Consistency, Complementalness, or Conflictation of Enterprise Ontology
and Normalized Systems Business Process Guidelines
Philip Huysmans, Dieter Van Nuffel and Peter De Bruyn
Normalized Systems Institute, University of Antwerp, Prinsstraat 13, B-2000 Antwerp, Belgium
{philip.huysmans,dieter.vannuffel,peter.debruyn}@ua.ac.be
Keywords:
Normalized Systems, Enterprise Ontology, Business Process Modeling, Enterprise Engineering.
Abstract:
Both Enterprise Ontology and Normalized Systems can be considered as design theories, which provide pre-
scriptive guidelines to design systems. Enterprise Ontology explicitly focuses on the design of organizations as
being social systems. Originally, Normalized Systems focused on the design of evolvable software systems.
However, it has been shown that, building on the Normalized Systems design knowledge, prescriptions for
other domains, such as the business process domain, can be proposed as well. This domain seems to overlap
at least partially with the domain of Enterprise Ontology, which is used to establish claims concerning pro-
cess design in various publications. However, both theories are based on completely different kernel theories.
Therefore, this paper analyzes to which extent the guidelines proposed for the Normalized Systems Business
Processes are consistent, complementing or conflicting with prescriptions from Enterprise Ontology. A con-
sistent set of prescriptions could lead to a more integrated approach for designing integrated organizations,
business processes and software systems.
1 INTRODUCTION
The design of organizations and their components
such as the organizational structure, business pro-
cesses, and software systems is an important topic in
both practical and scientific communities (Galbraith,
1974; Avenier, 2010). Notwithstanding the attention
for this subject, explicit design knowledge in these
fields seems limited. For example, Mendling et al. ar-
gue that many theoretical frameworks are too abstract,
and that more practically-oriented guidelines lack em-
pirical and theoretical support (Mendling et al., 2010).
As a result, design of organizational components is
often considered as craftsmanship, rather than engi-
neering.
The enterprise engineering paradigm introduces a
set of prescriptive design theories which seek to rem-
edy this issue (Dietz et al., 2013). It specifically men-
tions the β and ν theories as well-founded theories
to guide design efforts. The ν-theory states that the
design of a system is normalized when a change con-
sists of a set of elementary changes, so that every el-
ementary change does not trigger combinatorial ef-
fects (Dietz et al., 2013, p. 101). Normalized Sys-
tems (NS) provides concrete guidelines and design
patterns to obtain such normalization in software sys-
tems (Mannaert and Verelst, 2009). Based on this ap-
proach, normalization of business processes has been
researched as well (Van Nuffel, 2011). This research
resulted in a set of guidelines which need to be ad-
hered to during business process design. While both
Normalized Systems and Normalized Systems Busi-
ness Processes (NSBP) originally aimed at obtaining
designs exhibiting stability as defined in systems the-
ory (Kelly, 2006), it has been argued that the result-
ing guidelines are in line with existing heuristics of
experienced designers. For example, the guidelines
presented by Van Nuffel are related to the existing
business process literature (Van Nuffel, 2011). Nev-
ertheless, the main contribution of both approaches
is the formulation of unambiguous and theoretically
founded guidelines based on the single postulate of
obtaining the systems theoretic concept of stability.
As a result, an approach which resembles traditional
engineering, rather than mere craftsmanship, arises on
these levels.
The β-theory states that enterprise architecture
should be defined as deliberate restriction of design
freedom, which should address the function design,
construction design, and implementation design of
systems (Dietz et al., 2013, p. 100). For example, En-
terprise Ontology (EO) prescribes how the construc-
tion design of an organization should be made (Di-
etz, 2006). EO prescribes a clear way of separating
54
Huysmans P., Van Nuffel D. and De Bruyn P.
Consistency, Complementalness, or Conflictation of Enterprise Ontology and Normalized Systems Business Process Guidelines.
DOI: 10.5220/0004774000540063
In Proceedings of the Third International Symposium on Business Modeling and Software Design (BMSD 2013), pages 54-63
ISBN: 978-989-8565-56-3
Copyright
c
2013 by SCITEPRESS – Science and Technology Publications, Lda. All rights reserved
different abstraction levels to be considered in organi-
zations (i.e., ontological, datalogical and infological)
and a systematic recurring pattern to model the onto-
logical level.
While the formulation of such theories has been
demonstrated to further the field in practice
1
, several
issues remain. One important issue is the current lack
of integration between the specific methods which in-
tegrate the different theories (Dietz et al., 2013). This
issue has been documented in many studies related to
enterprise architecture (Kaisler et al., 2005; Dreyfus,
2007). While certain frameworks, such as TOGAF,
focus explicitly on a method to integrate high-level
activities such as strategy formulation with detailed
software design, no prescriptive methods are present
in such frameworks. As a result, the integration of dif-
ferent prescriptive methods can remain very complex.
Enterprise architecture researchers have shown that
most reports focus on a single architectural layer, and
do not address this integration (Schenherr, 2008). In
practical projects, this results in local optimizations,
which restrict the success of an organizational design
as a whole (Kaisler et al., 2005; Dreyfus, 2007).
