Project Title Project Summary Project Plan Project Participants
Ontologically-based Evaluation, Comparison and Engineering of Integrated Process Modelling Techniques
Integrated process modelling techniques such as UML and ARIS form the conceptual platform for many management and IT projects. Though most IS development tools contain these techniques, anecdotal evidence indicates many shortcomings. This project uses a well-established theory developed in philosophy and applied in information systems domains for the evaluation of these techniques. The expected outcomes are evaluations of ARIS and UML. Thus, this project contributes to the development of two of the most popular modelling techniques. Based on the theory used and the results of an international empirical study, suggestions for the further development of these techniques will be derived.
Project Motivation and Objectives
Process models form the conceptual basis for a wide range of organisational
and IT projects. Thus, contributions that try to improve the quality of process
models are of immediate significance and importance for many approaches such as
the management of enterprise systems, workflow management, or systems
development. The proposed research will start with the development of a meta
model for the constructs and relationships of the BWW models. This work is the
platform for this research as well as already providing a contribution of
general importance. A more easily understood meta model will help to spread the
idea of an ontology and it will advance the knowledge base of disciplines using
ontologies.
The planned outcomes of this project will be the first comprehensive ontological
evaluation and comparison of two popular grammars for integrated process
modelling. It will constitute a theoretically founded state-of-the-art
evaluation of two widely used modelling grammars.
The innovation of this project is not only seen in the domain of process
modelling. The applied methodology is also seen as unique and novel. An
ontological evaluation based on meta models is seen as a more rigorous and
transparent approach than existing methodologies. In project reviews, it will be
possible to evaluate easily the specification of the BWW models and the selected
modelling grammar using these meta models. Furthermore, the comparison will
become transparent as corresponding relationships become obvious. Ontological
weaknesses such as overload, excess, or redundancy will be easy to identify
based on the lack of corresponding constructs in these meta models. Overall, the
meta model approach will simplify the understanding of the BWW models ensuring
useful educational and research outcomes. Such a meta model allows also critical
evaluation of the BWW models and identification of such aspects as inherent
weaknesses in the BWW models. Finally, it will be possible to derive new or at
least modified modelling grammars based on this meta model. Figure 2 summarises
the advantages and application areas of such a meta model. The approach to use a
comparison of meta models will be intensively documented and might stimulate
research in related areas in which such a methodological approach could be
appropriate.
Figure 2: Application areas of a meta model for the BWW models
The proposed research will be divided into three main phases, viz.,
evaluation, comparison, and engineering of integrated process modelling
techniques.
In the first phase, Evaluation, this work will be based on, and continue, the
research of Green and Rosemann [2000]. The first activity will be to convert the
existing parts of the ER-based meta model for the representation model into a
UML model. This activity will be followed by the completion of the meta model
for the entire representation model as well as the design of meta models for the
state-tracking model and the good decomposition model. This meta model is
targeted for completion within the first six months after the project start. The
critical review of the meta model will be the first milestone of this project.
This completed meta model will form the basis for an ontological evaluation of
ARIS. As ARIS is already documented in a UML-based meta model (Scheer 2000),
this task will include the bi-directional mapping of constructs and
relationships of the BWW model and ARIS. Initial work done by Green and Rosemann
[under review with Information Systems] will serve as input in this phase.
However, a detailed and complete evaluation has not taken place yet. The outcome
of this phase will be an evaluation of the ‘ontological goodness’ of ARIS
including a structured list of ontological weaknesses.
Subsequent work will focus further on these identified weaknesses. These
weaknesses are based on the application of a generic ontology while ARIS was
developed for the specific needs of business modelling. Thus, it is a consequent
next step to empirically test the practical relevance of these weaknesses. For
these purposes, propositions and hypotheses will be derived from the outcomes of
the analysis. These hypotheses form the structure of a questionnaire and the
basis of an international empirical study. Intensive and experienced users of
the ARIS-Toolset will be identified from IDS’s customers. They will be
approached to participate in this study. In addition to the questionnaire,
selected interviews will take place with experienced international users. The
consolidation of the identified ontological weaknesses together with the results
of this empirical study will critically be reviewed in the second milestone of
this project. They also will be discussed with August-Wilhelm Scheer, the
developer of ARIS. The complete analysis of ARIS will take nine months.
Based on the experiences gained through the analysis of ARIS, a similar
ontological evaluation of UML will take place. The developed BWW meta model will
be applied to the meta model already available for UML - the de facto standard
for object-oriented modelling. Again, the identified weaknesses will be
consolidated and propositions will be derived. An empirical study using a
questionnaire with a structure corresponding to the one used for the ARIS study
will be distributed. The participants for this study will be users of the
modelling tool Rational Rose, developed and distributed by Rational Technology.
Ontological weaknesses and empirical results will be consolidated and critically
discussed with the project advisors, Yair Wand and Andreas Opdahl. It is planned
to spend nine months for the entire analysis of UML.
