Business Process and Business Rule Modeling: A ...Document Transcript
Business Process and Business Rule Modeling:
A Representational Analysis
Michael zur Muehlen Marta Indulska Gerrit Kamp
Stevens Institute of Technology The University of Queensland Stevens Institute of Technology
firstname.lastname@example.org email@example.com firstname.lastname@example.org
Abstract pirical research has identified representational weak-
nesses in process modeling languages and led to
Process modeling and rule modeling languages are speculation that business rule modeling languages
both used to document organizational policies and might fill this gap . It is an open question whether
procedures. However, little work has been done to the two language types should be used in combina-
understand their synergies and overlap. Understand- tion in order to increase the representation capability
ing the relationship between the two modeling types for process modeling. There is a need for a rigorous
would allow organizations to maximize synergies and analysis of the overlap of the two types of languages
reduce their modeling effort. In this paper we use the in order to identify their potential synergies.
well-established Bunge-Wand-Weber (BWW) repre- Therefore, the main goal of the work we present
sentation theory to compare the representation capa- in this paper is to investigate the representation capa-
bilities of both types of languages. We perform a rep- bility of two rule modeling languages, viz., the Sim-
resentational analysis of two rule modeling lan- ple Rule Markup Language (SRML) and Semantics
guages, viz., SRML and SBVR. We compare their of Business Vocabulary and Business Rules (SBVR),
representation capabilities with those of four popular and to place this evaluation in the context of previous
conceptual business process modeling languages, representation capability evaluations of conceptual
and focus on the aspects of maximum ontological process modeling languages. Our evaluation of repre-
completeness and minimum ontological overlap. The
sentation capability is based on the well-established
outcome of this study shows that no single language
Bunge-Wand-Weber (BWW) representation theory.
is internally complete with respect to the BWW repre-
sentation model and that a combination of two lan- Our two research questions are:
guages, viz. SRML and BPMN, is better suited for RQ1: What is the representational capability,
process modeling than any single modeling language. with respect to the BWW representation theory, of
SRML and SBVR?
1. Introduction RQ2: Are SRML and SBVR complementary or
substitutive to process modeling languages?
Business Process Management (BPM) has been The remainder of this paper is structured as fol-
identified as the number one priority of CIOs for the lows. In the next section we present a brief review of
last few years . Organizations are increasingly business rules and business processes. The section
interested in improving their core process portfolio, also provides a review of related work on the integra-
and in identifying and quantifying processes with tion of the two approaches and discusses recent em-
outsourcing potential. pirical studies that identified deficiencies in process
Both process modeling languages and rule mod- modeling languages. Section 3 presents the justifica-
eling languages offer constructs to represent work tion for the use of the BWW representation theory as
operations and constraints. This situation presents a a suitable benchmark for the analysis of representa-
selection dilemma for organizations where little guid- tion capabilities of process and rule modeling lan-
ance exists. While a significant amount of work has guages. The following section describes the research
been done in terms of evaluating the representational methodology adopted in this work and provides a
capability of process modeling languages , the justification for the selection of languages under con-
evaluation of rule modeling languages has received sideration. In section 5, we present a summary of the
considerably less attention. Furthermore, recent em- results of the BWW-based representation analysis of
SRML and SBVR and discuss the results of the declarative modeling category in that they focus on
analysis in light of representation capabilities of specifying what is required to take place, rather than
process modeling languages. We conclude the paper how something is accomplished.
in section 6 with a discussion of limitations and fu-
ture work directions in this area. 2.2 Business Processes
Business Processes are sets of activities that cre-
ate value for a customer. Organizations are increas-
While, to the best of our knowledge, no repre- ingly automating processes using workflow systems,
sentational evaluation of rule modeling languages has and are building elaborate management systems
been carried out, the background related to this work around their processes. Such management infrastruc-
exists in the documented attempts at the integration tures integrate modeling, automation, and business
of rule- and process-based modeling approaches, intelligence applications.
some surveys/comparisons of different approaches to A variety of modeling languages exists for the
business rule specification, as well as empirical stud- specification of process models; they can be classi-
ies investigating the weaknesses of process modeling fied according to their focal modeling construct:
languages. Activity-centered; processes as a network of tasks
2.1 Business Rules Process object centered; processes as the legal
sequence of state changes of the process object.
