Copyright 1998 Nikolas S. Boyd. Permission is granted to copy this document provided this copyright statement is retained in all copies.
The role of graphical conceptual models is similar to that played by object models - they allow one to grasp several concepts and the relationships between them at a glance. So, while such pictures are not exactly worth "a thousand words," they do help one to visualize (sometimes complex) relationships between concepts that might otherwise be difficult to integrate and understand from a purely textual description.
There are several elaborate graphical notations that could be used for conceptual modeling, ranging from those used in the past for conceptual graphs [Sowa 1982] to those presently used for object modeling [UML 1997]. However, if lay people cannot readily understand the graphical models we produce, they cannot validate whether our models are correct. So, this paper introduces a new graphical language for building conceptual models.
Natural Conceptual Modeling Language (NCML) was intentionally kept very simple, addressing only the primary syntactic elements of natural languages - nouns, verbs and the prepositions associated with verbs. Because of its relative syntactic transparency, NCML can be used to closely represent the semantics of simple sentences from any natural language that contains nouns and verbs.
The elements of NCML are labeled rectangles and labeled arrows and lines. Labeled rectangles represent sentence subjects and objects. Labeled arrows represent verbs and labeled lines represent prepositions that contribute to the predicates of the verbs. This notation serves as a reminder of the metaphoric relationship between natural language and the object-oriented paradigm - the sentence subject (or an object) may later become the receiver of an object-oriented message.
The elements
of NCML may be combined to create graphical models that represent simple
sentences. The most frequent kind of English sentence has a simple structure -
subject verb object (SVO). This kind of sentence identifies a binary
relationship - i.e., a relationship between two nouns - the subject and the
object.
English
sentences can also express complex relationships between a subject and multiple
objects. Such sentences identify n-ary relationships - e.g., relationships
between three (ternary), four (quaternary), or more parts. This kind of
sentence contains one or more prepositions that contribute additional nouns to
the predicate of the verb. When the verb identifies an action initiated by the
sentence subject, the additional nouns may become the arguments of an
object-oriented message. When the verb identifies a static (often structural)
relationship between the subject and the objects, the nouns may become the
fields of a record in a relational table or the fields of another object -
i.e., when the verb can be transformed into a noun through nominalization.
Some ideas
include or subsume one or more parts. There are two kinds of hierarchical
organizations for such conceptual inclusion.
For the sake of notational simplicity, distinction between the kinds of hierarchies are deferred until object-oriented analysis. So, both kinds of inclusion are depicted using nested rectangles, where the rectangle(s) that represent the included concept(s) are nested inside of the rectangle that represents the inclusive concept.
Another
kind of conceptual inclusion exists, namely abstraction. Some concepts can be
defined in terms of relationships between other concepts. Such abstraction can
be especially useful when breaking complex sentences down into simpler ones.
Some sentences contain complex combinations of dependent and independent
clauses, especially sentences that contain conditions that express invariants,
constraints, rules, etc. Abstraction provides a way of representing a
relationship or a complex set of relationships as a single concept. For
example, consider the following abstraction. Here, the verb "employs" has been
nominalized by the -ment suffix, producing the noun "employment." The
employment concept abstracts the relationship "employer employs
employee."
[Sowa, 1984] J. F. Sowa. Conceptual Structures: Information Processing in Mind and Machine. Addison-Wesley, Reading, Massachusetts, 1984.
[UML, 1997] Grady Booch, Ivar Jacobson, James Rumbaugh. The Unified Modeling Language. Rational Software, 1997.