A Survey on Modelling Knowledge-intensive Business Processes from
the Perspective of Knowledge Management
Christoph Sigmanek and Birger Lantow
University of Rostock, 18051 Rostock, Germany
Keywords: Knowledge Management, Business Process Oriented Knowledge Management, Knowledge-intensive
Processes, Process Modelling.
Abstract: Existing modelling approaches for knowledge-intensive business processes try to match the character of
these processes by specific modelling concepts and methods. The approaches differ significantly depending
on the focus of modelling. DeCo and KIPN for example recommend to be less strict on control flow
orientation. KMDL allows for modelling down to the level of individuals. SBPM and KPR as well
emphasize a detailed model and additionally underline the importance of distributed modelling. GPO-WM
in contrast suggests avoiding too much details. However, which approach or what level of abstraction is
now suitable for which modelling task from the perspective of knowledge management? Can the models be
reused for other tasks? The search for the "right" way for modelling knowledge-intensive processes and
issues derived therefrom are in the focus of discussion.
1 INTRODUCTION
Nowadays, knowledge is recognized as an important
enterprise resource. Thus, knowledge management is
derived as a management task. Here, business
process oriented knowledge management aims at the
ways of dealing with knowledge and requirements
for knowledge and knowledge activities (use,
production, and transfer of knowledge) in business
processes. Remus puts knowledge-intensive business
processes in the focus of a process-oriented
knowledge management (Remus, 2002, p.108). Here
lies the biggest success potential for knowledge
management. Table 2 summarizes the typical
characteristics of knowledge-intensive processes.
They are commonly found in knowledge-intensive
domains and are characterized by a high degree of
complexity. Control flow varies widely, so that a
high coordination and communication effort is
required. Knowledge-intensive processes are often
poorly structured, show a high number of
participants (experts), and are difficult to plan. Due
to their nature, it is difficult to reassign tasks to
different individuals (Remus, 2002, pp. 104-117).
Heisig sees as the most relevant criterion of
knowledge-intensive processes that required
knowledge can be planned ahead only in a limited
manner (Heisig, 2002).
In order to model the knowledge support of
processes, it is no longer sufficient to restrict the
process to a sequence of activities, events and
decisions consequently. Rather, it is necessary that
important components from the perspective of
knowledge management can be presented and that
the modelling methodology fits to the special
characteristics of these processes.
Considered to model components are:
1. Knowledge activities
a. Knowledge creation and knowledge use
(Allweyer, 1998, pp. 165)
b. Knowledge transfer
2. Knowledge resources
a. Knowledge carriers (Allweyer, 1998, pp.
165)
b. Knowledge sources
3. Knowledge Structure Conditions
a. ICT-involvement, the tech-nologies used
(Scheer, 1998, pp.63-65)
b. Organisational requirements and corporate
culture (Lehner et al., 2007)
On the same hand, the high complexity and
variability of knowledge-intensive processes has to
be considered by the modelling methodology.
In business processes, activities are performed in
Sigmanek, C. and Lantow, B..
A Survey on Modelling Knowledge-intensive Business Processes from the Perspective of Knowledge Management.
In Proceedings of the 7th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2015) - Volume 3: KMIS, pages 325-332
ISBN: 978-989-758-158-8
Copyright
c
2015 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
325
a logical sequence by individuals who act in roles.
For the successful completion of knowledge-
intensive tasks the individuals must satisfy their
need for knowledge with the help of knowledge
activities. Here, they interact more or less efficiently
through various media with other people and IT
systems. During these interactions new knowledge is
generated constantly. However, it can only partly be
preserved. In order to transfer these processes into a
model, a structured approach is needed, which
selects a proper level of detail. Since good models
represent an appropriate part of the real world, it is
possible to draw an analysis on how processes can
be improved in the real world based on these
models. Furthermore, the modelling provides the
advantage that knowledge about business processes
can be documented and shared. In the following
section, state-of-the-art methods to collect and
analyze knowledge-intensive business processes are
presented. They should ensure a correct model of
implicit and explict knowledge. The third section
then provides an evaluation of the presented
approaches in relation to the initially formulated
modelling requirements. The final section
summarizes the found open challenges for modelling
knowledge-intensive processes.