A business rule is a statement that aims to influ- Resource centered; process as a network of proc-
ence or guide behavior and information in an organi- essing stations that interact with each other.
zation . Business rules can be categorized in ac- Process languages appear as Graph-based lan-
cordance to their source or structure: guages (e.g. BPMN, EPC), Net-based languages (e.g.
Mandates; published policies that must be fol- Petri-nets, flow nets) and Workflow Programming
lowed, or consequences will ensue. Examples are Languages (e.g. BPEL). Such languages, in general,
the payment of taxes and adherence to the law. are considered to be of the procedural modeling type,
Policies; published standards that should be fol- in that they focus on specifying the step-by-step ac-
lowed to adhere to acceptable company behavior. tivities that are required to take place in order to per-
Examples are budgets or mission statements. form an action. While they do not provide the same
Guidelines; rules that may or may not apply, de- level of precision or formalism as rule modeling lan-
pending on circumstances. Examples are method- guages, their strengths stem from their relative user
ologies and management styles. friendliness and structural properties .
The different structural categories of business
rules are : 2.3 Integration of Business Rules and Busi-
Integrity (or constraints); For example: each pro- ness Processes
ject must have one and only one project manager.
Derivation (conditions resulting in conclusions); Early work on the integration of business rules
For example: platinum customers receive a 5% and business processes started appearing shortly after
discount. John Doe is a platinum customer. As a the introduction of the rule modeling concept [7, 8].
conclusion, John Doe receives a 5% discount. Krogstie et al.  were the first to suggest that busi-
Reaction (Event, Condition, Action, Alternative- ness process and rule modeling approaches should be
action, Post-condition); For example: an invoice is merged to improve the capturing of temporal infor-
received. If the invoice amount is more than mation for information systems development. They
$1,000 then a supervisor must approve it. presented a top-down approach for model specifica-
Production (condition, action); For example: if
tion that involves the use of the External Rule Lan-
there are no defects in the last batch of cars then
guage (ERL) for specification of process logic at the
the batch is approved.
lowest level of decomposition. This concept was fur-
Transformation (change of state); For example, an
employee’s age can change from 30 to 31, but not ther enhanced by McBrien and Seltveit , who pre-
from 31 to 30. sented a way to define the structure of rules within
Business rule modeling languages are typically the process model. Knolmayer et al.  introduced a
based on formal logic and have strong and precise framework where process modeling is refined and
expressive power . In general, they belong to the linked to workflow execution through some layers of
Reaction Business Rules. Kappel et al.  use Reac- lack the ability to adequately represent business rules.
tion Business Rules to model the coordination in The work of Rosemann et al.  suggests that the
workflow systems. Kovacic  developed a meta- same approach of incorporating business rule model-
model that represents important business constructs ing languages might apply in the process modeling
(goal, process, activity and events) and technical con- domain.
structs (data objects, software components, actions in In order to effectively integrate graphical busi-
Information Systems). He demonstrates how rules ness process modeling approaches with business rule
can link these two categories of constructs. Charfi modeling approaches, one must understand their
 argues that business rules are often hard-coded synergies and overlap. At the time of writing, the
into web services and proposes a hybrid approach of authors are not aware of any attempts to evaluate the
separating business processes and business rules. expressiveness of rule modeling languages, nor their
Meng et al.  introduced a dynamic workflow relationships to conceptual process modeling
management system for modeling and controlling the approaches. The only related work appears to be that
execution of inter-organizational business processes. of Lu and Sadiq , who carried out a comparison
The system uses an event and rule server to trigger of graph-based and rule-based modeling approaches.
business rules during the enactment of workflow No specific rule modeling languages were considered
processes in order to enforce business constraints and as the work was focused on workflow modeling in
policies at run-time. particular rather than conceptual modeling in general.