This means that, within the enterprise engineer-
ing community, additional research is required which
works towards an integrated method consisting of
different prescriptive design theories. For example,
both EO and NSBP seem to provide a similar kind
of guidelines when used in practical projects, which
could indicate that both approaches could be used as
complements in various projects. However, a clear
obstacle when aiming to apply both approaches si-
multaneously, is their difference in theoretical back-
grounds and abstraction. Therefore, an in-depth anal-
ysis regarding the possible compatibility of the guide-
lines resulting from both approaches is required up-
front. Such approach would investigate the extent
to which these guidelines are (1) similar in both ap-
proaches (i.e., consistent), (2) providing additional
guidelines (i.e., complementary), or contradicting one
another (i.e., conflicting). It should be noted that this
approach does not result in a theoretical analysis of
EO and NS(BP). On the one hand, NSBP cannot be
theoretically EO-compliant, since the distinction ax-
iom is not adhered to: no separation of ontological,
infological and datalogical concerns is made. On the
other hand, EO has not been developed based on the
concept of systems theoretic stability. Notwithstand-
ing this reservation, we are convinced that an analy-
sis of the consistency, complementalness, or conflic-
tation of the practical guidelines of both approaches
can contribute to an integrated use of both EO and
NSBP in various projects.
1
See for example
www.demo.nl
for case studies
2 BACKGROUND
2.1 Normalized Systems
Normalized Systems theory is aimed at studying how
modular structures behave under change (Mannaert
and Verelst, 2009). Initially, the theory was devel-
oped by studying change and evolvability at the soft-
ware architecture level, by applying concepts such as
stability and entropy to the study of the modular struc-
ture of the software architecture. Considering the ap-
plication of systems theoretic stability to software ar-
chitecture, stability implies that a bounded input func-
tion should result in bounded output values, even as
T . In other words, stability demands that the im-
pact of a change is only dependent on the nature of the
change itself. If the amount of impacts is related to the
size of the system, a combinatorial effect occurs. Re-
search has shown that it is very difficult to prevent CE
when designing software architectures. More specifi-
cally, it has been proven that CE are introduced each
time one of four theorems is violated (i.e., separation
of concerns, data version transparency, action version
transparency and separation of states).
Various studies have shown that combinatorial ef-
fects do not occur solely on the level of software ar-
chitectures (Van Nuffel, 2011; Huysmans, 2011). On
the business process level, it has been argued that
business processes at their most basic level (i.e., the
“elementary tasks and elementary sequencing and de-
sign of these tasks” (Van Nuffel, 2011)) can be con-
sidered to be modular structures as well. In this
context, business processes have been compared to
production lines (Van Nuffel et al., 2009a). In this
analogy, a business process flow performs operations
on instances of a specific life cycle information ob-
ject. Although production lines may seem highly in-
tegrated, they are actually loosely coupling. Every
single processing step requires the completion of the
previous steps on that instance of a particular product,
but it does not require any knowledge of the previous
processing steps, nor of the subsequent steps. As a
result, changes to individual processes or tasks do not
impact other processes of tasks (Van Nuffel, 2011).
Put differently, no combinatorial effects occur. More
generally, a business process which does not contain
combinatorial effects is called a Normalized Systems
Business Process (NSBP). In order to achieve such
processes, a set of guidelines has been developed,
which are based on the more fundamental theorems
of Normalized Systems. Together, these guidelines
allow the design of business processes without intro-
ducing combinatorial effects.
Consistency, Complementalness, or Conflictation of Enterprise Ontology and Normalized Systems Business Process
Guidelines
55
2.2 Enterprise Ontology
Enterprise Ontology (EO) provides an organizational
theory (Dietz, 2006) which is based on the Language-
Action Perspective (LAP). Consequently, it consid-
ers an organization as a social system, and focuses
on actor roles as the essential components of orga-
nizations. This is important for the goal of this pa-
per, since this background results in the claim that
EO provides “a modular framework for business pro-
cesses” (Dietz, 2003b, p. 1). The EO theory consists
of four axioms (i.e., the operation axiom, the transac-
tion axiom, the composition axiom and the distinc-
tion axiom) (Dietz, 2006). These axioms allow to
specify in more detail what is meant with the “modu-
lar construction of business processes” (Dietz, 2003b,
p. 18). Business processes are considered to con-
sist of three levels of building blocks. A first type
of building block (the atoms) refers to the individual
acts performed by actors, as explained by the opera-
tion axiom. These atoms can be combined in higher-
level building blocks (i.e., molecules), which repre-
sent the transactions as explained in the transaction
axiom. Multiple transactions can be required to fulfill
a certain service to a stakeholder. The collection of
these transactions (i.e., a fiber) is then considered to
be a business process.
Rather than merely defining business processes
using EO concepts, various studies have focused on
the design of business processes. For example, the
main research question of the paper Basic Notions
Regarding Business Processes and Supporting Infor-
mation Systems is “how business processes can be
understood in such a way that their continuous and
concurrent (re)designing and (re)engineering can be
performed more effectively than what is currently
the case” (Dietz and Albani, 2005). Another exam-
ple is the paper Enhancing the Formal Foundations
of BPMN by Enterprise Ontology, which states 11
propositions which can be derived from EO axioms
(Van Nuffel et al., 2009b). Based on the axioms, addi-
tional prescriptions for designing business processes
are available. For example, the operational cycle (Di-
etz, 2006, p. 163) states that an actor role needs to
be added when a transaction cannot be performed in
the same cycle of other transactions. Put differently,
this implies that the executor actor of an enclosing
transaction needs to be the initiator actor of an en-
closed transaction (cf. the composition axiom). Con-
sequently, EO prescribes that certain end-to-end pro-
cesses which are often defined in practice (e.g., order-
to-cash processes) need to be separated.
