MetaFrame in Practice: Design of Intelligent Network Services - Correct System Design: Recent Insights and Advances - page 399

4.3 Tailoring of the Visible Complexity

The described view concept provides all user groups of the development envi-
ronment with a tailored representation of the system under construction. These
representations, which arise as abstracted versions of the overall system repre-
sentation in terms of a flow graph-like structure, are formally derived on the basis
of hiding, abstract interpretation, model collapse and renaming concepts. The
whole system development can be reduced to stepwise modications of adequate
role- or topic-specic views, like the error view shown in Fig. 4.

4.4 Semantics-Based Control

Systems, as well as their underlying domain model, can be loosely specied in
terms of modal-logic constraints [3]. provides a model checker for

the corresponding verication, and a synthesis tool for guiding the development
process [25]. Design errors like those spotted in Fig. 4 are in fact automatically
detected on-line within seconds, even for libraries of hundreds of constraints.
Important feature is here the full automation of these tools, which are indeed
protably used without requiring any specic knowledge of the underlying formal
methods.

M eta Frame

4.5 Incremental Formalization

The key to the acceptance of our system development method in the industrial
context was its incremental formalization character [21, 6], ranging from `no ex-
tra specication', resulting in the old-fashioned development style, to `detailed
specication', exploiting full tool support. In fact this property was probably
psychologically most important for the acceptance of the tool within the SNI
designers group and by their customers, because it builds upon familiar devel-
opment habits.

4.6 `Evolutionary' Application Programming

The idea to this development style, which provides non experts with restricted
programming power, rst came up in connection with the ETI platform [19, 7],
but turned out to be nicely applicable to the IN scenario (cf. Section 3.6). The
proposed combination of inputs in terms of user dialogues , controlled extension
by means of our synthesis feature, and validation in terms of animation guaran-
tees reliability while allowing service subscribers a flexible and intuitive means
to customize their services, including modications of the Service Logic. This
flexibilization and simplication of the service customization process reduces
the costs for tailored intelligent network services and therefore provides a key
to a service-on-demand market. Currently, we are exploring these concepts as a
powerful means for personalizing internet services.