While the integration of the two approaches has The authors used a set of workflow patterns as a basis
been the subject of some early investigation in the for the evaluation, and found, among other results,
research community, anecdotal evidence shows that that rule- and graph-based modeling approaches had
organizations struggle with effectively capturing similar levels of expressiveness in terms of the con-
business processes and rules. In a recent empirical trol flows specified by the workflow patterns.
study of the representational capabilities of Business Accordingly, there is a need to provide research
Process Modeling Notation BPMN  we found that and practice with theory-backed guidance as to which
organizations supplement their BPMN process mod- rule modeling language provides the best representa-
els with textual annotations of business rules. This tional power.
practice introduces problems with rule consistency,
reuse, and enforcement – problems that are acknowl- 3. Representation Theory
edged by some of the organizations making use of the
textual business rule annotations. An ontology, or representation theory, can be
While the need to improve the representation of used as a benchmark to make predictions about the
business rules within process model diagrams is ap- capabilities of a grammar to provide complete and
parent, little is known about which representation clear representations of a real-world domain .
aspects, if any, are unique to each of the two types of The application of an ontology for such a purpose is
modeling languages. Previous work by Recker et al. known as representational analysis.
 has identified a general lack of process modeling Representational analysis is performed by com-
language capabilities to adequately model business paring the constructs of the chosen representation
rules. Similarly, Green and Rosemann  found theory with the constructs of the modeling grammar
limitations with respect to modeling business rules in and by identifying any representation equivalence
their BWW-based investigation of all five views of between these. Any deviation from a one-to-one
Architecture of Integrated Information Systems mapping relationship between these constructs indi-
(ARIS), a popular framework for integrated process cates potential representational deficiencies in the
modeling. grammar. Two principal evaluation criteria are onto-
Rule modeling languages are likely candidates to logical completeness, i.e., the extent to which the
fill such gaps. Indeed, an earlier study by Herbst et modeling grammar has a deficit of constructs that
al.  suggests that rule specification languages map to the set of representation theory constructs,
should be considered as a potential addition to and ontological clarity, i.e., the extent to which the
graphical representation languages when modeling modeling grammar constructs are deemed over-
for Information Systems (IS) design. While their loaded, redundant, or excessive . These criteria
analysis is not based on any formal framework, they provide a theoretical basis on which conceptual mod-
find that many of the popular IS modeling techniques eling languages can be compared with regard to their
completeness of representation and clarity. more than one grammatical construct maps to one
In this study we use the Bunge-Wand-Weber BWW construct. Higher levels of construct overlap
(BWW) ontology , specifically the representation will create confusion and conflict in the work of the
model, since it is understood to contain all necessary users. Together, the application of the MOC and
constructs to describe things, and the interaction be- MOO theories is known as overlap analysis and is
tween things, in the real world. While the BWW done with the intended purpose of identifying lan-
model is not without its criticisms (for example ) guage combinations with highest expressive power
and other ontologies could also be applied (for exam- but low construct overlap.
ple Chisolm’s ontology  or the Enterprise Ontol- We make use of the BWW representation model,
ogy ), several studies have shown that BWW is a together with the conduct of overlap analysis, in or-
good basis to study the representational capabilities der to analyze the representational capabilities of
of conceptual modeling languages [13, 14]. This suit- SRML and SBVR, and identify combinations of
ability has been empirically demonstrated in a large business rule and business process modeling lan-
number of cases, as summarized in . The choice of guages that are likely to be used in combination.
BWW representation theory allows us to compare our
results to previous BWW-based analyses of process 4. Methodology
The BWW representation model consists of 4.1 Selection and Analysis of Rule Modeling
some 40 higher-level abstract constructs, which can Languages
be grouped into four categories: Things and their
Properties, States of a Thing, Events and Transforma- A variety of rule languages have been developed
tions, and Systems and their Composition. If a proc- over the past decade, as shown in Figure 1. For our
ess or rule modeling language construct is found to initial study we selected two rule modeling language
have a representation for each of the BWW represen- specifications that provided a comprehensive expla-
tation model constructs then that language fulfills all nation of their vocabularies. The Simple Rule
the representation requirements criteria necessary to Markup Language (SRML) was selected as a repre-
model things and their interactions in the real world sentative example of a rule modeling language with a
(with respect to the BWW representation model), small vocabulary. A clear definition of its constructs
without limiting the user’s representation capabilities. is available and is not based on any other vocabulary
While this may be the case, the language may still . The Semantics of Business Vocabulary and
suffer from lack of clarity (e.g. an overload of con- Business Rules (SBVR) was selected since it repre-
structs), which impacts its usability. A language that sents an attempt at the definition of a standardized
is complete and has the lowest levels of construct rule modeling vocabulary . SBVR presents a vo-
overload, redundancy and excess should be chosen. cabulary that is intended to become a standard upon
When no one language provides the required repre- which many grammars can be based. For this reason
sentation capability in terms of completeness, Green its inclusion in the ontological analysis is useful.