Various claims have been made that EO can in-
deed lead to better results when (re)designing pro-
cesses. The abstractions discussed in the distinction
axiom are claimed to be “a tremendous advantage
for discussing business process optimization” (Di-
etz, 2006, p. 183), (van Reijswoud, 1999). More-
over, the dedicated model within the DEMO method-
ology to represent business processes (i.e., the pro-
cess model) has been claimed to “facilitate the discus-
sion about the redesign of business processes” (Dietz,
2006, p. 183).
2.3 Is it possible to Compare Both
Theories?
Caution should be applied when comparing the Enter-
prise Ontology and Normalized Systems theory, since
their intentional application domains vary greatly.
Normalized Systems theory focuses on evolvability
of software architectures, while Enterprise Ontology
attempts to describe coordination in organizations.
Nevertheless, the Design Science paradigm argues
that the application of theories of related fields is
useful to make scientific progress. Moreover, Win-
ter and Albani claim that different design theories
can be combined in certain projects (Winter and Al-
bani, 2013). Both the Normalized Systems and Enter-
prise Ontology theory have already been positioned
in a Design Science research framework (Huysmans
et al., 2012; Winter and Albani, 2013). Comparing
these frameworks indicates an important difference
between both theories: Enterprise Ontology builds on
communication theories (i.e., the theory of commu-
nicative action, the language-action theory and sys-
temic ontology) while Normalized Systems builds on
system theoretic and thermodynamicconcepts such as
stability and entropy.
Notwithstanding this clear difference in kernel
theories, remarkable similarities between Normal-
ized Systems and Enterprise Ontology have been dis-
cussed as well (Huysmans, 2011). For example, con-
sider the Separation of States theorem. It states that
“the calling of an action entity by another action en-
tity needs to exhibit state keeping in normalized sys-
tems” (Mannaert and Verelst, 2009). It therefore pre-
scribes how action elements can interact. This im-
pacts, for example, the workflow element, which ag-
gregates action elements. A workflow can reach dif-
ferent states by performing state transitions. A state
transition is realized by an action element. The suc-
cessful completion of that action element results in a
defined life cycle state. The workflow specification
determines which state transitions can be made. Sim-
ilarly, the state of a transaction in enterprise ontology
is determined by the successful performance of acts.
The result of such an act results in the creation of a de-
Third International Symposium on Business Modeling and Software Design
56
fined fact. Despite the different terminology, a clear
resemblance between Normalized Systems and Enter-
prise Ontology emerges: state-keeping is enforced in
Normalized Systems by defining states, and in En-
terprise Ontology by creating facts. These Normal-
ized Systems states are the result of executing actions,
whereas the Enterprise Ontology facts are the result
of executing acts. Which actions can be performed is
determined by the state transitions in Normalized Sys-
tems, and occurrence laws in Enterprise Ontology.
Moreover, other attempts have been made to in-
tegrate Normalized Systems and Enterprise Ontology
theory more directly (Huysmans et al., 2010; Krouwel
and Op’t Land, 2011; Op ’t Land et al., 2011). It
should be noted that in these efforts, an inductive ap-
proach based on concrete artifacts is used, which can
be contrasted to a more theoretical approach. Simi-
larly, this paper does not attempts to provide a theoret-
ical comparison, but aims to compare similar compo-
nents of both theories on an overlapping domain. The
similar components refer to the formulation of pre-
scriptive guidelines by both theories. This is impor-
tant, given the different kinds of theories available in
literature (e.g., descriptive theories, explanatory theo-
ries, or design theories). In Normalized Systems, such
guidelines are referred to by stressing the determinism
of design (Van Nuffel, 2011). In Enterprise Ontology,
we find clear references to the importance of such
guidelines in the definition of architecture, which is
“the normative restriction of design freedom” (Dietz,
2006). The overlapping domain is the domain of busi-
ness processes, which is clearly addressed in Normal-
ized Systems Business Processes (cf. Section 2.1.
While business processes are defined within EO as
well, it should be noted that we interpret the prescrip-
tions of EO not only on the ontological level. In any
organization, the ontological models eventually need
to be extended to include the infological and datalog-
ical layers, and to specify an implementation. Imple-
mentation means “the particular subjects that fulfill
the actor roles at a particular time, the particular way
in which C-acts are performed, and the particular way
in which P-acts are performed. Several publications
focused on this subject, which show that a design is
obtained which is influenced by EO prescriptions, but
which can no longer be considered to be a design of
a social system by itself, or be entirely on the on-
tological level. For example, we mention research
to define use cases for information systems based on
DEMO models (Dietz, 2003a). This is in line with in-
sights from the generic systems development process
(GSDP) (Dietz, 2006, p. 71), which states that a func-
tional specification of an object system needs to be
made based on the constructional model of the using
system.
3 APPROACH
In order to compare the guidelines of EO and NSBP,
four categories should be considered: (1) Consistent:
guidelines from NSBP and EO prescribe the same de-
sign; (2) EO-ignorant: an NSBP guideline which has
no similar EO guideline; (3) NSBP-ignorant: an EO
guideline which has no similar NSBP guideline; (4)
Conflicting: a NSBP guideline, which prescribes a
different design than an EO guideline, or vice versa.
