SD in IS Development - Transforming DFDs to SD models

[Bruce Campbell]

A number of people replied to me personally re my previous posting.
Rather than reply to everyone separately I will place both my MSc(Hons)
thesis, and a list of references on the JRCASE web page. This may take a
couple of days. I will advise those people who replied individually when
the site is ready.

Most of these people were interested in transforming data flow diagrams
(DFDs) to an SD model. There is no formal, or automatic, way of doing
this. However, if you observe both a Level 1 DFD and an SD model, both
consist of stocks and flows (processes). Flows in SD models are often
named to indicate that they are a process - a common naming convention
lists them as "doing" objects. A "doing" object is a process within a
DFD. Data flows within a DFD are often equivalent to either information
flows (connectors) within an SD model or a main chain. The relationship
between a stock and a database, or file, is obvious.

Just use a bit of imagination in the transformation. The main trick is
to only use a Level 1 DFD. Any diagram lower than this will contain too
much information (not aggregated high enough) to be usefull within an SD
model.

On a number of occasions I have developed a Level 1 DFD concurrently
with an SD model using the same object names. This makes it easy for
other people to see the relationship between the two. As previously
mentioned, the DFD can then be exploded to show more detail, while the
SD model can be expanded to include important feedback loops. In this
way it should be possible to model the IS at enough detail to be useful
in its development, AND model its likely behaviour within the
organisation in which it will be implemented.

Note that the above was outside my area of research - it was an
interesting observation whilst looking at something else. Consequently,
I have not used the technique in anger, although I have used it with
some real data. Theres probably a PhD in here for some enterprising
person.

Nilsson (1999) makes the observation that development within
organisations normally takes place at three levels - strategic, process
and system. According to Nilsson the development at each of these levels
takes place without consideration of development at other levels, with
familiar results to SDers. SD can be used during both strategic and
process development. DFDs can be used during system development. The
concurrent development of both an SD model and DFDs could provide a
link between these levels of development as recommended by Nilsson. This
is an hypothesis - I have not investigated it at all. A PhD for another
enterprising person. Whoever takes this one on should be well versed in
the effects of organisational politics!

I mention these areas of futher research as my contract at JRCASE is
coming to an end and I will have to go and find a real job. However,
both areas are worthy of investigation and I put them out in the hope
that someone will pick them up.

Regards,

Bruce Campbell

Nilsson A.G. 1999 in Nilsson A.G., Tolis C. and Nellborn C. (eds)
"Perspectives on Business Modelling: Understanding and Changing
Organisations", Springer-Verlag, Germany


--
Bruce Campbell
Joint Research Centre for Advanced Systems Engineering
Division of Information and Communication Sciences
Macquarie University 2109
Australia

E-mail: Bruce.Campbell@mq.edu.au
Ph: +61 2 9850 9107
Fax: +61 2 9850 9102