2 APPROACHES TO
MODELLING AND ANALYSIS
OF KNOWLEDGE-INTENSIVE
BUSINESS PROCESSES
In the literature there is a variety of approaches to
modelling knowledge-intensive processes and
modelling of knowledge management aspects in
process models.
This section discusses a selection. A part of
approaches has been selected based on an analysis of
the citation (> 50 citations in Google Scholar). Thus
a high scientific impact can be assumed.
Additionally, DeCo and KIPN (> 10 citations) have
been selected, which have been published more
recently. They do not have a comparable citation
count. However, we assume that the limited
timeframe is the major reason for that. Thus we
discuss well established approaches (>50 citations)
and current trends and ideas (DeCo and KIPN).
2.1 Knowledge Modelling and
Description Language
The Knowledge Modelling and Description
Language (KMDL) is a method for modelling
knowledge-intensive business processes that is still
Table 1: Attributes of knowledge-intensive processes according to Remus (2002).
Attribute class Typical attribute values of knowledge intensive processes
Process independent attributes
(knowledge intensive domain)
often decentralized networking organization showing a goal oriented support of
knowledge transfer, e.g. by incentives
Knowledge intensive domain (key technologies)
Complex relations between processes
Attributes concerning the process
(variablility)
Complex processes having many dependent single activities, actors that work in
interdisciplinary teams
Many exceptions, unpredictable control flow and results
Poorly structured, only ex-post modelling possible
High coordination and communication effort between the actors, needs knowledge
from different domains
Generating knowledge-intensive products and services
Only imprecise controlling possible, often qualitative goals
Low number of process instances having long running times
Process case driven, no standard process
Attributes concerning tasks (poor
transferability)
Productivity of knowledge work usually not measurable
Long learning and and training periods necessary
Chaotic workplace
Tasks are communication oriented, require a lot of information, case driven
Typical tasks are: decision making, problem solving, analysis and evaluation,
controlling and management
Attributes concerning actors (experts)
Highly autonomous actors
Unstructured and individualized rules and routines
High competency, learning aptitude, creativity and innovation required
Attributes concerning resources
(complex resources)
Wide use of knowledge management instruments, informal knowledge transfer
Knowledge often hardly accessible and highly depending on the context
High amount of handled knowledge, cost intensive knowledge acquisition.
RDBPM 2015 - Special Session on Research and Development on Business Process Management
326
under active development. (Gronau, 2009, pp. 76-
79) (KMDL, 2014).
KMDL is distinct from other approaches due to
its person or individual related knowledge
modelling. Here, the method provides an explicit
modelling of individual knowledge conversions as
introduced by Nonaka and Takeuchi (1995). In
addition, the method provides various analysis views
and comparative patterns for analysis.
The KMDL method is based on a nine-phase
process model. The active participation of the
project partner is required for successful knowledge
management projects. First, in phase 0 the
organizational framework is set. In Phase 1, the
definition of the intended objectives of the project
follows. Then, in Phase 2, the processes of the
project partner are iteratively registered, refined and
validated. This is the base for deriving knowledge
intensive processes in phase 3. In Phase 4 these
knowledge-intensive processes are iteratively
modelled and sequentially analysed for possible
improvements in phase 5. Specifically, the focus lies
on finding weaknesses in order to derive suggestions
for improvement from them. Then they are classified
and evaluated, and finally there is an assessment of
the potential for improvement. In the following 6
th
phase to-be concept is developed with the partners,
which will be implemented in phase 7. In the final
phase 8 the whole process will be evaluated together
with the project partner. (Gronau, 2009, pp.75)
KMDL defines three views for the different
requirements of modelling knowledge-intensive
processes. The process view shows the business
processes at an abstract level. Here, individual
activities are displayed in their logical order in
conjunction with the involved resources. Activities
are broken down to knowledge transformations (e.g.
socialization) in the activity view. The activity view
is also the basis for the communication view, which
describes how individual knowledge transformations
are performed in conversations. Conversations are
characterized by location (the same location /
different location) and time (synchronous /
asynchronous). (Pogorzelska, 2009, pp. 21-45)
KMDL analysis is based on these models. In the
first analysis the frequencies of knowledge objects
and conversation types (e.g. socialization) are
counted and evaluated accordingly. A high number
of socialization activities may for example indicate
that too little knowledge is explicated. A recurring
knowledge resource or a person who is involved in
many activities in contrast may point to a possible
bottleneck or a key function. Thus, knowledge needs
are matched with knowledge services, and there is
an assessment of the models regarding specific
patterns. There are concrete improvement actions
indicated for each pattern. (Pogorzelska, 2009, pp.
