Crispin (et al):

The short answer is that we are still defining _the relevant part_ of the
API, and are looking for input from FISH.

The longer answer is:

THE UNDERLYING DATA STRUCTURE

-- Although it will be hosted at data.pastplace.org, it will be accessing
the same database as our Vision of Britain web site (which actually covers
quite a bit more than Britain by now). That means that the underlying
database distinguishes between two kinds of high level entities:

(A) PLACES: these are mostly but not necessarily "populated places", i.e.
towns, villages and broader regions. We see these as inherently a little
fuzzy, require them to have a point coordinate and do not allow them to
have anything more.

(B) (Administrative) UNITS: these exist essentially in law, and in
historical documents, and over time there has been a tremendous variety of
kinds of unit, so our typology currently includes over 200 kinds although
not all existing in England and Wales; our task is to record their actual
legal status, not try to assign them to some simplified classification we
invented. In principle, units usually had boundaries but there are not
necessarily maps of those boundaries and even if there are they have not
necessarily been digitised. Units often first appear in our system from
textual sources which contain no maps or other locational information at
all. Consequently, units may have associated polygonal boundaries; or just
a point coordinate; or no coordinate data at all (we do require them to
have at least one "IsPartOf" relationship with a higher level entity).

Places and Units are related in two distinct ways:

-- Units are named after places. Any substantial town in England will have
10+ units named after it: a Borough, maybe a Hundred, a Poor Law Union, an
Urban Sanitary District and so on. Although our gazetteer of places is not
that big, by now it contains just about everywhere in Britain which gave
its name to a parish, many different kinds of districts, most
ecclesiastical units (e.g. "Rural Deaneries") and so on.

-- Units contain places, although we can only establish this when we have
boundary polygons.

Both places and units are identified by numbers, not names. Each unit and
each place has a preferred name, but each unit can have any number of
other names; and each place inherits all the names of the units named
after it, and additional names harvested from historical gazetteers and
travel writing.

Although we are expanding it in various ways, this structure already
exists and work on the API is independent of the expansion work.

THE API

What we have implemented so far is a simple API for searching for places:

(a) You give us a name and we return all places with that name: the ID,
the coordinate and the preferred name. They are returned in descending
order of frequency of attestation.

(b) You give us a coordinate and we return the n nearest-neighbouring
places.

Those two calls are already working, but at an address we do not want to
publicise. Although the arguments are different, the result sets are
identical, and available as HTML, JSON or a simple XML format specified by
Open Street Map. We will also definitely be implementing calls which
return more information about a specific place or unit:

(c) You give us a place ID and we return a name, a point coordinate,
textual description from a 19th century gazetteer, a list of units named
after the place, and a list of units containing the place. One aim is to
follow a developing standard for "points of interest" data being developed
by Raj Singh of the Open Geospatial Consortium (nothing to do with WFS(G)).

(d) You give us a unit ID and we return preferred and alternate names, a
point coordinate if available, information on dates of creation, abolition
and legal status, and information on related units. One available format
will definitely be RDF, although most of the terminology and certainly the
unit typology will have to be our own.

For now, the main focus is sorting out the hosting environment so that all
this will be available at pastplace.org while running on the same server
as Vision of Britain, but what is obviously missing from the above is
direct searching for administrative units.

NB where the starting point is a name, you will be able to access admin
unit data: do a name search (a); select the right place from the result
set; request more information about the place (c); select the unit of
interest from the linked units. However, where you start from a coordinate
something else is needed. The simplest thing we could offer is a search
that returns ALL units which cover the specified location, sometimes with
dates at which that coverage started and ended, and we leave it to you to
decide which if any meets your definition of "a parish" or "a county".

There may be a case for more specialised APIs that return just one unit,
but (a) that still requires the system making the request to understand
our typology, and (b) I am nervous about ending up with something as
complex as the Geonames API:

http://www.geonames.org/export/ws-overview.html

You will realise from the above that once we have decided what those unit
search services should look like, the infrastructure will already be in
place to implement them pretty quickly. This all needs to happen in the
next couple of months.

