"Seegrid will be due for a migration to confluence on the 1st of August. Any update on or after the 1st of August will NOT be migrated"

Data Model



At this stage it has agreed that the data model for the gazetteer will be based as closely as possible on the ISO 19112 standard. However it is recognised that for pratical functionality purposes the model will need to be amended to cope with requirements.

-- StuartGirvan - 18 Nov 2005

"Core" Place Names model

ISO 19112 standard

The core of the Place Names data model for the gazetteer will be based on the ISO 19112 standard.

The object of most immediate relevance to the project is SI_LocationInstance. That object has the following attributes and definitions:

SI_LocationInstance attributes according to ISO 19112

Attribute Meaning Cardinality Mandatory
geographicIdentifier Name/Unique Identifier 1 Yes
alternativeGeographicIdentifier Alternative Names 0..n No
geographicExtent Extent of geogrpahic feature 0..1 No
temporalExntent Usualy the date the entry was created 0..1 No
administrator Name of the organisation respnsible for the location instance info 1 Yes
position Coords of a representative point for the location instance 1 Conditional - must be there if geographic identifier contains insufficient information to identify location

The model for location instances also includes the ability to record parent/child relationships between location instances.

Context of the SI_LocationInstance and Issues With the Model

1) Resolving the place name uniquely

The ISO 19112 model describes a system for gazetteers that infers that each geographicIdentifier must be unique within its own context or at least can be made unqiue with the inclusion of a parent instance. In the case of Place Names, there are no real parent-child relationships as everything is simply a "place name", which is a very abstract concept.

This provides us with a bit of a problem. We know that there are many duplicate names in the place names gazetteer and the only way to differentiate between them, besides coordinates, is by an ID. There are a couple of ways of solving this:

  • We add an extra attrbiute to our version of the model for a unique ID or

  • We use the ID as the geogrpahicIdentifier attribute (and there's nothing wrong with this) and simply put the common name into one of the alternativeGeographicIdentifier attrbutes

The more I think about it the more I'm starting to like the second option.

  • < IvanPrice - 28 Dec 2005 -- ok, but the unique id will only be used in the GA emalgumated version of the data cause the custodians don't want to have to adopt a new id and the avoidance of collisions etc is a problem.. so do we have an id field that is custodian-prefixed.. e.g. NT1 and NSW1 for placename_id=1 for both custodians.. this way the ids will be as permanent as the custodian own ids.. a bit yucky but maybe necc with the lack of comms betwen the various placenames maintenance systems across the country.

  • the problem however with using the primary english name in the alternative name slot is that how do you designate the primary versus the alias names ? it would be confiusing to have a schema where the primary name is inside the alternate_names 'tag' where primary_name=1 or something.. -- IvanPrice - 28 Dec 2005 >

Either way, if we are to have interoperability between the jurisdictions we need to make sure that we are unqiuely identifying the points in the same way. Lynette would you be able to put a comment here about how GA comes up with the IDs it uses in the national gazetteer?

2) Geographic Extent

In the compiled national gazetteer extent is not recorded and from this I'm guessing that it is not recorded consistently around the jurisdictions. As such we will not be able to able to populate the field. Or at least not with anything more meaningful than "Uknown" or "Unspecified".

What we can do is have that information provided in the responses by the jurisdictions that have that data (see UseCases). This may even be used to update the map graphic - but that may be later down the track.-- StuartGirvan - 24 Nov 2005

  • < IvanPrice - 28 Dec 2005 -- i'm happy for the extent to be contributed as a point (identical to the position) until we get sophisticated enough to provide polygons.. this way its a custodian issue and when the data is ready the model doesn't need to be changed. -- IvanPrice - 28 Dec 2005 >

3) Locaction Types or Feature Types

The 19112 standard defines the properties of a Gazetteer (a gazetteer is just a collection of many location instances, but the gazeteer itself has its own meta data eg custodian, scope, etc) and one of the most important is Location Type. Location Type appears to be equivalent to Feature Type. The model appears to infer that a location instance gets it's Location/Feature Type from the gazetteer it is held within. However the model apperas ambiguous because a single gazetteer can cover many Location/Feature Types - and in that case, if the Loctaion/Feature Type is not recorded against the individual location instacne how would it know what Loctaion/Feature Type it was? I think we need some more info on this.

Feature Type is a vitally important attribute for all UseCases because much of the end user searching is performed by Feature Type (eg people often want to look up road names or mountain names) and it is also important for display (people want to know wheter they're looking at a town, road or mountain).

