Control Maquinas

avoid “clippings”, a correct match

is to be found between the contents which have

to be inserted in .the model, and the “expressive

capabilities” of the language. The representation’s

quality therefore is a function of both the language

and the object being modelled. For example,

IDEFO does not support the description of

organisational structures, and process models in

general are not capable of representing “soft”

elements such as corporate culture, negotiation

processes and so forth.

(b) Distortions. The MLF may limit some aspects

of the modlel maker’s view and enhance

other ones. While clippings may actually make

the adoption of a MLF essentially incorrect, distortions

usually do not lead to such dramatic

consequences. Halwever, it is up to the agents

involved in modelling to be aware of distortions

and take them into account.

(c) Biases. The MLF may arbitrarily introduce

implications or “preconditions” in the model,

which influence the subsequent decision making.

Biases are in some sense inevitable, but being

conscious of them may help in choosing languages

or formalisms whose biases agree with the

decision maker’s fundamental choices. As a typical

example, structured methodologies promote

hierarchical views of the world, and their use is

appropriate if hierarchical structures are desired.

Three corresponding questions are therefore

to be asked when proposing the use of a MLF:

Question 3a. Is the MLF able to support the

description of aspects which are relevant in the

model being built?

Question 3b. Does the MLF introduce distortions

to the aspects which are relevant in the model

being built? Are the agents involved in modelling

aware of them?

Question 3c. Do the biases and “preconditions”

introduced by the MLF comply with, or contradict,

the way with which the model is to be used?

The availability of software tools for writing

and maintaining the model also is an important

topic in the modelling-as-communication process.

Writing, reviewing and validating a model is a

lengthy and time-consuming task, and using computer-

based editors instead of pencil and paper

allows to achieve results of higher quality and in

less time. The following question arises:

Question 4. Does the MLF allow the use of some

kind of software editor? Is it easy to use?

At last, the communication process requires

that the adopted MLF has to be well known to

both the model maker and to the decision maker.

This point is not trivial since the erroneous use of

the language’s rules may lead to inconsistent or

misinterpreted models and, consequently, to bad

decision-making choices. Closely connected to this

point is the “readability” of the models. It is

known that it is not easy for humans to concentrate

on more than a limited number of concepts

at the same time, and that therefore all represen288

P. Brandimarte, M. Cantamessa /Computers in Industry 25 (1995) 281-293

tations should comply with this limitation. Some

structured formalisms (e.g. IDEFO) take this into

account, while many others (e.g. low-level Petri

nets) do not, therefore leading to models which

become very difficult to understand, use and

maintain. The last question to be placed therefore

is:

Question 5. Given the complexity of the object to

be modelled, do both the model maker and the

decision maker master the MLF with sufficient

proficiency?

...