49-79)
KMDL provides a holistic approach to modelling
and improvement of knowledge-intensive processes.
For modelling with KMDL, the tool K-Modeler (K-
Modeler, 2014) is available. The method has been
criticized for the extra effort that is induced by the
collection of the individual knowledge
transformations. It can only be justified by better
coverage of improvement measures for knowledge
management (Krallmann et al., p 417). Thus, this
method is very time consuming and the results
strongly depend on the trust of the interviewees and
the skills of the interviewer (Müller et al., 2012,
pp.362).
2.2 Knowledge Process Reengineering
The Knowledge Process Reengineering (KPR)
approach (Allweyer, 1998, pp.163- 168) is a seven-
step approach to improve the handling of the
resource "knowledge". In particular, the approach
aims at effective knowledge sharing, good
documentation and easy access to knowledge. KPR
was developed for use in enterprises and can be
supported by ARIS models. The individual phases,
starting with the strategic knowledge planning,
going on about the actual analysis and target
conception, to implementation, run linearly in KPR.
Re-entering a completed phase is not considered.
Instead, the approach provides an ongoing testing
and improvement process in the final phase.
KPR starts with strategic knowledge planning.
Here is determined how knowledge management can
support the company's strategic objectives. Models
which relate the core business processes to the
strategic business objectives help in this phase.
Subsequently, an as-is modelling of knowledge
usage and transformation is performed. The KPR
approach uses EPC for process description due to its
tight coupling to ARIS. Then knowledge carriers,
knowledge categories and knowledge needs must be
captured in knowledge structure diagrams,
knowledge maps and additional information in the
EPC diagrams. (Allweyer, 1998, pp.164-166)
Once the as-is situation has been modelled, its
analysis begins. Here, critical knowledge
monopolies, unsatisfied knowledge needs,
inadequate knowledge profiles of employees etc. are
revealed. The following development of a to-be
concept for knowledge handling provides measures
to solve the previously found issues. This is done for
A Survey on Modelling Knowledge-intensive Business Processes from the Perspective of Knowledge Management
327
example, by target knowledge profiles for
employees or changes in business processes. After
the to-be concept is set, realisation concepts for the
organisation and the ICT are developed. The
realization concept regarding the organisation
includes staff trainings regarding new processes and
new IT systems. The ICT realization concept
includes the selection of appropriate IT solutions,
the definition of content structures and system
integration.
After implementing the to-be concept by the
developed realization concepts, a phase of testing
and possibly improving starts. (Allweyer, 1998,
pp.166-168)
KPR thus offers an approach, which aims to
anchor technologies of knowledge management in
the working procedures of employees. The strong
dependence on the underlying ARIS architecture, the
requirement to model all the knowledge of a
company, and the lack of a detailed description of
single method steps are critical issues of KPR. In
consequence, KPR is only a specific process model
for the integration of IT in knowledge-intensive
business processes.
2.3 PROMOTE
Hinkelmann et al., (2002, pp. 65-68) presented with
process-oriented methods and tools for knowledge
management (PROMOTE) a technology-
independent method for the management of
functional and process knowledge. It is an evolution
of the business process management system
framework (BPMS) (Karagiannis, 1995) and
supplements this by the software tool PROMOTE
(BOC, 2014). PROMOTE focuses on the
identification, modelling and integration of
processes that require and generate knowledge. The
software supports the processing of knowledge-
intensive activities by knowledge processes can be
activated context-specific. Furthermore, it provides
knowledge maps and topic maps as configurable
knowledge management tools and. Finally,
PROMOTE provides management capabilities for
knowledge flows and a model-based indexing of
documents with process- and role-specific access
rights.
