|
Call For
Papers 
There is a significant disconnect between the ideals of the
‘on-demand’ business and the way that we currently develop
information systems. Growing competition, globalization, increasing
regulation, merger and acquisition and innovative business models
require organizations to be flexible and adaptable. The ability of
existing computer-based information systems to meet the challenges
of on-demand business is, however, limited. Historically, business
models have been embedded in code that does not distinguish those
models from the assumptions and platform constraints of particular
technology approaches. In many cases, models are not semantically
well-formed: The description of business things and relationships
vary greatly across different systems and the code is, in fact, the
only representation of the model. In addition, in many legacy
environments the human ‘memory’ of the business knowledge has
evaporated as staff retire and/or move on to other jobs. The reality
of this situation is evidenced in the general cost profile of
systems development, where the bulk of spending and effort lies not
in the development of new systems, but in the (evolutionary)
maintenance, integration and interoperability of existing systems.
This situation provides a challenge to the generally accepted
approaches to systems development.
In response, model-driven development has emerged as a potential
means of divorcing business issues from the underlying technology
platforms, in a way that makes change more manageable. Model-driven
approaches see the primary system development artifacts only as
models and their transformations; the technological element of an
information system is simply generated from models. To help manage
the journey from ‘real world’ to code, the prevalent architectural
approach partitions models into those that are (a) computation
independent, (b) platform independent and (c) platform specific.
Such frameworks are not easily translated into practice however. As
an example, the computation independent layer is not well understood
but implicitly makes strong demands in relation to understanding
of the types of things and relations that exist in the real world.
Similarly, the transformations between models should be considered
as first class models in their own right and demand strong semantic
treatment. Across the board, models also need to be developed at an
appropriate level of (a) abstraction, (b) generalization and (c)
precision and accuracy. In practical terms, the approach therefore
requires significant thinking in relation to both model development
and the modeling process if it is to be of value to organizations.
|