Certain guidelines are expected to be consistent, since
both EO and NSBP consider business processes as
modular structures, and propose guidelines to opti-
mize their design. However, given the different kernel
theories of both approaches, and their non-identical
goals, certain conflicting guidelines could be identi-
fied. Moreover, neither EO or NSBP claim to be com-
plete. The claim from Dietz that “we do not intend
to claim that ... even the whole ψ-theory is a suffi-
cient basis for achieving optimally performing enter-
prises” (Dietz, 2006, p. 81) indicates the validity of
the EO-ignorant category. The claim from Van Nuffel
that NSBP guidelines are necessary, but not sufficient,
indicates the validity of the NSBP-ignorant category.
We will adopt the work of Van Nuffel as our start-
ing point as it explicitly lists a set of 25 guidelines,
whereas the guidelines from EO have not been for-
mally consolidated in such list exhaustively enumer-
ating all guidelines incorporated in the method. Fur-
ther, given this starting point to determine for each
guideline to which category it belongs, the NSBP-
ignorant category will not be required in this paper.
The authors of this paper independently made a
classification of the NSBP guidelines. After integrat-
ing the result, differences were discussed, and the as-
sessment was iteratively refined. All three authors
have a sufficient background in both EO and NSBP.
The NSBP PhD dissertation (Van Nuffel, 2011) and
EO book (Dietz, 2006) were used as reference ma-
terials. Several academic publications were used
for additional details. Moreover, several cases (see
e.g., (Van Nuffel, 2011), (Dietz, 2006),
http://www.
demo.nl
) were consulted as an application of the
guidelines.
4 COMPARISON
Within this section, the actual comparison between
the practical guidelines resulting from the two the-
oretical approaches is made. Our discussion will
Consistency, Complementalness, or Conflictation of Enterprise Ontology and Normalized Systems Business Process
Guidelines
57
follow the division made within the PhD of Van
Nuffel (Van Nuffel, 2011): first, the general guide-
lines with respect to identifying business processes
are discussed. Second, the comparison continues with
the three additional guidelines that in specific cases
identify business processes. Third, the comparison
continues with the guidelines determining individual
tasks, and finally, the auxiliary guidelines are inves-
tigated. The business process patterns discussed in
the PhD of Van Nuffel (Van Nuffel, 2011) focus on
issues not discussed by Enterprise Ontology, and are
therefore not taken into account. This section lists the
names of the guidelines in italic and bold font. Next,
the guideline is summarized in italic. Then, the con-
sistency, complementalness or conflict with EO is dis-
cussed. An overview of these discussions is provided
in Table 1.
4.1 General Business Process
Guidelines
1.1 Elementary Business Process. A Business pro-
cess denotes a constrained sequence — i.e., sequence,
iteration or selection — of individual tasks represent-
ing state transitions in the life cycle of a single life
cycle information object. Within Enterprise Ontology
(EO), a P-fact is a factum, which is defined as “the
result or the effect of an act” (Dietz, 2006, p. 42).
Therefore, facta “can be conceived as status changes
of . . .an object in some class” (Dietz, 2006, p. 42).
Furthermore, the order in which facta occur is de-
termined by so-called occurrence laws (Dietz, 2006,
p. 43). The transaction is thus about a unique P-
fact transcending the transaction pattern, which can
be considered to be somewhat consistent with a NS
business process which is about state transitions of
a single life cycle information object as stated by
NSBP. The one-to-one relationship between a transac-
tion and a P-fact is in our opinionconceptually consis-
tent with the one-to-one relationship between a single
life cycle information object and a business process.
1.2 Elementary Life Cycle Information Object.
an information object not exhibiting state trans-
parency is a life cycle information object. Whereas
NSBP prescribes the criterion of state transparency
(i.e., when no proper state transitions should be made
explicit (Van Nuffel, 2011, p. 118)) to define whether
an information object is a genuine life cycle informa-
tion object processed in a business process, Enterprise
Ontology does not explicitly state a rule, criterion or
law that in all circumstances denotes what a single P-
fact is. There are evidently ways and requirements a
P-fact should adhere to, but no general identification
mechanism seems to be made explicit:
“We conceive the result of a production act as a
particular change in the state of the system’s ob-
ject world” (Dietz, 2006, p. 58);
“The object world reflects the produced things
(e.g., goods or services) that are delivered to the
elements in the environment” (Dietz, 2006, p. 58).
As a consequence, – although it could be argued that
only most fine-grained production facts exist (and
therefore, that production facts are defined unambigu-
ously), but that they can be aggregated to simplify
models it seems that identification of production
acts in EO is not unambiguous: it depends on what
is considered to be the system and environment, and
different production facts can be identified depending
on the aggregation level taken into account. More-
over, elementary life cycle objects can also refer to in-
fological and datalogical production facts. Therefore,
the authors categorize this guideline as EO-ignorant.
1.3 Aggregated Business Process. In order to rep-
resent an aggregated business process, an aggregated
life cycle information object has to be introduced. In
EO, a business process is based on the composition
axiom: “a business process is a collection of causally
related transaction types, such that the starting step
is either a request performed by an actor role in the
environment or a request by an internal actor role to
itself (Dietz, 2006, p. 103). Based on this defini-
tion, the operational cycle (Dietz, 2006, p. 163) can
be understood, which specifies that certain end-to-end
processes cannot be considered as causally related
transactions. Since the NSBP guideline is explicitly
aimed towards representing any required end-to-end
process, both theories are conflicting in most situa-
tions.
