 ccording
to a recent analysis of Essential
Science Indicators ,
the work of Dr. Arthur ter Hofstede achieved the highest percent
increase in total citations (Rising
Stars) in the field of Computer Science. Dr. ter Hofstede’s
current record in this field includes 32 highly cited papers
cited a total of 277 times to date. Dr. ter Hofstede is an
Associate Professor in the Faculty of Information Technology at
Queensland University of Technology in Brisbane, Australia, and
Co-Leader of the Business Process Management Group in the
faculty. In the interview below, he talks with in-cites about
his highly cited work. |
Would you give us some
background on your education and early research?
I studied computer science at the University of Nijmegen in
The Netherlands, and upon graduation I embarked on a Ph.D.
funded through a project at the Software Engineering Research
Centre in Utrecht. I received my Ph.D. from the University of
Nijmegen in 1993. In my early research the focus was on
conceptual data modelling, in particular formalization of NIAM
(now more commonly referred to as ORM or Object-Role Modelling)
and later the introduction of extensions to NIAM and the
development of a formal data manipulation language.
What do you consider the main focus of your research?
Working in the general field of information systems has
influenced me in the sense that on the one hand I became very
interested in conceptual modelling, where one abstracts from
technology-oriented considerations in order to focus on the
essence of the problem at hand, and on the other hand,
observing a fairly widespread lack of rigor in the definition of
methods and techniques. I became very interested in mathematical
foundations of (conceptual) modelling approaches.
|
|
|
|
 |
|
“The workflow patterns provided a first
comprehensive and systematic insight into workflow modelling
constructs and into modelling capabilities of various approaches
to the specification of workflows.” |
|
|
|
|
So if a common thread can be observed through the research
that I have been involved in, it is that on many occasions I
have been involved in the formalization and analysis of
specification techniques (existing ones or newly developed).
Over time my emphasis shifted from static aspects (data
modelling) to dynamic aspects (process modelling/workflow
management) as in the field of workflow management there still
was ample scope for significant foundational work. My research
agenda concerns the formal, conceptual, and technological
foundations of workflow management.
Your most-cited paper is "Workflow patterns," (Distributed
and Parallel Databases 14[1]: 5-51, July 2003). Is there a
reason this paper has attracted so many citations?
Yes, I believe it is a case of the right research at the
right time. There was, and still is, much confusion in the field
of workflow management, and more generally business process
management (BPM), about modelling concepts and their meaning.
This has resulted in many different proposals for modelling
languages, both from academia and industry. Many of these
languages are not well defined and it has been hard to
understand comparative strengths and weaknesses. The workflow
patterns provided a first comprehensive and systematic insight
into workflow modelling constructs and into modelling
capabilities of various approaches to the specification of
workflows. This work can provide, and in fact has provided,
assistance with the selection of an appropriate modelling
language or tool for a specific project as well as the
development of new languages and tools.
It can further be added that the work had the right mix of
authors and that it was presented in such a manner that it was
accessible to a wider audience than just academic researchers.
In addition, a web site around the research was set up in an
early stage, which helped considerably in the dissemination of
the work. Finally, I also think the paper has attracted a group
of researchers who have enabled us to extend this work over time
to have significantly more impact than was probably originally
anticipated.
Please tell us about your involvement in the Workflow Patterns
Initiative and the YAWL Initiative.
These are joint initiatives between Queensland University of
Technology (QUT) and Eindhoven University of Technology. I
manage them from QUT’s side and Prof. Wil van der Aalst manages
them from Eindhoven’s side. We, and more generally our groups,
have been collaborating for many years. The YAWL (Yet Another
Workflow Language) initiative started during one of Prof. van
der Aalst’s visits to QUT, where we developed a workflow
language aimed at providing comprehensive control-flow patterns
support. YAWL can be considered a reference language for
workflow specification. We wanted to demonstrate that it is
possible to provide comprehensive patterns support, and later,
that it is possible to build a support environment for such a
language. This support environment was made available as
open-source software and we have since had some fruitful
collaboration with industry. I believe that the YAWL environment
provides unique capabilities for dealing with workflow
specification, verification, dynamic workflow, and exception
handling.
Where have you taken this research since the publication of the
2003 paper?
Since the publication of the 2003 paper, the patterns
collection has been extended with patterns for data definition
and manipulation, for resource allocation, and for exception
handling. Also, the original collection of control-flow patterns
of the 2003 paper has recently been thoroughly revised,
formalized, and extended. As such the current collection of
patterns provides, I believe, a very thorough foundation for
workflow management.
As mentioned before, the original patterns collection also
provided the basis for YAWL, where Petri nets, a
well-established theory for concurrency, have been extended with
constructs inspired from these patterns.
If you are free to discuss them, please tell us about your
current projects.
In one of the projects we have recently started we are
looking at simulation. Now that the foundations of workflow
modelling have more or less crystallized, it is opportune to
provide support for situations where people would like to
examine the consequences of deploying certain workflows and
provide assistance with exploring various alternatives.
Workflows are typically quite complex; one cannot fully examine
them analytically, and therefore it is imperative to also have
access to simulation capabilities. This project intends to
examine the use of historical data to fine-tune simulations and
to provide support for simulations that take the current state
as a starting point.
Another project is in the field of configurable reference
process models. Basically the idea is that process model
variations can occur in different contexts and that making these
variations explicit results in models that are easier to
understand, deploy, and maintain. Challenges include ways of
making configuration manageable for domain experts, finding
proper modelling abstractions, and providing appropriate tool
support.
A third project that I would like to mention is conducted in
the context of the ARC (Australian Research Council)
Centre of
Excellence for Creative Industries and Innovation. In this
project we are looking at the potential of the application of
BPM techniques and technology in the context of the film and TV
industry, so as to provide benefits without interfering with the
creative process.
What are your expectations for this particular field in five or
ten years?
I would hope for more consensus about specification
approaches, for more genuine appreciation for the necessity of
proper formal foundations, and for more understanding of what
having a proper formal foundation really means. I am also hoping
for more emphasis on ease of specification, in particular the
removal of the need to understand low-level programming or
scripting languages when specifying workflows. It would be great
to have BPM environments with powerful modelling capabilities,
which are easy to use, and with sophisticated analytical
support, both at design time and at runtime.
Arthur ter Hofstede, Ph.D.
Queensland University of Technology
Brisbane, Australia
|
cites