et al.  show that users will make use of combina- In order to reduce subjectivity and increase in-
tions of languages that allow them to obtain maxi- ternal validity of our research, we employed the ex-
mum representation capability. Green et al.  dis- tended representational analysis methodology as sug-
cuss two theories for selecting two or more grammars gested by Rosemann et al. . We followed the
for Information Systems modeling. The first, Maxi- reference methodology as closely as possible. In par-
mum Ontological Completeness (MOC), states that ticular, to increase objective comparison, the ERD
users will select combinations of languages that, to- BWW meta-model was obtained from the authors of
gether, afford them the maximum possible represen-  and an ERD meta-model of the SRML language
tation power for their domain, i.e., if more constructs was created to guide the mapping between SRML.
from an underlying ontology are incorporated in the SBVR was not transformed into an ERD diagram
chosen grammar, the expressive power of the result- because the SBVR specification contains many UML
ing language combination will be higher. The second, diagrams of the language constructs, which were suf-
Minimum Ontological Overlap (MOO), states that, ficient for a thorough understanding of the language.
when selecting languages to satisfy MOC, users will The three researchers independently conducted
prefer languages with minimum overlap in ontologi- the representation analysis of the languages. The in-
cal constructs, i.e., language combinations where no dependent analyses were followed by coordination
As a second step, we apply the process of over-
lap analysis in order to determine a pair of languages
that provides, with respect to the BWW representa-
tion model, the highest representation modeling
power while having the lowest amount of construct
overlap between the languages. Sine the MOC theory
takes precedence in such analysis, we focus on all
language pairs that provide MOC and select the
pair(s) that have the lowest construct overlap.
Fig. 1. Development of Rule Languages
5. Discussion of Analysis Results
The summary of the BWW representational
sessions, during which consensus was gained about
analysis of SRML and SBVR is shown in Table 1. A
the construct mappings. The analyses and respective
tick indicates that the rule modeling language was
consensus development meetings were conducted in
found to have capability to represent the correspond-
turn for each language. First, SRML was analyzed
ing BWW representation model construct. The full
and the analysis was finalized, then SBVR analysis
details of the mapping reasoning are omitted due to
paper length limitations.
The analysis shows that the two chosen rule
4.2 Selection of process modeling languages
modeling languages are less expressive than their
and overlap analysis
process modeling counterparts, with respect to the
The selection and analysis of conceptual process BWW representation model. This is perhaps not sur-
modeling languages was based on existing work. prising given that their focus is narrower that that of
Many BWW-based representational analyses of proc- graphical process modeling languages. For example,
ess modeling languages have already been published, while we would expect business rules modeling lan-
and some of them empirically tested (for an overview guages to provide a corresponding representation for
please refer to ). We specifically focus on con- a BWW construct such as state law, the lack of
ceptual modeling languages rather than executable graphical representation might lessen the need for a
languages such as Business Process Execution Lan- representation of system decomposition or system
guage (BPEL), since we concentrate on the documen- structure. However, this observation implies that
tation of policies and processes rather than their business rule modeling languages like SRML or
execution. Accordingly, we chose Petri Nets (in their SBVR in isolation do not provide an equivalent or
original specification), Event-driven Process Chains better means for modeling processes than some of the
(EPC), Integrated DEFinition methodology – Process established and popular process modeling languages
Description Capture Method (IDEF3), and Business (most notably, BPMN).
Process Modeling Notation (BPMN) as the basis of Closer investigation of Table 1 shows that, even
our comparison. We want to determine if business when used in combination, none of the popular proc-
rule languages can indeed contribute to these popular ess and rule modeling languages considered provides
graphical process modeling languages, as was specu- complete coverage for all BWW constructs. In par-
lated by Herbst et al.  for the IS domain. ticular, representations for the BWW representation
The analysis of how complementary business theory constructs of history, conceivable event space,
rule and process modeling languages are was per- and lawful event space are missing across all lan-
formed based on the results shown in Table 1. As a guages under consideration. This implies that even
first step, we compare the representational capabili- combinations of conceptual process modeling lan-
ties of the six languages and derive a set of BWW guages and business rule modeling languages are
representation model constructs that do not have a neither able to represent audit trails of activities nor
corresponding construct across the chosen languages. the sets of all the possible or allowed events that can
This situation implies that full completeness of repre- occur in a given situation. More specifically, while
sentation cannot be achieved with the selected lan- the addition of business rule modeling languages ap-
guages with respect to the BWW representation pears to alleviate some of the empirically validated
weaknesses of the popular BPMN language in par- plies an inability to model the set of all possible
ticular , we still expect the following shortcomings events that can occur within a process. While this
to manifest in practice: representation capability may not always be required
Table 1. BWW Analysis results, including mate- in the process modeling domain, the lack of it in-
rial from  and  creases the complexity of identifying events of inter-
Language est to the process being modeled and events that are
2004 v1.0 BPMN
not allowed to have a triggering impact.