As a prerequisite for the approach, the following
assumptions are made:
1. Knowledge processes can be modelled the same
way as business processes
2. Activities in business processes use knowledge.
Base for the use of PROMOTE the method steps
which provide high degree of freedom. Depending
on the context the order of these steps and the final
results may vary. The general goal is a support of
knowledge flows between knowledge-intensive
business processes. This knowledge flows can occur
within a business process, across business processes,
within a project, and even across projects. In
addition, external knowledge inflows by training,
internet research, etc. are possible. (Hinkelmann et
al., 2002, pp. 68-71)
Realization is done in the five phases “Aware
Enterprise Knowledge”, “Discover Knowledge
Process”, “Modelling Knowledge Processes and
Organisational Memory”, “Making Knowledge
Processes and Organisational Memory operational”
and “Evaluate Enterprise Knowledge”. In the first
phase corporate goals are adjusted and strategically
determined. These are for example products,
services, financial requirements and the
development of core competencies. The aim is an
alignment of the knowledge strategy with the
business strategy. (Hinkelmann et al., 2002, S. 73-
76)
In the ensuing “Discover Knowledge Process”
phase process knowledge is modelled. That means
knowledge of the logical sequence of activities
within a process, including participating
organizational units, application systems and
resources. In addition, knowledge with high
potential impact is identified by experts. This
includes decision-critical knowledge and knowledge
to create a service or a product (functional
knowledge). In addition, types of processed
knowledge, knowledge carriers, knowledge flows
and the forms of knowledge representation are
recorded. After the modelling of business processes
modelling and mapping of knowledge processes
takes place in the third phase. Knowledge processes
should replace knowledge flows by giving the
knowledge flow a methodology. If an agent requires
knowledge to carry out a task, then there are
different options to obtain this knowledge. For
example, he can turn to his colleagues or look up an
expert using yellow pages. To make documents
retrievable and therefore available for future use,
they are enriched with metadata. A document gets a
modification date, content keywords (tags) from
folksonomies (collections of tags), an author and
other information that are ideally already set by the
appropriate knowledge structures and form the
technical part of the organizational memory.
(Hinkelmann et al., 2002, pp. 76-84)
Phase 4 “Making Knowledge Processes and
Organisational Memory operational” implements
these knowledge processes in existing software.
RDBPM 2015 - Special Session on Research and Development on Business Process Management
328
Hence, during his work an agent can see
immediately which options he has to satisfy his
knowledge needs. For example, a direct link to
yellow pages for expert search can be provided,
having context specific search parameters already
set. An evaluation of the use of PROMOTE takes
place in the 5th phase. Thus, the contribution of
knowledge management can be measured by the
success of the company. (Hinkelmann et al., 2002,
pp. 84-90)
2.4 Declarative Configurable
Declarative Configurable (Deco) is a combination of
declarative modelling, model verification and
variability modelling to capture knowledge-intensive
processes. In DeCo, the knowledge-intensive
processes are modelled on three layers. The most
abstract layer is at design. Here, a configurable,
nondeterministic specification is created in
accordance with the process goals. In the at-
deployment layer, the process is configured in a
context that is close to the application domain.
Finally, a fully deterministic process execution trace
that maps a single process instance is created in the
at execution layer. (Rychkova and Nurcan, 2011, pp.
1-2)
Business processes are divided into prescriptive
processes and descriptive processes. Prescriptive
processes have predictable process flows, simple
tasks, and can be fully specified at design time. At
the opposite pole are the descriptive processes,
which include the knowledge-intensive processes.
.These complex tasks are based on cooperation
between different actors, can only be outlined at
design time. Principles underlying DeCo are: "Very
little is certain at design-time" and "Fixed constraint
often means lost opportunities". Therefore, nor
control flow is required in the at-design layer for
descriptive processes. Thus, the configurability
remains unlimited and critical decisions can be made
later on. (Rychkova and Nurcan, 2011, pp. 2-5)
Processes are configured in a specific context in
the at-deployment layer in order to allow
implementation for a certain application. Some
details may not be pre-configured because of their
vagueness. For configurable processes tasks are
assigned to roles, tasks are arranged or selected rules
are applied for example. In the at-execution layer,
the pre-configured processes are finally carried out,
leaving process tracks that are stored and thus
contribute to the construction of a knowledge base
and contribute to improving future processes.
(Rychkova and Nurcan, 2011, pp. 2-9)
Hence, DeCo helps with the controlled assembly
of important process specifications from predefined
process parts or process variants. The design phase
is controlled by central questions and after each
execution possible new paths are incorporated in the
initial or adapted model. The DeCo notation is an
adaptation ofthe BPMN standard: Optional objects
are marked by dashed lines, configurable objects by
bold lines. Furthermore, objects are enriched by tags
(e.g. <IN> for detailed information) in order to
describe knowledge-intensive processes. Mainly the
variability of knowledge-intensive processes is
covered by this approach. (Rychkova and Nurcan,
2011, pp. 5-10)
2.5 GPO WM
Heisig (2002, pp. 47-59) shows with "Business
Process Oriented Knowledge Management" (GPO-
WM) is an eight-phase model for the introduction of
knowledge management. Furthermore, the
company’s strengths and potentials related to the use
of the resource "knowledge" can be determined.