Best wishes

Humphrey




On 24/05/2012 22:15, "Crispin Flower" <[log in to unmask]> wrote:

>Hi Humphrey 
>Your API sounds very interesting. If you'd like to describe it in more
>detail (perhaps off list) and point us at it I can quickly tell you
>whether it is likely to be useful to HBSMR users (in current or future
>versions). We can already pull in data from SOAP/XML web services so it
>shouldn't be too hard technically, if the user community wants it.
>cheers
>Crispin
> 
>
>________________________________
>
>From: Technical advisory panel to the Forum on Information Standards in
>Heritage. on behalf of Nick Boldrini
>Sent: Thu 24/05/2012 12:15
>To: [log in to unmask]
>Subject: Re: [FISH-TECHNICAL] MIDAS Heritage and Mandatory recording of
>Admin area
>
>
>
>Hi Humphrey
>
>in answer to your question
>
>Is there a demand from the HER community for a system that returns the
>identity of non-current administrative units in response to a grid
>reference?
>
>I would say yes, almost certainly.
>
>I would imagine something of this sort would also be of interest to
>County Record Offices
>
>best wishes
>
>Nick Boldrini
>
>Historic Environment Record Officer
>Durham County Council
>Tel: 0191 3708840
>Fax: 0191 3708897
>[log in to unmask]
>VPN 7777 8840
>
>NOTE: Durham County Council Archaeology Service is moving on 20th June
>2012 and the HER will be shut from 18th - 22nd June.
>
>2012 edition of "Archaeology: County Durham" is now available.
>
>**Now available ** Order your copy of "Faverdale, Darlington: excavations
>at a major settlement in the northern frontier zone of Roman Britain" by
>Jennifer Proctor.
>Both these publications and more are available direct from the
>Archaeology Section (Archaeology publications for sale - Durham County
>Council)
>
>Web: www.durham.gov.uk
>Follow us on Twitter @durhamcouncil
>Like us at facebook.com/durhamcouncil
>
>-----Original Message-----
>From: Technical advisory panel to the Forum on Information Standards in
>Heritage. [mailto:[log in to unmask]] On Behalf Of Humphrey
>Southall
>Sent: 24 May 2012 10:23
>To: [log in to unmask]
>Subject: Re: [FISH-TECHNICAL] MIDAS Heritage and Mandatory recording of
>Admin area
>
>I am someone who mainly lurks on this list, but my ears pricked up when I
>saw stuff about parishes being abolished, and for once I think I should
>jump in.
>
>
>As some of you know, I am the main creator of the Great Britain Historical
>GIS, which holds OLD administrative area boundaries. Up to now, this has
>been accessible mainly (a) via EDINA's UKBORDERS download system, which is
>open only to academics, and (b) via the web site A Vision of Britain
>through Time. However, we are currently putting in place a search API.
>
>Is there a demand from the HER community for a system that returns the
>identity of non-current administrative units in response to a grid
>reference?
>
>We did start discussing this with the Archaeology Data Service a couple of
>months ago, mainly in the context of historical counties, but they went
>dead on us after we pointed out that it was not just an issue of a single
>set of counties whose boundaries varied over time, there were multiple
>systems of county which existed concurrently, with a broadly similar set
>of names but often substantially different boundaries. One inevitable
>consequence is that a simple county name does not uniquely identify a
>particular administrative unit.
>
>My favourite example is "Cambridgeshire", where the apparently simple
>question "was Wisbech in Cambridgeshire" is anything but simple to answer,
>but similar issues arise with parishes. For the record, we have Hoar
>Cross, Staffordshire, as "a township in Yoxhall and Hanbury parishes", and
>an Ecclesiastical Parish after 1876. We have a point coordinate but not a
>boundary, although we normally would if it had been either a Civil Parish
>or an Ecclesiastical Parish.
>
>We are currently discussing the API mainly with people interested in
>demographic/genealogical records, although there is still the problem that
>most people, even people with expertise in those kinds of records, do not
>understand the gross complexity of British administrative geographies.
>
>Best wishes,
>
>
>Humphrey Southall