P3: The lack of corresponding representation for
Year / Version the BWW construct of lawful event space, similarly
to proposition P2, has a negative impact on the mod-
eling of allowable events in a process model. Specifi-
BWW Constructs cally, there is no modeling construct that would allow
Thing for the representation of all events that are legal in a
Property given process context, thus impacting exception han-
The overlap analysis of the six chosen languages
is summarized in Table 2. Each cell in the table is a
Conceivable State Space
1. Unique: The number of BWW constructs repre-
State Law sented distinctly by the given combination of
Lawful State Space languages, free of overlap;
Stable State 2. Overlap: The number of BWW concepts that can
Unstable State additionally be represented by both languages,
History with construct overlap;
Event 3. Process Add: The number of non-overlapping
Conceivable Event Space BWW constructs contributed by the process
Lawful Event Space modeling language shown in the table column to
External Event the rule modeling language shown in the table
Internal Event row; and
Well-defined Event 4. Rule Add: The number of non-overlapping
Poorly-defined Event BWW constructs contributed by the rule model-
Transformation ing language shown in the row to the process
Lawful Transformation modeling language shown in the column.
Coupling While the overlap analysis shows synergies be-
Acts on tween rule modeling languages and process modeling
System languages, these do not appear to be as dramatic as
System Composition we expected. It is clear that business process lan-
guages are a better choice for modeling organiza-
tional procedures, and that BPMN is a construct-rich
process modeling language that could be enriched by
the addition of SRML for this domain.
Investigation of the overlap results summarized
in Table 2 indicates that incorporating the use of
# BWW constructs repres. 10 7 12 11 11 19
SRML with any of the four popular conceptual proc-
P1: The lack of corresponding representation for ess modeling languages allows users the ability to
the BWW construct of history may have an impact represent between fifteen and twenty-three represen-
where a log of an entity’s state changes is required. tation theory constructs (i.e. MOC = 23). Minimal
The lack of such explicit representation can, for ex- ontological overlap is equal to 6, implying that
ample, impact exception modeling, in particular re- whichever combination of languages is chosen, a
covery. minimum of six constructs will be overlapping in the
P2: The lack of corresponding representation for language pair. Considering both MOC and MOO
the BWW construct of conceivable event space im- theories, the analysis clearly shows that the combina-
Table 2. Overlap analysis of the selected languages BPMN and the associated costs of converting their
Petri Nets EPC IDEF3 BPMN models. Last, once a pair of languages is in use, or-
8 5 9 5 9 5 17 13 ganizations may use the results of the analysis (Table
7 3 6 4 6 4 6 4 1) as guidance for the development of consistent
13 9 14 9 8 6 14 13 workaround policies to alleviate the weaknesses their
3 4 2 5 5 2 6 1 modelers will encounter while using the language
LEGEND FOR EACH QUADRANT pair to model organizational procedures.
1. Unique 3. Process Add
2. Overlap 4. Rule Add
tion of BPMN and SRML provides users with the
highest representation power, while having minimal This paper presents the first theory-based analy-
ontological overlap. On the other hand, the analysis sis of representational capabilities of two rule model-
also shows that the combination of SRML and Petri ing languages, viz. SRML and SBVR. The considera-
Nets is not a good option, given the higher level of tion of our analyses, together with existing represen-
overlap and representation of only fifteen representa- tational analyses of four popular conceptual process
tion theory constructs, as compared to twenty-three modeling languages, has allowed us to provide some
from the BPMN/SRML combination. initial direction, which combinations of languages
A closer look at SBVR reveals that SRML is a provide users with the best representational capabili-
superior option for the integration with conceptual ties. Our findings show that the combination of
process modeling languages, since it provides BPMN with SRML provides users with the highest
representation of three additional representation representation power while suffering an amount of
theory constructs at the same level of construct construct overlap that is no higher than that of other
overlap. The representational capability of SBVR in language pairs. However, the analysis also shows
combination with any of the process modeling that even this combination of languages is still defi-
languages ranges from thirteen to twenty constructs, cient in some constructs, viz. history, conceivable
while overlap ranges from two to six. MOO indicates event space, and lawful event space.