Figure 2 depicts these project activities and the proposed milestones for the
first two years of this project. All these activities can be summarised as
ontologically-based evaluation of integrated process modelling techniques.
The next main phase, Comparison, will involve the ontologically-based comparison
of integrated process modelling techniques. In this phase, the results of the
ARIS and UML analyses will be consolidated. Ontological weaknesses that both
grammars have in common will be identified. As this task might indicate general
ontological weaknesses, further selected studies in other process modelling
techniques (e.g. Petri Nets) might verify this situation. Furthermore,
differences in the results could be used for a study into the individual
advantages/disadvantages of ARIS and UML respectively. However, of equal
importance is the potential finding that common so-called "weaknesses" of ARIS
and UML might indicate shortcomings in the BWW models. Thus, to what extent
these results might motivate a further development or a modification of the BWW
models will be investigated. This second phase will have a duration of four
months.
The final phase, Engineering, will cover the ontologically-based development of
integrated process modelling techniques. The results up to this phase will serve
as input for the selective development of improvements for ARIS and UML. These
suggestions will be measured by the degree to which they help to overcome
identified ontological weaknesses that were perceived as relevant in the
empirical studies. The last eight months of this project will be invested in
performing this third phase.
Procedures. On the basis of the analyses of ARIS and UML, hypotheses will
be developed for each modelling grammar respectively. These hypotheses will then
be translated into two survey instruments. The construction of these survey
instruments will be based on that constructed, validated, and used for a similar
purpose by Green [1997]. The two newly constructed instruments will be validated
in pilot tests conducted with experts within IDS, Rational Rose, and some
selected customers. In 2002, the ARIS-related instrument will be issued to 400
IDS customers. Similarly, in 2003, the UML-related instrument will be issued to
400 Rational Rose customers. Customers in both instances will either be selected
on a random basis from the full customer list or 100 percent sampling will be
used if the numbers are small.
Figure 2. Project phases and milestones of the first two years
Structured interview protocols will be developed from the respective validated instruments. These interview protocols will then be administered in tape-recorded interviews with a small number of customers of IDS and Rational Rose respectively not included in the surveys. The purpose of these qualitative interviews is two-fold:
Analyses. The results of the surveys will be summarised and analysed using standard statistical tests for significance available in such packages as SPSS. The tape-recorded structured interviews will be transcribed into standard Word documents. These text documents will then be coded and analysed using a standard qualitative text pattern matching analytical package like Nud-ist.
Michael Rosemann will contribute his knowledge in the areas of process
modelling, ARIS and UML. Thus, he will be in charge for the meta models for ARIS
and UML. He also will be responsible for the development of suggestions for
improvements for these two modelling grammars. Michael has for many years worked
with representatives from the IDS Scheer AG. These contacts will be very
valuable in the design and execution of the empirical study.
Peter Green is a well-recognised researcher in the area of ontological analyses.
He will contribute his comprehensive and in-depth knowledge of the selected
ontology, the BWW models. Peter will be responsible for the semantic correctness
of the meta model of the BWW models and the correct application of the BWW
constructs. As Peter is also an experienced empirical researcher, he will be in
charge of the empirical study.
Peter and Michael will both together conduct the actual ontological evaluation
of ARIS and UML as this task requires the consolidation of knowledge related to
ontology and process modelling.
A full-time research assistant will support the Chief Investigators during the
design of the meta models, the actual evaluation, and especially in conduct of
the empirical surveys. Further personnel will be employed on a part-time basis
for tasks like publishing the results adequately on the web or translation
activities related to the empirical survey.
In addition to this proposed team of researchers, an advisory board will be
established. It will include Ron Weber, University of Queensland, August-Wilhelm
Scheer, University of Saarbruecken, Germany, Yair Wand, University of British
Columbia, Canada, and Andreas Opdahl, University of Bergen, Norway. Prof. Ron
Weber together with Yair Wand developed the BWW models. His involvement in this
project is of high importance. He will especially be involved in the critical
review of the meta model for the BWW models. Prof. August-Wilhelm Scheer
developed over the last decade the Architecture of the Integrated Information
Systems (ARIS). He also founded a company, the IDS Scheer AG, that developed a
software tool for ARIS (ARIS-Toolset). The ARIS-Toolset currently boasts more
than 25,000 licences sold - the most widely distributed upper-CASE tool. Scheer
will be involved in the discussion of outcomes of the ontological analyses of
ARIS. Selected users of the ARIS-Toolset will participate in the empirical study
of these results. Prof. Yair Wand is the second developer of the BWW models. He
has comprehensive knowledge in applying the models to object-oriented methods.
Assoc. Prof. Andreas Opdahl has investigated object-oriented methods (OML) using
the BWW models. Thus, he brings knowledge in the areas of ontology and UML that
is very valuable for this project. Opdahl will contribute to this project in
regard to the analysis of UML. The three advisors will also get all progress
reports of this project.