1.4 Aggregation Level. Tasks performed on a dif-
ferent aggregation level denote a separate business
process. Although in the PSD-diagrams the causal
and conditional links are enriched with cardinalities
that describe the relationship between different trans-
actions, nowhere is indicated that when an analyst dis-
covers an one-to-many relationship between two can-
didate transactions, both should be separated. Fur-
thermore, this latter relates to the aggregation level on
which production facts are defined, since a produc-
tion fact defines a transaction. Again, this does not
result in a guideline to actually separate the transac-
tions. Therefore, EO seems to be ignorant with re-
spect to this design issue.
Third International Symposium on Business Modeling and Software Design
58
1.5 Value Chain Phase. The follow-up of an orga-
nizational artifact resulting from a value chain phase
denotes a different business process. While some
arguments can be made for the consistency of this
guideline, the most important argument seems to in-
dicate a conflict. For example, the operation axiom
might indicate value chain phases as separate trans-
actions, although it is dependent on the aggregation
level on which the P-facts are defined. Moreover,
the composition axiom illustrates the possible nesting
required to integrate the different phases. However,
the transaction axiom results in design decisions like
explicitly stating that the Order phase belongs to the
Delivery process in a typical Customer Order process
scope. With respect to the latter, NSBP clearly state
that these phases should be separated as they denote
separate concerns (Van Nuffel, 2011, p.132-34). In
this way, NSBP seems to consider concerns a level
“deeper” as it explicitly considers a delivery not to
belong to the Order Phase, but as a separate process
in the aggregated business process Customer Order.
Therefore, both theories seems to disagree with re-
spect to this design issue.
1.6 Attribute Update Request. A task sequence to
update an attribute of a particular life cycle informa-
tion object that is not part of its business process sce-
narios, is represented by an Attribute Update Request
business process. The guideline prescribes to separate
state transitions dealing with modifying an attribute
of a life cycle information object that does not be-
long to the business process scenarios (i.e., included
in the process). Enterprise Ontology however, con-
siders such requests to change to be part of the trans-
action. Mostly, they can be represented by one of the
four cancelation patterns. As such, this represents a
conflict between the two theories, although they com-
ply with each other on modifications that do belong to
the business process scenarios.
1.7 Actor Business Process Responsibility. Actor
business process responsibility indicates a separate
business process if different actors are responsible for
a different set of tasks, of which the task allocation be-
longs to different process owners. The operation ax-
iom declares actor roles to denote chunks of authority,
responsibility and competence. Furthermore, follow-
ing Enterprise Ontology, a single transaction can only
be executed by a single actor role. In this way, this
notion is equivalent to stating that state transitions of
a particular life cycle information object being part
of the responsibility of a particular process owner de-
note a separate business process. Furthermore, in ad-
dition to EO, also NSBP identifies the only vaguely
described notion of process ownership within litera-
ture. As a consequence, NSBP opts for a clear iden-
tification of such process ownership, which seems to
be very closely related to EO’s notion of authority.
1.8 Notifying Stakeholders. Because notifying, or
communicatinga message to, stakeholders constitutes
an often recurring functionality in business processes,
a designated business process will perform the re-
quired notification. EO considers notifying stake-
holders as performing coordination acts, which are
part of an ontological transaction that creates a sin-
gle P-fact. However, NSBP identifies the concern of
notifying stakeholders to clearly differ from the con-
cerns taken care of by other business process (e.g., de-
livering an order, recruiting an employee, etc.): “de-
livering a message in the correct format to the in-
tended recipients at the right time in an unchanged
format, with the related fault handling (Van Nuffel,
2011, p.143). These concerns refer to implementation
details, which are not considered on the ontological
level. Therefore, EO theory is ignorant with respect
to this design guideline.
1.9 Payment. Because paying a particular amount
of money to a particular beneficiary constitutes an
often recurring (technical) functionality in business
processes, a designated (technical) business process
will perform the required payment. The payment
business process/transaction is identified by both the-
ories, and can be considered as consistent. Various
DEMO cases illustrate this. It should be noted that
NSBP requires that at least the execution phase of
payment processes is implemented using a reusable
business process, in order to preventcombinatorial ef-
fects. This is not clear from the DEMO cases, which
explicitly define multiple payment transactions.
4.2 Business Process Guidelines
2.1 Product Type. A different type of product or
service denotes a main concern, and thus indicates
a different business process. The composition ax-
iom seems to indicate that Enterprise Ontology also
recognizes the existence of transactions that although
being enclosed by the same transaction, do consti-
tute individual and independent transactions based on
a product structure. But again, no clear rules could
be identified, implicitly stated by “one could apply a
finer-grained product structure” (Dietz, 2006, p. 170).
The notion of a product type defined by Van Nuffel
(Van Nuffel, 2011, p. 149) allows some interpretation
as well, namely the domain expert who will identify
the characteristic dimensions on which product types
Consistency, Complementalness, or Conflictation of Enterprise Ontology and Normalized Systems Business Process
Guidelines
59
exhibit similar properties. As a consequence, we cat-
egorize this design issue as an EO-ignorant one.
However, if the Logistics example discussed by
NSBP is taken into account, the design issue also
seems to indicate conflicting statements by the two
theories. The NSBP separate the Logistics processes
based on the following types: non-food, food, quickly
rotating, slowly rotating, and so on. On the other
hand, EO theory seems to declare that these product
types do not cause another type of P-fact to be cre-
ated, and thus no separate transaction to be executed.
This could indicate a potential conflict.