in this case that a combination of SBVR and EPC is While our initial findings encourage further in-
good from a clarity perspective, although such a vestigation of the integration of process and rule
combination offers just sixteen representation theory modeling languages, there are some known limita-
constructs. Table 1 also shows that such a combina- tions to our current approach. First, different authors
tion would be lacking the representation of the con- performed the analysis of EPC and IDEF3, and their
ceivable and lawful state space, constructs that, intui- interpretation of the language constructs may differ
tively, are important for modeling organizational from ours. Second, the analysis of Petri Nets is based
policies and rules. Accordingly, when a higher on an old version of Petri Nets and can be extended
representation capability is required, the SBVR and to incorporate newer versions (e.g. colored Petri
BPMN combination could be chosen, if SRML is not Nets), which may alter the results. Third, the pub-
an option. initial findings are significant for three
These lished representational analyses of modeling gram-
reasons. First, they provide guidance to the develop- mars generally do not include analysis of representa-
ers of modeling languages in terms of which areas tional capability of combinations of the modeling
require improvement. Clearly, there are a number of grammar constructs, focusing instead on representa-
constructs missing across the board and further inves- tion of each construct in isolation. Fourth, we assume
tigation is necessary in terms of their criticality and that each BWW representation model construct is
potential addition in future revisions of these model- equally important for the process modeling domain.
ing languages. Second, our results can provide guid- In the future, we will conduct an expert study to in-
ance to organizations in adoption of a specific set of vestigate a more refined ranking of ontological con-
modeling languages for their process documentation structs in order to determine the criticality of missing
efforts. For example, an organization already using representations. Finally, the results of the mapping
BPMN for process modeling has a theoretical basis need to be tested against a real-world example.
for the choice of SRML over SBVR. An organization Work is currently underway to include additional
already using Petri Nets would be inclined to adopt rule languages in the evaluation. In particular, SWRL
SBVR to obtain higher representational power with is being considered to complement the analysis of
lower construct overlap, or investigate a switch to SBVR, and an evaluation of RuleML is in progress.
Other rule modeling languages are also being evalu-  A. Kovacic, "Business renovation: business rules
ated for inclusion since more suitable combinations (still) the missing link," Business Process Manage-
of languages may be identified if more rule languages ment Journal, vol. 10, p. 158, 2004.
 A. Charfi and M. Mezini, "Hybrid web service com-
are included. Further work is also required to develop position: business processes meet business rules," in
a cost/benefit calculation that would indicate whether 2nd international conference on Service oriented
the additional representational capability provided by, computing, New York, NY, USA, 2004.
for example, SRML over and above BPMN, is worth  J. Meng, S. Y. W. Su, H. Lam, and A. Helal, "Achiev-
the complexity of adding an additional language. ing dynamic inter-organizational workflow manage-
ment by integrating business processes, events and
Last, we see the need for further research that focuses
rules," in 35th Hawaii International Conference on
on how to achieve a meaningful and seamless inte- System Sciences, Hawaii, 2002.
gration of business process and business rule model-  J. Recker, M. Indulska, M. Rosemann, and P. Green,
ing languages. "Do Process Modelling Techniques Get Better? A
Comparative Ontological Analysis of BPMN," in 16th
7. References Australasian Conference on Information Systems,
Sydney, Australia, 2005.
