Identifying things

[Jeffery Candiloro]

G'Day all,

One of the issues that has emerged from our work on building the list of
representatives and elected bodies in Australia is a need for a
consistent identification system for them (I know Adam mentioned this to
a few people at the Hackfest).

I'd like to start a discussion about that with anyone who is interested
- and to begin with elected bodies and their chambers.

There is a range on information that needs to be expressed. This includes
* Level of govt
* Jurisdiction [1]
* Geographic bounds
* Chamber name
* Division name
* Year of addition (This would be the first year from which this data is
valid. A subsequent entry with the same details but a later year of
addition would supersede but not replace the entry)

We would also like to recognise the different objects - chamber,
division, ward, etc.

Right now we are leaning towards using URIs. So, for the Federal House
Of Representatives after the last general Federal election we would have
something like:

http://server/id/house/Federal/House of Representatives/2007

And the Federal electoral division of Lowe would be:

http://server/id/division/Federal/House of Representatives/New South
Wales/Lowe/2007

In this way the Senate would be:

http://server/id/house/Federal/Senate/2007

and the NSW block in the Senate:

http://server/id/division/Federal/Senate/New South Wales/2007

At a state level you could have

http://server/id/house/New South Wales/Legislative Assembly/2007

and

http://server/id/division/New South Wales/Legislative
Assembly/Strathfield/2007

Please note - the names used here are mostly placeholders. They may not
be ideal for the final specification (if the spec wants to easily
support things like having the paths be filesystem-friendly or what not).

If you have any thoughts on the above - or even on how we could best
have this discussion - please reply.

Cheers

Jeffery

[1] - http://www.agls.gov.au/agls/AglsJuri/

[Domain Admin]

I'd add to this introduction that this issue is of critical importance
right now, because for people like fixmystreet that want to consume,
say, a Geo API from someone else they need a way to join the
identifiers coming out of one service, with the data stored in their
own database, and to link THAT to other services like Hansard et al.

We've got a number of very promising different systems in play at the
moment, but if we try to start connecting them together things are
going to get slow and painful pretty fast if we aren't speaking a
common language.

Adam K

[Rob Manson]

Hi all,

On Wed, 2009-06-17 at 20:09 +1000, Domain Admin wrote:
> We've got a number of very promising different systems in play at the
> moment, but if we try to start connecting them together things are
> going to get slow and painful pretty fast if we aren't speaking a
> common language.
>
> Adam K

Adam, I couldn't agree more.

Based on this comment, the systems surrounding it, the Open Data model
discussions at #publicsphere and what I've observed in general I'd like
to kick start a group data modeling project.

http://wiki.openaustralia.org/index.php/Open_Data_Model

First I'd like to use this wiki page and this discussion list to build a
general consensus that this problem "needs" to be addressed and then,
how it relates to OA's goals or scope.

Then, I'd like to define the scope (ideally as comprehensive as
possible) for this Open Data Model.

Then, I'd like to engage as many directly impacted or related groups as
possible - e.g. myrepresentative (hey jefferyc), fixmystreet (hey jake),
the tweetMP team (if they're not already on this list) and the relevant
Federal, State and Local data model architects or custodians/librarians.

@reemski and @broccolini are two great examples here at the Local level.

I'm keen to use irc for real-time detailed discussions with the logs
made available at a consistent URL. Then bind this back to a mailing
list - either a thread here or a separate google group.

Then the agreements and decisions from this discussion can be codified
and documented in a wiki (e.g. the OA wiki if that's appropriate).

>From there we can generate usable schemas, libs and transformation
rules.

Obviously I don't believe we can mandate anything at all...but if we can
make a productive and useful start then I'm confident we can gather some
momentum and save ourselves and Australia in general a lot of effort,
money and confusion.

I hope you all think this is as important as I do...

roBman

[Craig Thomler]

Some questions to consider:

1) Electoral boundaries change through years (not at end of years) -
how does the history of an electorate get tracked over time if there
are two records in a year?
2) Electorates can be created and destroyed during a year - how are
these catered for?

Cheers,

Craig

[Jeffery Candiloro]

G'Day Craig,

Electoral boundary changes only become effective at elections - so there
is a defined date at which they start (and end).

This is something that is being addressed. The current thinking is to
have a an "effective from" date as part of the identifier for a
particular division/electorate. That date would most likely be just the
year with a letter increment if there is a second election within the
same calendar year (eg, /2007 and /2007a).

Any service would then need to maintain their own list of "current"
electorates.

Cheers

Jeffery

[Craig Thomler]

FYI Ben Raue of Tally Room is contactable via Twitter - http://twitter.com/benraue

Cheers,

Craig