I completely agree with Jeff.
In order to implement such distinction, clearly linking the two facet
metadata structures, and trying to answer to Ben, I think that we should
work on the DC "Content" core element.
In particular, such element contains the following fields: Coverage,
Description, Type, Relation, Source, Subject, Title; each of them are
characterised by qualifiers.
For instance, for a dataset, encoded in a given XML dialect (e.g. ESML)
such element can be:
Field: qualifier Content
Mandatory
----------------------------------------------------------------------------------------
Coverage: spatial DCMI Box scheme
N
Coverage: spatial TGN (The Getty Thesaurus of Geographic
Names) N
Coverage: temporal DCMI period
N
Coverage: temporal W3C- DTF (W3C Encoding rules for dates and times
N
- a profile based on ISO 8601)
Description: abstract Text (A summary of the content of the
resource) Y
Type: resource type "Dataset" (from the DCM Type Vocabulary)
Y
Relation: requires THREDDS Encoding Scheme for Relation(*)
Y
Relation: ....... Other relationships with other DC
documents N
Source URL scheme (e.g. the catalogue resource,
the dataset N
belongs to)
Subject THREDDS/DLESE or other existing encoding
schemes Y
Title Text
Y
Title: alternative Text
N
-----------------------------------
The link field is the qualification "Relation: requires": it means that
the described dataset resource requires the referenced resource to
support its function, delivery, or coherence of content.
(*)The encoding scheme for such field should be formatted as follow:
Component Definition
Default
-------------------------------------------------------------------------------------------------
URI The referenced resource URI
---
scheme The GI encoding scheme used for the resource
description ?
(e.g. ISO 19115, FGDC CSDGM, etc. )
name A name for the referenced resource (optional)
---
..... ........
In such way, THREDDS/DLESE is able to reference more than one GI
metadata schema format; in my opinion it is important for the
interoperability and extensibility sake.
Naturally, that is a very first proposal; any comment is more than welcome.
--- Stefano Nativi
Jeff de La Beaujardiere wrote:
Stefano Nativi writes:
# From my point of view, a metadata standard like Dublin Core can be used
# for describing the first Spatial Dataset facet, while a specific
# Geo-information metadata standard (i.e. ISO 19915, GDC CSDGM profiles,
# etc.) should be used for describing the other facet.
I think this type of approach can work if you build in a well-defined
link from one to the other: the DC metadata references detailed FGDC
CSDGM or ISO 19115 metadata at a separate URL, which in turn indicates
that this data is part of a THREDDS data collection with DC metadata
somewhere else.
The OGC Web Map Service takes a similar approach: the mapping service
and its data content are described by enough metadata to describe what
maps are available, and pointers to FGDC dataset descriptions can be
embedded. The latest FGDC CAP grants exercised this linkage
<http://www.fgdc.gov/funding/cap2001.html>.
-Jeff DLB