Important paradigms of GPO-WM are:
1. There should not be too much details in the
models.
2. There should be a close connection between the
method expert and the organization.
One possibility to keep a close connection to the
organisation that is subject to a GPO-WM project
could be the use of a company-specific modelling
language. In order to put the focus on relevant tasks
and processes, the central question “Does the task
contain base activities of knowledge management?”
is proposed. Basic tasks of knowledge management
are generating knowledge, storing knowledge,
transferring knowledge and applying knowledge.
During analysis, the focus is not on optimizing
particular activities such as storing explicit
knowledge in a database, but rather on consideration
of the entire frame. Hence, questions like “Where is
the generated knowledge reused?” are in the focus.
Problems are identified based on guiding questions
and possibly solved by best-practice solutions for
knowledge management. Thus, problems can be
discovered, which are not shown in a model. As a
result, knowledge management modules are
implemented and integrated into the respective
business processes. (Heisig, 2002, pp. 59-64)
2.6 KIPN
França et al., (2012) noted that there are already
numerous methods for modelling knowledge-
A Survey on Modelling Knowledge-intensive Business Processes from the Perspective of Knowledge Management
329
intensive processes and examine to what extent these
can map knowledge-intensive processes regarding
their specific process characteristics. Like Gronau
(2009, pp. 69-71) in a similar study, they conclude
that no approach from the literature covers all
relevant aspects. França et al. made a step further
and examined already established process modelling
languages such as BPMN and EPCs based on the
same criteria. It revealed that EPCs and BPMN
already meet many of the requirements for the
modelling of knowledge-intensive processes as
defined by Remus (Remus, 2002, pp 115-116).
Shortcomings are in the representation of poorly
structured processes, the relationships to other
business processes, knowledge transfer and the short
half-life of knowledge.
As a result, França et al. propose an ontology
(KIPO) for knowledge-intensive processes (França
et al., 2012, pp 499-504) as the basis of the
Knowledge Intensive Process Notation (KIPN).
KIPN is a graphical notation which is composed of
five diagrams. In the KIP diagram, activities are
represented including business rules, relations and
the level of abstraction. Modelling the control flow
of individual activities is not mandatory in KIP
diagrams. Communication between the actors, i.e.
exchanged messages, knowledge acquisition and
knowledge transfer, are shown in the socialization
diagram. Finally, alternatives and their advantages
and disadvantages are listed in a decision diagram.
In addition, the notation provides diagrams for goal
and for role modelling (França et al., 2013).
3 REVIEW OF THE PRESENTED
APPROACHES
In the previous section, different approaches to
handle knowledge-intensive business processes have
been introduced. In the various approaches it is clear
that knowledge-intensive processes need to be
treated differently from normal processes due to
their nature. All authors claim that setting the right
focus of modelling is crucial for the output of an
analysis. DeCo and KIPN recommend to diverge
from the control flow orientation of many modelling
languages. An alignment of knowledge or corporate
objectives respectively is the starting point of any
modelling or analysis project. KMDL provides the
ability to model on the level of individuals and
requires a strong incorporation of the modelled
organisation in the modelling process. This results in
a very context specific model which might not be
transferable and might have limited maintainability
due to the variability of knowledge-intensive
processes as described in DeCo.
KPR recommends a distributed modelling. Due
to a separate specification of concepts on one hand
and concurrent activities at the other hand, semantic
consistency can be guaranteed throughout the model.
For both - distributed modelling as well as a
modelling in a central model – several modelling
phases are proposed. In some approaches, the
detailed modelling has a high priority for subsequent
analysis, while GPO-WM discourages too high
detailing. Most methods solve identified problems
by the introduction of concrete knowledge
management tools and their integration into the
business processes. GPO-WM provides best
practices that cover certain problem categories. The
variety of objectives led to a multitude of different
modelling languages. However, none of them was
able to prevail in the literature to date.
