Information Technology Reference
In-Depth Information
with the various requirements of describing actual agent-based models.
Furthermore, a formal specification should not impose unnecessary
limitations or hurdles (cp. [69]). A current overview of formal specific-
ation approaches is given in [11].
Within the M&S-community, a number of modeling formalisms
are available, like e.g., process calculi, timed automata, DEVS or
(labelled) transition systems. Currently, widely accepted standards
for describing reference models are not available but an open research
topic [34, p. 13ff.].
As important as the intended purpose of a formalization is to know
the activities which the chosen formalism has to support. Therefore, a
second reason for formalizing a reference model would be the intention
to support some kind of formal verification or automatic processing
(e.g., reasoning, inferencing). Also, direct transformation of a formal
specification into executable code is of interest. However, currently
no such activities are intended.
With reference to the intended purpose and an addressed modeling
level of (primarily) the conceptual model, a formal specification is
not considered a must-have requirement in this thesis. This does, of
course, not imply that a reference model is allowed to be inaccurate
or ambiguous. On the contrary: a reference model should strive for
maximum clarity and exactness. Therefore, parts of a reference model
may be formalized (where appropriate) using a formalism suitable for
this purpose.
Furthermore, a reference model may provide suggestions which
parts of a problem-specific model should be formalized and which
formalisms might be suitable. Future revisions of a reference model
should built upon first experiences gained with the current reference
model and strive for a more formal specification.
