Concepts for Automating Systems Integration
6.3.3 Object and information modeling
In describing the functions performed by a system, it is necessary to identify the objects and information on which
the system acts, and the objects and information it produces. When the process is decomposed into separate
activities implemented by separate components, further details of the objects and information, including new
intermediate objects, become a part of the specification. Thus models of the shared objects and information of a
system, called the universe of discourse of the system, become an important part of the engineering specifications,
and are especially critical to integration.
the system acts, and the objects and information it produces. When the process is decomposed into separate
activities implemented by separate components, further details of the objects and information, including new
intermediate objects, become a part of the specification. Thus models of the shared objects and information of a
system, called the universe of discourse of the system, become an important part of the engineering specifications,
and are especially critical to integration.
The universe of discourse has four elements: objects, relationships, properties, and operations. Various modeling
methodologies address some or all of these elements in somewhat different ways.
methodologies address some or all of these elements in somewhat different ways.
An information model, or conceptual schema, is a formal description of the possible states of the objects, properties,
and relationships within a system. Information analysis is the process by which these objects, properties, and
relationships are discovered. A data model is a formal description of an organization of information units
conforming to an explicit or implicit information model.
and relationships within a system. Information analysis is the process by which these objects, properties, and
relationships are discovered. A data model is a formal description of an organization of information units
conforming to an explicit or implicit information model.
Object-oriented analysis is the process of identifying the objects and operations in a universe of discourse, and then
classifying objects by the set of operations they support. The information units that are attached to an object, and
the relationships among objects, are then determined from the need to support these operations. An object model is
a formal description of the object classes and the operations they support, and usually includes the required
information units and relationships.
classifying objects by the set of operations they support. The information units that are attached to an object, and
the relationships among objects, are then determined from the need to support these operations. An object model is
a formal description of the object classes and the operations they support, and usually includes the required
information units and relationships.
The remainder of this section describes popular methods for information modeling and object-oriented modeling.
The Entity-Attribute-Relationship method
The Entity-Attribute-Relationship (EAR) me
thod [91] is the oldest accepted information analysis method. It places
each element of a universe of discourse into one of the following categories:
each element of a universe of discourse into one of the following categories:
· Entity: any interesting object
· Value: an information unit having a simple representation
· Relationship: an association between two or more entities
· Attribute: an association between an entity and a value
Entity types are distinguished by the set of attributes and relationships the member entities possess. More advanced
EAR models attach particular significance to the relationships "is a part of" and "is a kind of." The latter
relationship is also referred to as a subtype relationship.
EAR models attach particular significance to the relationships "is a part of" and "is a kind of." The latter
relationship is also referred to as a subtype relationship.
EAR methods were closely associated with the development of relational databases, and as a consequence, often
have problems representing multi-valued attributes and relationships (situations in which an entity may have the
same type of conceptual association to more than one entity or value), which cannot be represented by a single value
in a column in a relational table. This gives rise to value structures (sets or lists of values), and to reified
relationships (entity types which represent associations as objects).
have problems representing multi-valued attributes and relationships (situations in which an entity may have the
same type of conceptual association to more than one entity or value), which cannot be represented by a single value
in a column in a relational table. This gives rise to value structures (sets or lists of values), and to reified
relationships (entity types which represent associations as objects).
EAR methods lead to a number of model representation languages, both graphical and textual. Two of these have
been standardized and are in common use:
been standardized and are in common use:
· IDEF1-X [13] has both a standard graphical representation and a less frequently used textual representation.
The graphical format is most frequently used for the publication of models.
· EXPRESS [20] has both a standard graphical representation (EXPRESS-G) and a standard textual
representation (EXPRESS). The latter is far more commonly used, and in fact the graphical form cannot
capture all the relationships representable by the language form.
capture all the relationships representable by the language form.
Models in these languages represent a common form in which input and output information sets are formally
documented, particularly in the case of engineering specifications for manufacturing products and processes.
documented, particularly in the case of engineering specifications for manufacturing products and processes.
58