Either way we must record the LocationType/Feature type as part of a loctaion instance.

We can again base what is needed to be recorded about a Location Type/Feature Type on the 19112 standard. The key elements are a name, theme, identification (ie by code or name), definition, territory of use and owner. Location Types/Feature Types can also have parent/child relationships - this may not be crictical in the first stages of this project but may become useful when we start to think about more complex uses of the gazetteer data.

The feature type list will be taken from the aggregated master list that currently contains NSW, VIC and GA data. As an action for this project it will also included dat afrom the NT and AAD.

Target Core Model Attributes

Location Instance (Place Name)

Attrbiute Meaning Cardinality Mandatory
id (? dependent on whether we need it or use geogrpahicIdentifier) Unique Identifier 1 Yes
geographicIdentifier Name/Unique Identifier 1 Yes
alternativeGeographicIdentifier Alternative Names 0..n No
geographicExtent Extent of geogrpahic feature 1 No - can't guaranteer everyone will have this data
temporalExntent 1 Usualy the date the entry was created No
administrator Name of the organsiation respnsible for the location instance info 1 Yes
position Coords of a representative point for the location instance 1 Yes
featureType Description of a feature type for the location 1 Yes

We know that there will be cases where alternativeGeographicIdentifier will be used many times and that it will also have clarification against what type of name it is (eg historic, ethnic etc), so that needs teasing out.

Location Type (Feature Type)

Attrbiute Meaning Cardinality Mandatory
name Name for the type of Feature 1 Yes
theme Property used as defiing characteristic eg administration, electoral, postal 1 Yes
identifier Method of unquely identifying location instaces, eg code or name - in our case name has been agreed upon 1 Yes
definition the way in which features are defined eg by area, single feature or collection of smaller features. Ours will be by single feature 1 Yes
territory of use Geographic area within which the location instances are defined 1 Yes
owner Name of the organsiation or class of organisation responsible for feature type 1 Yes
parent location type name of parent location type 0..n No
child location type name of child location type 0..n No

Initial Schemas for Place Names Gazetteer

Rob Atkinson has delivered the informational models and schemas for the gazetteer.

  • CGNA.xsd: Schema for CGNA Place Names Gazetteer

This is the schema with which Place Name GML documents produced by CGNA members should conform to. It may require a few changes once we begin testing in anger.

The second document is very handy for understanding exactly what an instance of a place name would look like - see CGNA_LocationInstance in the document.

Alternatively look at the PrototypeResults page.

-- StuartGirvan - 10 May 2006

Attributes Wishlist

What will the comprehensive model look like and include? Please jump in here guys and add all the things you'd like to see in the extended place names model.

-- StuartGirvan - 21 Nov 2005

Here's the extended Place Names data model from NT Place Names Draft (thanks Ivan). It would be worth AAD and NSW having a look at this and adding anything they'd like covered that is not there.

-- StuartGirvan - 22 Nov 2005

Additional attributes from NT draft model:-

  • suffix
  • place_status
  • history_origin
  • last_updated
  • last_updated_by

Place Alias which is probably equivalent to alternative geographic identifier has
  • alias_type
  • variant_name
  • verified_date
  • comments

Place LGA
  • council

Other fields/objects that may be of interest
  • reference, which has a whole refernce schema associated with it
  • pronunciation
  • ethnicity
  • community, which has a whole additional schema associated with it

-- StuartGirvan - 05 Dec 2005

  • < IvanPrice - 28 Dec 2005 i'll get stuart to see if there's anything else we'd like to see in the data model.. otherwise as long as we are no querying on thes eaddional attributes can we not add them later as optional elements ? -- IvanPrice - 28 Dec 2005 >

  • CGNA.xsd: Schema for CGNA Place Names Gazetteer

Topic attachments
I Attachment Action Size Date Who Comment
CGNA.xsdxsd CGNA.xsd manage 5.7 K 10 May 2006 - 12:38 StuartGirvan Schema for CGNA Place Names Gazetteer
cgna_gazetteer.xsd.htmlhtml cgna_gazetteer.xsd.html manage 163.2 K 10 May 2006 - 12:39 StuartGirvan Documentation for model and instance examples
Topic revision: r13 - 15 Oct 2010, UnknownUser
 

Current license: All material on this collaboration platform is licensed under a Creative Commons Attribution 3.0 Australia Licence (CC BY 3.0).