a talk about open source geo, metadata and INSPIRE

[Jo Walsh]

dear all,

On Tuesday I am giving a talk at a conference organised by the
Romanian Space Agency which forms part of their preparation for the
INSPIRE directive after Romania joins the EU next January.
http://portal.rosa.ro/conferinte/2006_12_event/index.php?page=infoday-inspire

The talk is catchily titled "An Open Source, Geospatial Web approach
to implementing INSPIRE" and is part open source geospatial plug, part
rant about neogeo specifications, part metadata + GeoNetwork
obsessing, part open data propaganda. I am not sure if i will have
time to cover it all, or really what my fellow attendees are likely to
expect, so my tune might change before then. Nonetheless i include my
notes for it below this email and any feedback would be appreciated.

Perhaps i will run into the full text of the elusive final INSPIRE
directive while i am there. It feels odd to be talking about it when
it doesn't publically exist yet - but "search services" will be mandated
and free of cost for sure, so metadata is solid ground.

cheers,


jo

===============================================================
An Open Source, Geospatial Web approach to implementing INSPIRE
===============================================================

Jo Walsh - OSGeo - 12-12-2006
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Now a compromise has been reached that will allow the INSPIRE Directive
establishing a framework for European spatial data infrastructure to become
law. Within 2 or 3 years its "implementing rules" governing the publication of
geospatial data and metadata, by state agencies that collect it. Now all across
Europe, public sector information providers will be asking, "How can we do this
as cheaply as possible?" and "How can we generate the most value from this
data?".

I think the answer to the first question is contained in Open Source
software and open standards for information exchange. The answer to the second
question is contained there too, but in a less obvious way. The perspective I
hold is biased by the fact that I am serving on the Board of the Open Source
Geospatial Foundation; but this in turn I am doing because of a conviction that
open source offers the best platform and the best model for the maintenance of
a civic information infrastructure.

The products comprised by OSGeo form a "stack"; data manipulation libraries
GDAL and GeoTools underpin web services interfaces Mapserver and GeoServer,
with GRASS and OSSIM providing heavyweight analysis and processing tools, and
Mapbender, Mapbuilder and OpenLayers offering a variety of web mapping client
functions, all oriented towards open standards; and MapGuide Open Source, the
more "all in one" open source geospatial project originating with Autodesk, and
most recently FAO GeoNetwork, the geospatial metadata catalogue service.

[picture of stack]

(These projects have a variety of backgrounds, from academic institutions to
"dot.org" social enterprises to the UN Food and Agriculture Organisation,
private consultancies and community cycling groups. They have many things in
common; their source code is openly published and once "incubated" from the
Foundation, guaranteed free of copyright and patent constraints, and available
for re-use without restriction. Developments in each project are overseen by an
active Project Steering Committee and subject to extensive peer review.

Open source projects are heavily involved in the open standards development
process as "reference implementations".) Also community involvement in
standards which hit the open source domain first, driven very directly by needs
of developers and immediate end-users, solving problems faster.  Strongly to
open source benefit to be 'interoperable' to have common interfaces. 
Emergent specifications focused on making it easier to contribute and 
redistribute geospatial data. 

Examples:

WMS --> TMS

[OpenLayers screenshot]
[GeoNetwork screenshot]

[redundant distribution - lessening infrastructure investment - 'bleeding edge' comes up with budget constrained things that work.]

GML --> GeoRSS
"long tail", 90% of applications, etc.
[Maybe a Mapbuilder/WFS-T/Geoserver screenshot?]

WFS --> WFS-Simple? 

ISO19915/FGDC --> ?

[uDig catalog screenshot]
[FAO GeoNetwork screenshot]

GeoRSS / WFS Simple etc advantages - ease of implementation + support. "Lego
model" No need for heavy lifting backend systems initially. 
Webserver orientation = low overhead, easy to distribute. 
When needed easy to build on more complex interfaces lego-wise.
Adaptability of component-style software architecture: start w/ PostGIS, add
GeoServer later.  

I want to talk about GeoNetwork a lot more because it is relatively unique and
represents an area where open source definitely has the edge on proprietary
software offerings when it comes to SDI implementation. Metadata collection,
management; registry + search of OWS services...

Metadata on both the bottom and the top of the open source geospatial stack.
Essential for managing any kind of repository as well as publishing. Metadata
is at the core of the INSPIRE directive and metadata access and search is the
only service that INSPIRE rules dictate must be offered to the public free of
cost.

Metadata extends a lot further than data about descriptions of geospatial
features. 80% of information collected by government has a spatial component.
Metadata is the main area in which "translation" and "transformation" problems
will arise, attempting to map between different taxonomies and cadastral models
used within each of the European member states. It is helpful to think of
INSPIRE as a building block for a more generalised data infrastructure. As
such, connection with related information exchange standards and interfaces is
of benefit. Idea of a "geospatial web" using existing frameworks, particularly
RSS, RDF and the semantic web, to publish data and metadata, achieving the same
ends as the web services comprising a Spatial Data Infrastructure.

Architectures of participation... ease of participation. Registering source
with resource like GeoNetwork. Filesystem feeder application, encouraging
"organic growth" of better information about data, more componentisation.

[screenshots]

Easy syndication of updates - RSS feeds or access to an OAI-PMH repository
holding references to data. Publish metadata it will get indexed, metadata
search services crawlers. Find other people doing your job for you. Become node
on GeoNetwork. Data on web more likely it will get used. Metadata is byproduct
of own internal systems management - needed to do job better - fact it is
available on web a sideeffect. But to really get value / knowledge from it this
way has to be in open.

Open data = similar ideas of collab data production. "Many eyes shallow bugs".
Contribution federated nearer to where data collection happens. Increased
timeliness through distribution. Prospect of interesting connections to
location-enabled services.

Openstreetmap example. Commercial operators admit it is viable. Local councils
start contributing data to it. Business value in exploitation of data.

[screenshot of Timisoara Free Map in progress in JOSM.]
[something from mapnik/OSMarender]

Business value in exploitation of data. Public data generation exponential more
value in general if open at the point of collection + initial distribution. See
table from Weiss paper. 

(Here in the UK the Office of Public Sector Information is working on RDF
project for publishing metadata about civic information. 'click-use' license -
'atomisation' whereby metadata attached to smaller items of information. Ease
of entry and encouragement of reuse both by other public sector bodies and by
small commercial enterprises. Making data available raw - with consistent
identifiers - and an open license. Machine-negotiable click-use licenses
planned. Big difference Ordnance Survey stance. General trend availability.
Look at GeoConnections - they also fund OSGeo projects. 

The trend in legislation and infrastructure is going to be to more open, so more open now saves time later.