2.2 Stakeholder Type. Stakeholder type should
principally be considered a cross-functional concern
(i.e., a concern which does not require a life cycle in-
formation object by itself), expect for those business
processes where the stakeholder type denotes the life
cycle information object. Whether the theories com-
ply, comes down to the question: does EO consider
a transaction to be independent from the actor role
for which it is potentially performed? In the PhD
of Van Nuffel, a case about Human Resources (HR)
processes is discussed in which it is clearly demon-
strated that the assignment processes for a statutory
employee and a non-statutory employee differ. Based
on the authors’ knowledge, EO does not provide any
rule or prescription about the potentially different na-
ture of a transaction. For example, in the Educational
Administration case, no separate transactions are cre-
ated based on different student types.
2.3 Access Channel. The concept of an access
channel indicates a cross-functional concern. In EO
publications no explicit referral to this design ques-
tion could be found. However, implementation is ex-
plicitly out of scope for EO: EO “fully abstracts from
the implementation [of C-acts]”, which includes “the
particular way in which C-acts are performed” (Di-
etz, 2006, p. 83). Consequently, it can be argued that
the theories comply as EO does not explicitly states
a different access channel denotes a separate transac-
tion. Consider in this context the pizzeria case (Dietz,
2006, p. 166). The transaction T01: Completion con-
tains all access channels to place an order.
4.3 Task Guidelines
3.1 A Single Functional Task - Overview. A task
represents a functional entity of work that either re-
sults in a single state transition of a single infor-
mation object type, or refers to an Update or Read
task on a single information object type. Where
NSBP specifically describes what a single task (or
step within a business process) can be, our analysis
of the EO fails to find equivalent rules. Of course,
the transaction axiom identifies single acts (result-
ing in facts) within the transaction which might in-
dicate consistency. However, the authors seem to find
more evidence to categorize it as EO-ignorant. For in-
stance, consider the acceptance of a stated P-fact con-
sisting of an evaluation of its quality by performing
three quality tests and then communicating the out-
come to the initiator actor role which is authorized,
responsible and competent to accept the P-fact, who
will communicate it to the executor. EO considers
this example to be part of the Accept C-act whereas
NSBP prescribes to separate it in five different tasks,
and two instances of the Notification business pro-
cess. Thus, based on our analysis, we consider it to
be EO-ignorant.
3.2 CRUD Task. Each of the Create - Read - Up-
date - Delete (CRUD) operations constitutes a single
task. Since these tasks are on the infological and dat-
alogical layers, this guidelines is EO-ignorant.
3.3 Manual Task. Every manual task of which the
initiation and completion has to be known, has to be
designed as a separate task. EO makes abstraction of
the implementation of C- and P-acts (also see discus-
sion of 2.3 Access Channel). Therefore, EO is igno-
rant with respect to this guideline.
3.4 Managing Time Constraint. The management
of a time constraint denotes a separate task because
it represents the individual concern of managing a
particular time constraint. In EO, a time aspect only
seems present in the time-aspect of the proposition of
a P-fact (Dietz, 2006, p. 84) and self-initiating trans-
actions (Dietz, 2006, p. 99). However, EO makes no
claims whatsoever with respect to (not) separating an
individual time constraint. As such, we categorize the
guideline to be EO-ignorant.
3.5 Business Rule Task. A single business rule
should be separated as a single task. An individ-
ual business rule should be isolated in its designated
task following NSBP. EO acknowledges that busi-
ness rules can sometimes be existential laws, as ex-
pressed in the state model, or action rules, which are
expressed in the action model (Dietz, 2006, p. 196).
In this sense, both seem to be consistent. However,
EO does not explicitly states that every single busi-
ness rule should be isolated. Therefore, EO seems to
be rather ignorant to this design issue.
Third International Symposium on Business Modeling and Software Design
60
3.6 Bridge Task. When a business process instance
operating on an instance of life cycle information ob-
ject type I has to create a business process instance of
another life cycle information object type L, this func-
tionality is designed as a bridge task that initiates the
creation of the instance of the life cycle information
object L, and represents a state transition on the in-
stance of I. As already illustrated above, the composi-
tion axiom of EO denotes the nesting of transactions.
As such, it is illustrated that the Request C-act can be
“triggered by another transaction (i.e., the executor
of an enclosing transaction can initiate an enclosed
transaction). The Result structure analysis step of the
DEMO methodology also adds to this. Conceptually,
this is what a bridge task represents: it triggers the
execution of another business process.
3.7 Synchronization Task. When a business pro-
cess instance operating on a life cycle information
object I has to inform a business process instance of
another life cycle information object L, a synchro-
nization task, representing a state transition on the
instance of I, alters the state of the business process
instance of L. The NSBP synchronization task con-
ceptually equals the waiting conditions specified in
the EO model based on the Result structure analysis,
following the composition axiom.
3.8 Synchronizing Task. A synchronizing task rep-
resents the task receiving information from another
business process’s execution, in order to continue
the business process control flow. Equivalent to the
Bridge task, also the Accept C-act in the EO trans-
action pattern represents conceptually the same as a
synchronizing task. It allows the enclosing transac-
tion/business process to continue, and thus is the end
of the waiting condition.
3.9 Actor Task Responsibility. A task cannot con-
sist of parts that are performed by different actors.
Here NSBP is consistent with EO, as the operation
axiom states that actor roles are elementary chunks
of authority, responsibility and competence. Thus the
fact that another actor role is authorized, responsible
and competent to perform a particular task, suffices to
split this task from any other task another actor role is
authorized, responsible, and competent to execute.
