Models

The LongSpine spine relies on the Linked Data technology stack (see Principles > Linked Data) for technical interoperability with lots of other Internet-based data and an overarching model (next section) for conceptual alignment. Additionally, all LongSpine Datasets use elements from a set of background models (below) which afford them conceptual alignment with many other datasets outside the spine itself, such as those published by the Australian Government Linked Data Working Group.

Models sections

• ^↓ Modelling Paradigm
• ^↓ Overarching Model
• ^↓ Background Models
• ^↓ Component Data Models
• ^↓ Vocabularies

§ Modelling Paradigm ^↑

This project implemented a series of Semantic Web data models, all formulated according to the Web Ontology Language (OWL). This means data according to the models is presented in Resource Description (RDF) formats, can be stored in an RDF triplestore (a form of graph database), be queried using the SPARQL query language and can be joined to any other data that is also presented according to any of the very many OWL models in RDF.

OWL was chosen as the modelling system as there are very many public OWL models in wide use of relevance to this project's domain (e.g. the Organization Ontology) and OWL models are specialisable, extensible, technically interoperable (with any other OWL models) and data made according to them can be presented online as Linked Data allowing human- and machine-access across institutions.

§ Overarching Model ^↑

• LongSpine Ontology

The structure of LongSpine's Dataset and Dataset relations (Linksets) (see Principles > Datasets and Linksets) adheres to the LocI ontology however the content of the Datasets does not, since LocI is about spatial objects and LongSpine is not. LongSpine content adheres to the very general LongSpine Ontology, an informal outline of which is given in Figure M1.

Figure M1: The LongSpine overarching model

The LongSpine ontology can be treated as the simples or fall-back data integrator: all data from all Datasets and Linksets, regardless of their specifics, can be viewed as LongSpine ontology data and used accordingly. The LongSpine Ontology documentation gives an example of specialised CRS Ontology data viewed as generic LongSpine ontology data.

§ Background Models ^↑

There are a series of well-known OWL models that are used for concepts relevant to LongSpine. These include both the technical, structural, models of how to represent data elements generally and also the conceptual models of particular domains' concepts. The following models, with notes on their role, are used by LongSpine:

Table M1: Background models used in LongSpine

The models above build on each other with last model listed, LocI, being an integrator of many of the others. Due to the reuse of the LocI ontology, Datasets and Linksets in LongSpine are structured identically to those in the Location Index, even though their content is very different. This does mean that structurally, LongSpine is interoperable with LocI.

All of these models are loaded into the LongSpine Cache as a series of Named Graphs which means, like individual Datasets and Linksets, they can be selected for use, or excluded, within individual queries against the Cache.

§ Component Dataset Models ^↑

Each LongSpine component Dataset has its own data model - ontology - to formally describe its content in a manner that aligns with the LongSpine Overarching Model several existing Australian government datasets. Each Dataset's model has a namespace for all items within it that define the structure of the dataset. The namespace, a web address, acts as both a persistent identifier for the dataset model and also as means to access it and metadata about it. The namespace web address is called a "PID" - Persistent Identifier - and the PIDs listed below are issued by the Australian Government Linked Data Working Group who have a system in place for this very purpose. The 5 component dataset models are given in Table M2.

Table M2: LongSpine Component Dataset Models

See the Data Page for links to the data of the Datasets built according to these models.

§ Vocabularies ^↑

In addition to the data models listed above, models of government functions used by LongSpine are presented as regular SKOS vocabularies, meaning SKOS vocabularies without any properties beyond the SKOS model's properties and basic annotation properties mentioned in SKOS, such as Dublin Core Term's source property to indicate sources for terms. The vocabularies are listed below in Table M3.

Table M3: LongSpine Vocabularies