On Roles of Models in Information Systems - Intentional Perspectives on Information Systems Engineering - page 28

ambition is to provide for semantic preserving transformations, so that when new
concepts of new world views are introduced and defined relative to previously
defined concepts, new information which is relevant to the new world view may
be produced from what is already present in the data bases.
The realization of the digital system components require that the appropriate
domain models are expressed in the modeling languages of computers, in the form
of data structures and computer programs. But the associated domain systems must
usually also be expressed in the relevant domain modeling languages so that the
system models can be understood in the domain culture. Data system languages are
keys to realization of the digital system components and cannot be avoided. So we
need to have parallel specifications, in three different modeling languages.
The mathematical basis and the modeling tradition of the various modeling cul-
tures differ. For example, most modeling cultures in technology and science require
mastering of differential equations, while the mathematical basis of data systems is
discrete mathematics and mathematical logic. Information systems apply additional
modeling approaches from linguistics and sociology, e.g., speech act theory, and rest
on the associated mathematical basis.
There is no automatic translation among the modeling languages. So it is very
labor intensive to maintain several specifications when the systems are changed.
And change is the normal state-of-affairs: laws and regulations change, products
have to be changed to meet competition in the marketplace, new system versions
are made to meet the needs in new markets. The usual situation is that only the
executable specification is maintained. That is, the details of a domain system
model are found only in the data system specification. The negative consequences
for comprehension and validation are obvious. Even for systems of medium size
comprehension and validation is impossible. This is acceptable only when system
failures are of little consequence.
The deep penetration of computers in all realms of society necessitates a change
in approach whereby domain problems being expressed in domain oriented lan-
guage can be made to co-exist with specifications of data systems and information
systems without everything having to be translated into data system oriented spec-
ifications. IT concepts, tools and theory consequently have to be better integrated
with the wide spectrum of domain specific modeling concepts and theories. In order
to reconcile the various approaches to modeling it seems to be necessary that the
different specification languages are based on similar ontology, and that there exist
an understandable relationship between data and what the data stand for.
In information systems engineering we have three groups of concepts: for con-
cept modeling in general, for behavior modeling, and for data modeling. The
concepts of state, event and process are used for behavior modeling. These are
separate concept classes and represent the evolution over time of properties of con-
crete systems. These concepts are used for behavior modeling in all three modeling
realms.
Over the years many approaches to concept modeling and behavior model-
ing have been proposed. Some influential modeling approaches were proposed by
Wand and Weber (the Bunge-Wand-Weber model) [ 34] and by Colette Rolland, first