4.4 Auxiliary Guidelines
4.1 Unique State Labeling. Each state of a life cy-
cle information object has to be unique. The first
auxiliary guideline, Unique State Labeling, states that
each state of a life cycle information object should be
unique. Thus, it indicates the necessity to uniquely
define the states a business process can transverse.
Also EO identifies unique labels as each coordina-
tion act and each transaction are uniquely labeled; and
even more it states that facts can be created, but can-
not be undone (Dietz, 2006, p. 82). Thus theories are
considered to be consistent.
4.2 Unique State Property. A life cycle information
object instance can only be in a single state at any
time. Also EO declares a transaction has a unique
status: the last performed fact, which is defined in EO
as a state transition in the C- or P-world (Dietz, 2006,
p. 82). Thus theories are considered to be consistent.
4.3 Explicit Business Process End Point. If a busi-
ness process type has multiple possible outcomes,
each of these scenarios should have its dedicated end
point reflecting the respective end state of a business
process instance. EO specifies through its transaction
patterns (basic-standard-cancelation) that every sce-
nario should be explicitly described. In this way, it is
consistent with NSBP as every business process’ exe-
cution results in a specific end point/state, and not in
a general state “finished”.
4.4 Single Routing Logic. A split/join element in a
business process’s control flow should only represent
a single split or join routing expression. Essentially
EO does not discuss this proposed guideline, so it is
considered to be EO-ignorant. However, it can be ar-
gued that both theories are consistent because within
the transitions between the different C-facts and P-
fact that are exhaustively described in the transaction
pattern, no violation to the NSBP guideline was iden-
tified. Further research should identify whether this
non-violation is purposefully and thus the theories
are consistent or rather by chance and thus remains
EO-ignorant.
5 DISCUSSION
Table 1 summarizes the comparison made in the pre-
vious section. A bullet denotes that the identified cat-
egory is determined without any doubt. An open cir-
cle means the categorization still needs further elicita-
tion as a unique categorization could not be identified.
When scanning the table, it can be argued that
the theories comply on many points (i.e., at least 10
out of 25 guidelines are consistent), indicating that
a surprising overlap exists between guidelines pre-
scribed by EO and NSBP, given their different the-
Consistency, Complementalness, or Conflictation of Enterprise Ontology and Normalized Systems Business Process
Guidelines
61
Table 1: Consistency of NSBP guidelines and EO.
Consistent
EO-ignorant
Conflict
1.1 Elementary Business Process
1.2 Elementary Life Cycle Information
Object
1.3 Aggregated Business Process
1.4 Aggregation Level
1.5 Value Chain Phase
1.6 Attribute Update Request
1.7 Actor Business Process Responsi-
bility
1.8 Notifying Stakeholders
1.9 Payment
2.1 Product Type
2.2 Stakeholder Type
2.3 Access Channel
3.1 A Single Functional Task -
Overview
3.2 CRUD Task
3.3 Manual Task
3.4 Managing Time Constraint
3.5 Business Rule Task
3.6 Bridge Task
3.7 Synchronization Task
3.8 Synchronizing Task
3.9 Actor Task Responsibility
4.1 Unique State Labeling
4.2 Unique State Property
4.3 Explicit Business Process End
Point
4.4 Single Routing Logic
oretical backgrounds. The EO-ignorant category is
mostly discovered in the NSBP task rules. Almost
all observations can be contributed due to the differ-
ent abstraction level (EO does not consider these de-
sign questions), or the lack of a clear available answer
in the different publications (e.g., Stakeholder Type).
Consequently, NSBP seems to answer design ques-
tions EO does not answer or does not consider. Re-
garding the conflicting guidelines, some genuine con-
tradictions (e.g., Aggregated Business Process) were
identified. These conflicts should be clarified in fu-
ture research, especially because most conflicts occur
in the core (i.e., the first twelve) NSBP guidelines.
Moreover, more in-depth analysis is required,
since the reason for consistent design decisions may
differ. For example, consider the library case. The
transaction T03: Reduced fee approval is a separate
transaction in EO because it is executed by a dif-
ferent actor (Dietz, 2006, p. 144). In NSBP, it is a
different transaction because “it denotes a separate
concern, because it recurs in at least two situations”
(Van Nuffel, 2011, p. 217) (i.e., when creating a new
member and when collecting the yearly fee). Ad-
ditionally, the NSBP-ignorant category needs to be
elaborated upon as well. For example, various coor-
dination acts are not required to be modeled in NSBP,
for example when they are implicit. The explicitation
of this category could especially aid the completeness
of NSBP models.
Nevertheless, the authors hypothesize that
given the consistency between both theories and un-
der the condition that the different abstraction levels
on which they clearly operate do outweigh the con-
tradictions, or that contradictions could be resolved
by clearly identifying the abstraction levels on which
both theories have their proven scientific importance
— a method combining both theories to analyze busi-
nesses can be proposed. We will further elaborate on
this method and its applications in future research.
6 CONCLUSIONS
In this paper, we explored to which extend the pre-
scriptiveguidelines related to the business process do-
main of EO and NSBP are consistent, complementary,
or conflicting. We explained how both approaches
offer theory-based guidelines to design business pro-
cesses, and discussed in detail the assessment of the
various NSBP guidelines. Moreover, we suggested
several possibilities for further research, to work to-
wards an integrated method for organizationaldesign.
