Database Reference
In-Depth Information
FUTURE WORK AND CHALLENGES
We are in the process of developing a workfl ow design tool that incorporates the ideas
we developed in this research with the notion of automating statechart analysis. The idea of
developing this tool is to allow the user to develop sophisticated workfl ow models without
having to be concerned with the underlying formalisms and algorithms we developed. In
theory, the tool will automatically fl ag potential workfl ow problems for the user to aid in
workfl ow redesign efforts. Here, we point out some of the challenges that we might face
and obstacles that we must overcome to realize such a goal.
Workfl ow verifi cation and validation are important topics in academia and industry
alike. Given a workfl ow design and a specifi cation, it is important to see if the workfl ow
design fulfi lls the requirements in the specifi cation. Similarly, the output of a workfl ow design
needs to be validated for the workfl ow to be ready for production. Considering the complex-
ity of today's business workfl ows, it would be extremely useful if the process of verifying
and validating workfl ow designs could be automated, or less-ambitiously, semi-automated.
However, the computer has certain limitations, particularly with algorithms, which must
be understood. The Church-Turing Thesis states that Turing machines precisely capture the
intuitive notions of algorithms. Turing machines, or any equivalent forms of computation,
have limitations, however. A well-known problem that does not have an algorithmic solution
is the halting problem of a Turing machine, which can be stated as follows: Given a Turing
machine and an input text string, it is not algorithmic to determine if the Turing machine
will halt on that input string. This important result in the theory of computation has serious
consequences. One of the consequences of the halting problem is that it does not have a
computer-based solution to determine if an algorithm possesses certain properties. Thus,
in general, it is hopeless to develop an automated software to accept a workfl ow design as
input and determine if the workfl ow design possesses certain nontrivial properties. However,
if we put certain constraints on the given workfl ow, then it would be possible to develop
an automated solution. Therefore, it is our job, as researchers, to determine the constraints,
or the bounds, that we must impose on the workfl ows for such an automated solution to be
feasible. This research, therefore, is a step in such a direction.
Of course we are speculating about the potential of our design tool until we can empiri-
cally validate it in the fi eld. As such, we intend to conduct an extensive case study of MWI
once our workfl ow design tool has been prototyped. It is hoped that additional case study
iteration will reveal the tool's capabilities in a more granular manner.
We intend to further explore business workfl ows at MWI and other organizations to
validate and extend our fi ndings. We recently visited (May 2001) an executive at MWI who
was not part of this study. The purpose of the visit was to initially verify the fi ndings that
we obtained from this study and discuss future work possibilities with MWI. The executive
we interviewed was very positive about our current fi ndings and has agreed to participate
in an extension of this study.
Our next study will focus on developing the prototype and further testing our theory
on a new workfl ow mutually agreed upon by us and the MWI contact. If we can replicate
the fi ndings we obtained in this study, it will greatly enrich context and theoretical validity.
As such, we hope to build a cumulative tradition over time.
Of course we realize that in-depth case studies tend to uncover many ideas, constructs,
and concepts that are unanticipated. Therefore, we will try to keep our study somewhat within
the scope of theory we have already generated to enable rigorous replication.
