For Shawn

["Jim" <Mitchelj@lic.gov.au>]

Shawn

Thanks for the comments.

I am looking at metadata from a slightly different angle.  We run
what is effectively a data warehousing operation.

My notional clients typically request a subset of existing datasets
and my objective is to supply that subset complete with metadata.

This then leads to the notion of "calculating" metadata values from
metadata attributes attached to individual data entities.  Given that,
then Lineage, Positional Accuracy and Attribute Accuracy might be
calculated as the dominant or average values represented in the
subset - with completeness and logical consistency being copied
across from tile or dataset level as appropriate.

We are already replacing "map-sourced" data by more accurate or
more current data from a variety of sources on a piece-wise basis.
For example, the realignment of a bend in a road could be inserted
with a positional accuracy of 1 metre, sourced from an engineering
plan with revised construction attribute such as "re-inforced concrete".

Both quality and metadata facts appropriate at the entity level can 
be considered as attributes - I hope we can build the recording of 
these into our routine data maintenance activity.  We cannot afford
to maintain metadata as an off-line process.

Regards

Jim Mitchell
New Temporary Title follows:
Project Manager (Infrastructure Data)
Geo-information Infrastructure and Management Program
Department of Land and Water Conservation