ACKNOWLEDGMENTS
P.D.B. is supported by a Research Grant of the
Agency for Innovation by Science and Technology in
Flanders (IWT).
REFERENCES
Avenier, M.-J. (2010). Shaping a constructivist view of
organizational design science. Organization Studies,
31(9-10):1229–1255.
Dietz, J. (2003a). Deriving use cases from business pro-
cess models. In Song, I.-Y., Liddle, S., Ling, T.-W.,
and Scheuermann, P., editors, Conceptual Modeling
- ER 2003, volume 2813 of LNCS, pages 131–143.
Springer Berlin Heidelberg.
Dietz, J. L. G. (2003b). The atoms, molecules and fibers of
organizations. Data Knowl. Eng., 47:301–325.
Third International Symposium on Business Modeling and Software Design
62
Dietz, J. L. G. (2006). Enterprise Ontology: Theory and
Methodology. Springer.
Dietz, J. L. G. and Albani, A. (2005). Basic notions re-
garding business processes and supporting informa-
tion systems. Requir. Eng., 10:175–183.
Dietz, J. L. G., Hoogervorst, J., Albani, A., Aveiro, D.,
Babkin, E., Barjis, J., Caetano, A., Huysmans, P.,
Iijima, J., Van Kervel, S., Mulder, H., Op t Land,
M., Proper, H. A., Sanz, J., Terlouw, L., Tribolet, J.,
Verelst, J., and Winter, R. (2013). The discipline of
enterprise engineering. International Journal of Or-
ganisational Design and Engineering, 3(1):86–114.
Dreyfus, D. (2007). Information system architecture: To-
ward a distributed cognition perspective. In ICIS 2007
Proceedings. Paper 131.
Galbraith, J. R. (1974). Organization design: An informa-
tion processing view. Interfaces, 4(3):28–36.
Huysmans, P. (2011). On the Feasibility of Normalized En-
terprises: Applying Normalized Systems Theory to the
High-Level Design of Enterprises. PhD thesis, Uni-
versity of Antwerp.
Huysmans, P., Bellens, D., van Nuffel, D., and Ven, K.
(2010). Aligning the constructs of enterprise ontology
and normalized systems. Lecture notes in business in-
formation processing, 49(1):1–15.
Huysmans, P., Oorts, G., and De Bruyn, P. (2012). Position-
ing the normalized systems theory in a design theory
framework. In Shishkov, B., editor, Proceedings of the
Second International Symposium on Business Model-
ing and Sofware Design (BMSD2012), pages 33–42,
Geneva, Switzerland.
Kaisler, S. H., Armour, F., and Valivullah, M. (2005). En-
terprise architecting: Critical problems. In Proceed-
ings of the 38th Hawaii International Conference on
System Sciences, volume 8, page 224b, Los Alamitos,
CA, USA. IEEE Computer Society.
Kelly, D. (2006). A study of design characteristics in evolv-
ing software using stability as a criterion. Software
Engineering, IEEE Transactions on, 32(5):315–329.
Krouwel, M. and Op’t Land, M. (2011). Combining demo
and normalized systems for developing agile enter-
prise information systems. In Albani, A., Dietz, J.,
and Verelst, J., editors, Proceedings of EEWC 2011,
pages 31–45. Springer-Verlag.
Mannaert, H. and Verelst, J. (2009). Normalized Systems—
Re-creating Information Technology Based on Laws
for Software Evolvability. Koppa, Kermt, Belgium.
Mendling, J., Reijers, H. A., and van der Aalst, W. M. P.
(2010). Seven process modeling guidelines (7pmg).
Inf. Softw. Technol., 52(2):127–136.
Op ’tLand, M., Krouwel, M. R., van Dipten, E., and Verelst,
J. (2011). Exploring normalized systems potential for
dutch mod’s agility. In Harmsen, F., Grahlmann, K.,
and Proper, E., editors, Proceedings of PRET 2011,
pages 110–121. Springer-Verlag.
Schenherr, M. (2008). Towards a common terminology in
the discipline of enterprise architecture. In ICSOC
Workshops’08, pages 400–413.
Van Nuffel, D. (2011). Towards Designing Modular and
Evolvable Business Processes. PhD thesis, University
of Antwerp.
Van Nuffel, D., Mannaert, H., De Backer, C., and Verelst, J.
(2009a). Deriving normalized systems elements from
business process models. In Boness, K., editor, Pro-
ceedings of ICSEA 2009, pages 27–32. IEEE Com-
puter Society.
Van Nuffel, D., Mulder, H., and Van Kervel, S. (2009b). En-
hancing the formal foundations of bpmn by enterprise
ontology. In CIAO! / EOMAS, pages 115–129.
van Reijswoud, V. (1999). Model based business system
transformation. In Pries-Heje, J., Ciborra, C., Kautz,
K., Valor, J., Christiaanse, E., and Avison, D., editors,
Proceedings of ECIS 1999, pages 600–615, Copen-
hagen.
Winter, R. and Albani, A. (2013). Restructuring the design
science research knowledge base. In Baskerville, R.,
De Marco, M., and Spagnoletti, P., editors, Designing
Organizational Systems, volume 1 of Lecture Notes in
Information Systems and Organisation, pages 63–81.
Springer Berlin Heidelberg.
Consistency, Complementalness, or Conflictation of Enterprise Ontology and Normalized Systems Business Process
Guidelines
63