Message confusion

[davaya]

The Introduction to NIEM Common Model Format discusses the N-squared problem as "a multitude of technologies requiring peer-to-peer conversions between technologies". But the technologies listed (RDF, SVG, UML, XML, SQL, JSON) are a mixed bag of "kinds" that don't support direct conversions. SVG is primarily a presentation format, JSON and XML are primarily data formats, RDF is primarily a knowledge graph format. What does it mean to "convert an SVG instance into an RDF instance" - what information does the input represent and what does the output represent?

Entity: Bob is a person with DNA and life experience
Logical/Knowledge Model: Defines a Person type with attributes: appearance, accounts, credit cards, a passport, family, employers, friends, plus an unlimited set of types that can refer to Person
Logical Instance: Bob is an instance of the Person type plus instances of Person and other types that can reach Bob through graph edges
Information Model: Defines the content restrictions and structure of a Document / Message independently of data format
Information Instance: a finite and immutable subset of Bob's logical instance plus additional information not included in the logical model. Represented as an abstract syntax tree (AST).
Document or Message: the result of serializing an Information Instance (AST) into a sequence of bytes in a specified data format, or the source byte sequence from which the identical AST can be parsed.

Nodes in a knowledge graph are not documents, but they can be serialized into documents. A document is not a node in a knowledge graph, it is an instance of a datatype (a data entity) that can be described by a node just as physical entity Bob can be described by a node.

(Note that, unlike an RDF container, the RDF "collection" type is not a graph node. It is a datatype that is identified by its value and can thus be used to define an information model.)

The diagram should clarify that the things translated across formats are documents / messages defined by Information Models. Those messages may, but are not required to be, described by knowledge graph nodes. They may carry but are not restricted to include only serialized node values. For example a passport photo may be referenced by a knowledge graph and carried in a message, but the image document is not required to be defined by CMF or translated across formats.