Information Modeling
The following diagram illustrates the relationships between information modeling concepts and the core structures provided by the Metaschema Framework.
An information model is an abstract representation of information from a given information domain. An information model is composed of a set of semantically well-defined information structures that each represent an information element in the information domain.
The primary purpose of the Metaschema Framework is to support the structured expression of an information model, which is represented as a Metaschema module. A Metaschema module is used to represent the whole or a part of an information model for a given information domain in an information-centric, format-neutral form.
Graph Theoretical Basis of Metaschema
In a Metaschema module, an information model is represented as a cyclic directed multigraph.
Information elements are represented as the nodes of the graph, which are connected by edges that represent the relationships between nodes.
A Metaschema module is a directed graph, since the edges represent a set of ordered node pairs. This allows a Metaschema module to represent sequences of relationships. In the example above, a
has two ordered relationships: a
-e1->
b
and a
-e3->
c
.
A Metaschema module is a multigraph, since two nodes may have multiple edges representing distinct relationships between the two nodes. This allows a Metaschema module to represent different relationships between the same two information elements, such as the relationships e3
and e4
between a
and c
below.
A Metaschema module is a cyclic graph, since a sequence of nodes and edges can form a circuit where the first and last nodes are the same. This allows a Metaschema module to represent recursive information structures. This is illustrated below where a
-e1->
b
-e2->
c
-e3->
a
form a cycle with node a
is the first and last node in the cycle.
In a Metaschema module a node is represented as a definition. Each definition represents a given information element in an information model. Each definition contains data and documentation about the meaning (semantics), structure (syntax), and use of a given information element.
In a Metaschema module an edge is represented as an instance, which establishes a relationship between two information elements in an information model. Each instance contains data and documentation about the relationship, including data about the cardinality, meaning (semantics), structure (syntax), and use of the relationship.
While a Metaschema module represents a cyclic directed multigraph, there are a few assumptions that ensure that data representations based on the information model conform to an oriented acyclic directed multigraph. This is necessary to support JSON and XML data representations which use a tree form.
- At least one root node MUST be defined in the information model. A root node ensures that orientation is supported, providing a root of the data tree. Multiple root nodes are also possible allowing multiple resulting oriented acyclic directed multigraphs to be produced.
- Nodes involved in a cycle MUST allow for cycle termination. This can be achieved by a zero minimum cardinality on an edge creating a cycle.
Object-Oriented Basis of Metaschema
In a Metaschema module, the cyclic directed multigraph representing an information model is represented as a hierarchy using a compositional approach to describe information elements. This approach aligns well with the hierarchical form of JSON, YAML, and XML; and also supports programming language data structures that are based on object-oriented principles.
In object-oriented programming languages, a class can be used to represent an information element. An object represents a data instance of a class, which can also be considered an instantiation of an information element. Each composite member of a class can be used to represent a relationship to an object of another class. Thus, a class member represents a specific type of relationship between two information elements. This compositional form is often referred to as a has-a
relationship.
For example:
In this example the Computer
class establishes an aggregate compositional relationship consistsOf
to a ComputerPart
class. This is a way of stating that a computer is made up of a set of parts.
In a Metaschema module, the information element concepts of a Computer
and a ComputerPart
can be represented as a definition. Thus, a definition represents a class. The consistsOf
relationship is represented as an instance. Thus, an instance represents a class member.
The following illustrates how the object-oriented approach used to describe a computer can be represented as a cyclic directed multigraph.
The duality of the Metaschema module as both a representation of a cyclic directed multigraph and a set of object-oriented classes is explored in this specification.
Definitions, Instances, and Local Definitions
Metaschema uses 3 types of definitions to represent information elements with different structural shapes: define-flag
, define-field
, and define-assembly
. These definition types are used as building blocks of a Metaschema-based model.
Field definitions and assembly definitions allow the identification of edges through the declaration of an instance. An instance represents an edge between the containing definition and another definition. Thus, an instance makes use of another definition, typically by reference.
Both field and assembly definitions optionally allow the inclusion of one or more child flag instances.
An assembly definition also has a complex model which contains a sequence of model instances, each model instance is either a field instance or an assembly instance.
Within a Metaschema module, the information model implementation consists of assemblies, each of which are composed of more assembly, field, and flag instances.
Field instances and assembly instances support the composition of complex information elements.
Flag instances may exist on fields and assemblies, providing identifying or qualifying data about their containing definition.
The following example illustrates the use of each type of definition, and the use of flag instances and model instances to create a more complex model through composition.
The example above declares 4 distinct object definitions, along with their instances.
- The flag definition
flg-def-1
represents a reusable flag. - The field definition
fld-def-1
represents a reusable field. - The assembly definitions
asmb-def-1
andasmb-def-2
represent reusable assemblies.
Through the compositional use of flag instances and model instances, it is possible to build a complex information element by declaring how smaller information elements are combined together through composition.
In the example above, the assembly definition asmb-def-1
and the field definition fld-def-1
both instantiate the flag defined as flg-def-1
. These instances, flg-inst-1
and flg-inst-2
respectively, are examples of flag instances. The assembly definition asmb-def-1
declares the flag instance flg-inst-1
as a composite child by referencing the flag definition flg-def-1
. Similarly, the field definition fld-def-1
declares the flag instance flg-inst-2
as a composite child by referencing the flag definition flg-def-1
.
Furthermore, the assembly definition asmb-def-1
has a complex model that instantiates the assembly definition asmb-def-2
, as asmb-inst-1
, and the field definition fld-def-1
, as fld-inst-1
. These are examples of model instances.
Assemblies and fields also allow inline definitions to be declared which represent a single use definition that is also an instance. In these cases the inline <define-flag>
, <define-field>
, and <define-assembly>
elements are used, which combine the data elements used to declare a definition and instance of the same type.