage and spatial specializations of it: how to encode extent and
crs? 
* dc:identifier: Any policies? Is it an URN? I'm tending to say it points to
metadata (not data access points).
 
I thought about which of these attributes are essential metadata as part of
any geodata and made a comparison between ArcCatalog, GeoMedia and OGRINFO
(on Shapefiles). That's what you can expect to be extracted automatically
from data resources:
* dc:title........ <= File name and path
* dc:date/modified <= Modification date
* dc:description.. <= Layer names, feature count, source system (as
concatenated strings)
* dc:coverage......<= Extent (bbox) and CRS
* dc:type......... <= vector/raster (from Geometry types)

I think there is a potential to include data set language and to insist on
modification date and CRS especially in databases.
 
WxS/OxS getcapabilities (and FGDC inside ArcCatalog) go beyond these
attribute values and include information not contained in the data resources
like contact addresses.
 
To me the question arises, if these attributes (contact, dc:publisher, etc.)
are maintained best near the local data resource too or better in a
webapplication? My current idea is to sync (one way) the metadata coming
from data sets under the control of a webbased metadata editor.
 
-- Stefan

 
2007/3/15, Jo Walsh <jo at frot.org>:
> dear all at geodatacommons,
> 
> I come in peace from the OSGeo Geodata Committee. Collectively there
> has been some hard thinking about geospatial metadata, with a lot in
> common with the position held in your MD whitepaper. We've also been
> coordinating a FOSS community response to the draft profile for
> metadata in Europe. It's more of a meta-profile, a very different beast
> from the new North American Profile guidelines based on ISO19115.
> http://wiki.osgeo.org/index.php/Reading_the_INSPIRE_Metadata_Draft
> 
> If you read that page, there is a mention of wanting to connect with
> the geodatacommons group on a response to the NAP Profile draft.
> I sat down and read it, realised there is no way of modifying or
> critiquing it without critiquing the whole ISO19115 model, and wrote
> this discussion draft: http://frot.org/terradue/minimal_metadata.html
> 
> On the one hand I am starting to hear real world success stories from
> people working, without other options, in an ebXML framework to carry
> metadata about OGC standard web services for geographic data around;
> on the other, GeoRSS/FOAF/Dublin Core between them provide adequate
> representation for a minimal model for 'non-expert' metadata to be
> published and collected. ISO19115 is falling between two stools, and I
> fear that as it is currently framed it is going to have as little
> utility to data producers and users, as the current FGDC core does.
> 
> I want to know what I am overlooking, what I am taking for granted and
> how the free and open source geospatial community could help out the
> geodata/metadata user community by building simpler interfaces and
> real use cases into cataloguing and search platforms...
> 
> cheers, and best of luck,
> 
> 
> jo
> --
> _______________________________________________
> Geodata mailing list
> Geodata at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/geodata
> 


_______________________________________________
Geodata mailing list
Geodata at lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/geodata

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/geodata/attachments/20070315/41be4ce3/attachment.html