Table 2 presents the main features of each
approach in a summary. The idea is to provide a
starting point for the selection of an existing
approach depending on the modelling requirements.
Furthermore, the limitations of existing should be
emphasized.
Table 2: Characteristics of the approaches.
RDBPM 2015 - Special Session on Research and Development on Business Process Management
330
In rows 1-3, general modelling aspects such as
goals, methodological support and required tools are
taken into account. These information can be used to
assess the practical applicability. First, the goal of
the modelling approach must fit to the goals of the
modelling project. Key point is the project focus - is
it a process improvement cycle on operational level
or is it a strategic alignment? Furthermore, a
modelling methodology as well as an appropriate
toolset should be provided for the applicability of an
approach.
The rest of the table addresses the specific
requirements of modelling knowledge-intensive
processes from Section 1 by a meta-analysis. Thus,
the approaches are matched against the theory of
knowledge intensive processes. Necessary modelling
artefacts are identified and the existence of
respective modelling constricts in the several
approaches is assessed. Regarding knowledge
activities (rows 4-5), there is a distinction between
knowledge use/generation and the representation of
knowledge transfers, because the latter is not
covered by all approaches while generally all
approaches address knowledge use and generation.
A supplement to this is then the modelling of
knowledge resources and their structures (row 6).
The possibility of taking into account the technical
infrastructure (ICTs) and organizational
environment is considered in rows 7-8. The last row
of the table aims at the ways how the approaches are
dealing with the complexity and variability in the
knowledge-intensive processes.
One result of the investigations is that all
approaches fail to describe the organizational
environment regarding knowledge intensive
processes. Additionally, only two address the
modelling of knowledge management system
components in terms of ICT support (KPR, GPO-
WM). Knowledge activities and knowledge
resources on the other hand are well covered. Thus,
the latter might be a starting point for model reuse in
different contexts because they are present in the
approaches independently from the defined goals.
4 CONCLUSION
Gai & Dang name three limitations of the process-
oriented knowledge management (Gai and Dang,
2010, pp. 3-4):
1. Not all knowledge activities are associated with
business processes. An example is the desire to
communicate during a coffee break.
2. The variability of the processes is not well
represented by many methods. Knowledge flows
are changing and are not tied to static processes.
3. Tacit knowledge is often treated inadequately.
Knowledge carriers are modelled as an attribute,
but this is not enough to represent the flow of
knowledge.
Limitation 1 generally applies to the approach of
business process-oriented knowledge management.
The context, in particular the organizational
conditions have a significant impact on the
performance of knowledge-intensive processes. This
applies not only to knowledge activities performed
outside the processes. The modelling approaches do
not take this into account (see table 2). However, the
context should be addressed in the models. The
limitations 2 and 3 are only partially addressed too.
As shown in table 2, not all of the approaches
explicitly model the different possibilities of
knowledge transfer. For dealing with the complexity
and variability of knowledge-intensive processes
two basic ways are being sought of: first, turning
away from the control flow orientation and second a
high abstraction level. It turns out that strategically
oriented modelling approaches (GPO-WM, KPR)
rely on a high level of abstraction, while approaches
to detailed specification and analysis of processes
(Deco, KIPN) have just little control flow
orientation. Besides this straight forward distinction,
guidelines for the application of particular method
components need to be developed: Which approach
fits best to what goals? How can the developed
models be the base for a sustainable knowledge
management? How can the effort and the benefits of
the approaches be evaluated?
In summary, there are only ideas and assistance
for addressing knowledge transfers in process-
oriented knowledge management, but not a complete
methodology. In a lot of cases, the consideration of
process variability, of the organizational
environment and a concrete methodology are
missing. Furthermore, effort and benefits of
knowledge management activities are poorly
addressed.
REFERENCES
Allweyer, T.: Wissensmanagement mit ARIS-Modellen.
In: Scheer, A.-W.: ARIS - vom Geschäftsprozeß zum
Anwendungssystem, 3. Auflage, Springer Verlag,
Berlin 1998, S. 162-168.
BOC Group: Wissensmanagement mit PROMOTE,
http://www.boc-group.com/de/landing-pages/wissens
management-mit-promoter/, (date of access:
A Survey on Modelling Knowledge-intensive Business Processes from the Perspective of Knowledge Management
331
22.12.2014).