 Gartner Group, Creating Enterprise Leverage: The  P. Green and M. Rosemann, "Perceived Ontological
2007 CIO Agenda. Egham, UK: Gartner, Inc, 2007. Weaknesses of Process Modelling Techniques: Further
 M. Rosemann, J. Recker, M. Indulska, and P. Green, Evidence," in 10th European Conference on Informa-
"A Study of the Evolution of the Representational Ca- tion Systems, Gdansk, 2002, pp. 312-321.
pabilities of Process Modeling Grammars," in Ad-  H. Herbst, G. Knolmayer, T. Myrach, and M. Schle-
vanced Information Systems Engineering - CAiSE singer, "The Specificatioin of Business Rules: A
2006. vol. 4001, E. Dubois and K. Pohl, Eds. Luxem- Comparison of Selected Methodologies," in Methods
bourg, Grand-Duchy of Luxembourg: Springer, 2006, and Associated Tools for the Information System Life
pp. 447-461. Cycle, A. A. Verijn-Stuart and T.-W. Olle, Eds. Am-
 J. Recker, M. Indulska, M. Rosemann, and P. Green, sterdam: Elsevier, 1994, pp. 29-46.
"How Good is BPMN Really? Insights from Theory  R. Lu and S. Sadiq, "A Survey of Comparative Busi-
and Practice," in 14th European Conference on Infor- ness Process Modeling Approaches," in 10th Interna-
mation Systems, Goeteborg, Sweden, 2006. tional Conference on Business Information Systems
 G. Steinke and C. Nikolette, "Business rules as the Poznan, Poland: Lecture Notes in Computer Science,
basis of an organization's information systems," Indus- Springer-Verlag, 2007.
trial Management + Data Systems, vol. 103, p. 52,  R. Weber, "Ontological Foundations of Information
2003. Systems," Coopers & Lybrand, 1997.
 G. Wagner, "Rule Modeling and Markup," in Reason-  M. Rosemann, P. F. Green, and M. Indulska, "A Ref-
ing Web, 3564 ed, N. Eisinger and J. Maluszynski, erence Methodology for Conducting Ontological
Eds. Msida, Malta: Springer, 2005, pp. 251-274. Analyses," in 23rd International Conference on Con-
 P. McBrien and A. H. Seltveit, "Coupling Process ceptual Modeling, Shanghai, 2004, pp. 110-121.
Models and Business Rules," in Information Systems  B. Wyssusek, "Ontology and ontologies in information
Development for Decentralized Organizations, Pro- systems analysis and design: a critique," in Americas
ceedings of the IFIP 8.1 WG Conference: Chapman Conference on Information Systems New York, 2004.
Hall, 1995.  R. M. Chisholm, "The basic ontological categories," in
 G. Kappel, S. Rausch-Schott, and W. Retschitzegger, Language, Truth, and Ontology. vol. 51, K. Mulligan,
"Coordination in workflow management systems - A Ed. Dordrecht: Kluwer Academic Publishers, 1992,
rule-based approach," in Coordination Technology for pp. 1-13.
Collaborative Applications - Organizations, Proc-  M. Uschold, M. King, S. Moralee, and Y. Zorgios,
esses, and Agents, 1364 ed, W. Conen and G. Neu- "The Enterprise Ontology," Artificial Intelligence Ap-
mann, Eds.: Springer, 1998, pp. 99-119. plications Institute, The University of Edinburgh 1997.
 G. Knolmayer, R. Endl, and M. Pfahrer, "Modeling  P. Green, M. Rosemann, M. Indulska, and C. Man-
Processes and Workflows by Business Rules," in ning, "Candidate Interoperability Standards: An Onto-
Business Process Management, Models, Techniques, logical Overlap Analysis," Data & Knowledge Engi-
and Empirical Studies, W. M. P. van der Aalst, J. De- neering, vol. 62 pp. 274-291, 2007.
sel, and A. Oberweis, Eds. London: Springer, 2000,  Coverpages.com, "Simple Rule Markup Language
pp. 16-29. (SRML)," R. Cover, Ed., 2001.
 J. Krogstie, P. McBrien, R. Owens, and A. H. Seltveit,  OMG, "Semantics of Business Vocabulary and Busi-
"Information Systems Development Using a Combina- ness Rules Specification, Interim Specification," 2006.
tion of Process and Rule-Based Approaches," in Third  P. F. Green, M. Rosemann, and M. Indulska, "Onto-
Nordic Conference on Advanced Information Systems logical evaluation of enterprise systems interoperabil-
Engineering: LNCS, Springer-Verlag, 1991. ity using ebXML," Knowledge and Data Engineering,
IEEE Transactions on, vol. 17, pp. 713-725, 2005.