França, J. B. S., Baião, F. A., Santoro, F. M.: Towards
Characterizing Knowledge Intensive Processes. In:
Proceedings of the 2012 IEEE 16th International
Conference on Computer Supported Cooperative
Work in Design, 2012, p. 497-504.
França, J.B.S., Baião, F.A., Santoro, F.M.: A Notation for
Knowledge-Intensive Processes. In: Proceedings of
the 2013 IEEE 17th International Conference on
Computer Supported Cooperative Work in Design,
2013, p. 190-195.
Gai, Y., Dang, Y.: Process-oriented Knowledge
Management: a Review on Strategy, Content, Systems
and Processes. In: IEEE International Conference on
Management and Service Science (MASS), August
2010.
Gronau, N.: Wissen prozessorientiert managen: Methode
und Werkzeuge für die Nutzung des
Wettbewerbsfaktors Wissen in Unternehmen,
Oldenbourg, Munich 2009.
Gronau, N., Müller, C., Uslar, M.: The KMDL Knowledge
Management Approach: Integrating Knowledge
Conversions and Business Process Modelling. In:
Karagiannis, D., Reimer, U.: Practical Aspects of
Knowledge Management. 5th International
Conference, PAKM 2004, Vienna December 2004.
Heisig, P.: GPO-WM: Methode und Werkzeuge zum
geschäftsprozessorientierten Wissensmanagement. In:
Abecker et al.: Geschäftsprozessorientiertes
Wissensmanagement, Springer Verlag, Berlin
Heidelberg 2002, p. 47-64.
Hepp, M., Roman, D.: An Ontology Framework for
Semantic Business Process Management. In:
Proceedings of Wirtschaftsinformatik 2007, Karlsruhe
Februar - März 2007.
Hinkelmann, K., Karagiannis, D., Telesko, R.:
PROMOTE: Methodologie und Werkzeug für
geschäftsprozessorientiertes Wissensmanagement. In:
Abecker et al.: Geschäftsprozessorientiertes
Wissensmanagement, Springer Verlag, Berlin
Heidelberg 2002, S. 65-90.
Karagiannis, D.: BPMS: Business Process Management
Systems. In: ACM SIGOIS Bulletin, New York August
1995/Vol. 16, No. 1, p.10-13.
KMDL Blog, http://www.kmdl.de/ (date of last access:
24.11.2014).
K-Modeler | KMDL Blog, http://www.kmdl.de/
?q=de/node/27 (date of last access: 24.11.2014).
Krallmann, H., Bobrik, A., Levina, O.: Systemanalyse im
Unternehmen: Prozessorientierte Methoden der
Wirtschaftsinformatik, 6. ed., Oldenbourg, Munich
2013.
Lehner, F. et al.: Erfolgsbeurteilung des
Wissensmanagements: Diagnose und Bewertung der
Wissensmanagementaktivitäten auf Grundlage der
Erfolgsfaktorenanalyse. In: Schriftenreihe
Wirtschaftsinformatik, Diskussionsbeitrag W-24-07,3.
ed., Passau 2007.
Müller, C., Bahrs, J., Gronau, N.: Considering the
Knowledge Factor in Agile Software Development. In:
Gronau, N. (Eds.): Modelling and Analyzing
knowledge intensive business processes with KMDL -
Comprehensive insights into theory and practice, Gito,
Berlin 2012.
Nonaka, I., Takeuchi, H.: The Knowledge-Creating
Company: How Japanese Companies create the
Dynamics of Innovation, Oxford University Press,
New York 1995.
Pogorzelska, B.: Arbeitsbericht - KMDL® v2.2: Eine
semiformale Beschreibungssprache zur Modellierung
von Wissenskonversionen, 07. 01. 2009.
Remus, U.: Prozeßorientiertes Wissensmanagement:
Konzepte und Modellierung. Dissertation, Universität
Regensburg, 2002.
Rychkova, I., Nurcan, S.: Towards Adaptability and
Control for Knowledge-Intensive Business Processes:
Declarative Configurable Process Specifications. In:
Proceedings of the 44th Hawaii International
Conference on System Sciences, 2011.
Scheer, A.-W.: ARIS - vom Geschäftsprozeß zum
Anwendungssystem, 3. ed., Springer, Berlin 1998.
RDBPM 2015 - Special Session on Research and Development on Business Process Management
332