=Paper=
{{Paper
|id=Vol-514/paper-7
|storemode=property
|title=Towards Collaborative Strategy Content Management using Ontologies
|pdfUrl=https://ceur-ws.org/Vol-514/paper7.pdf
|volume=Vol-514
}}
==Towards Collaborative Strategy Content Management using Ontologies==
https://ceur-ws.org/Vol-514/paper7.pdf
Towards collaborative Strategy Content
Management using Ontologies
Simon Paradies, Sonja Zillner, and Michal Skubacz
Siemens AG
Corporate Technology CT IC 1
Otto-Hahn-Ring 6
81739 München, Germany
http://www.ct.siemens.com
{simon.paradies.ext,sonja.zillner,michal.skubacz}@siemens.com
Abstract. We propose the use of ontologies for managing strategy con-
tent, which is generated and processed during strategy planning pro-
cesses. We introduce a Protégé extension aiming at facilitating this task.
It provides views on the strategy content codified in the ontology in form
of diagrams and supports the concurrent graphical editing of these by
multiple users at the same time.
Key words: Strategic Management, Strategy, Ontology, Protégé, Plu-
gin, Collaboration
1 Introduction
“To be, or not to be: that is the question” [1]. This question not only is funda-
mental to Hamlet but also to the majority of organizations that struggle for their
justification to exist (in most cases the increase of the shareholder value) or the
achievement of other (subsequent) goals. In the best case, decisions and actions
resulting thereof are coordinated accordingly in the best possible manner. Such
coordination is often called a Strategy.
The field of strategic management research is among other things dealing with
the question how strategies emerge, are planned, formulated, and implemented.
Strategic management research is a heterogeneous field; less reductionist but
constructivist; meaning that theories are built on individual experiences rather
than built on the description of causal systems. It employs theories and methods
inspired from economics, and others coming from an eclectic set of behavioral
sciences like psychology, social psychology, and sociology [2]. The practical dis-
cipline of strategic management reflects this heterogeneity, as it makes use of a
multitude of different methodologies to (amongst other things) generate strategy
content (which includes strategies).
The contribution of the paper is to analyze how ontology structures and
collaborative tools can be used for strategy content management. We will very
CEUR
ceur-ws.org
Workshop ISSN 1613-0073
Proceedings
briefly introduce the characteristics of strategy content in Sect. 2, how it is com-
monly managed, and weaknesses of those approaches. In Sect. 3, the usage of
ontologies and collaborative tools to structure, integrate, and manage strategy
content is proposed. Since we have special requirements managing it, we devel-
oped a plugin for the Protégé ontology editor, which is introduced in Sect. 4.
Section 5 describes the application of the plugin prototype in an example use
case scenario, thereby showing its functionality. In Sect. 6, we introduce related
work. Section 7 discusses the technical aspects of the plugin and possible im-
provements. Finally, Sect. 8 draws some final conclusions on the approach and
shows ways for further research.
2 Strategy content management
Strategy content1 makes up the strategy knowledge body of an organization. It
describes how the organization means to act and at best even motivates why. In
most cases, it is developed collaboratively by a team of experts with different
backgrounds, particularly to allow for a greater variety of perspectives on the
domain under observation or to establish coordination between distinct units
of the organization. These different backgrounds at the same time may impede
efficient collaboration because each expert - for efficient mental processing and
filing - encodes information in his own mental schema, which may cause prob-
lems if other experts are unable to follow. Yet, the persons involved in strategy
planning processes need a common understanding about the matter they are
dealing with and means to effectively collaborate.
Overall, different sources of information are analyzed and brought into con-
text. Examples include market and competitor data derived from business and
competitive intelligence, appraisals of business development based on gut feel-
ings of executives, or objectives of the super ordinate department. The resulting
strategy content is complex in its characteristics. It encompasses multiple aspects
and multiple domains, for instance markets and customers, as well as technol-
ogy and trends, people and competences. It is used to describe the current and
the desired state of an organization in the real world and the actions it plans
to implement this transition. For capturing these different aspects, practition-
ers and scientists over times developed a multitude of specialized methods and
methodologies that contribute to the “crafting” of a strategy. Examples include
the Balanced Scorecard [4], Blue Ocean [5], Porter’s Five Forces [6], and many
more. Each employs different views on an organization, thus being capable of
capturing a particular part of the whole “strategy picture” and generating spe-
cific strategy content. However, there is no standard way to consolidate this
specific strategy content into an integrated strategy picture.
1
Cf. [3] for a discussion about the distinction between strategy process, content, and
context
2.1 Common strategy content management
Organizing strategy content is not an easy task, as there are basically two con-
flicting concerns. On the one hand, the flexibility to support a multitude of
methods with different underlying concepts, and on the other hand, the need to
structure and integrate the generated strategy content to be able to relate the
knowledge, provide means for common understanding, foster reuse, communi-
cation and collaboration, and ideally to provide means for automation (i.e. an
enhanced IT-support). By means of a central storage, furthermore the data man-
agement, revision, versioning, information access and exchange is made consider-
ably easier. In typical scenarios in organizations however, the strategy knowledge
that is generated often is fragmented and stored in multiple repositories, includ-
ing wordprocessing and spreadsheet programs, or maybe special databases at
least. Using a fragmented, decentralised storage concept, unfavorable side-effects
occur:
– Problems related to document management in general in a larger and/or
collaborative environment apply, e.g.:
• Synchronization problems: Multiple versions of a document or sets of
documents that are mutually not consistent.
• Versioning complexity
• Collaboration obstacles: Concurrent document editing is not possible.
– Knowledge integration and linking barriers:
• The inter-relation of concepts is complicated (concepts need to be copied
back and forth between or within the documents).
• There is no mechanism to provide an overview about selected concerns
and to easily navigate or drill down to the details.
– Little guidance for structuring strategy content (except plain forms or ta-
bles):
• The strategy modeler is not supported in structuring his stream of
thought into separate and meaningful concerns and needs to figure out
an conceptual scheme himself.
• Reuse is impeded: A second modeler likely is unfamiliar with the con-
ceptual scheme of the first modeler.
• Secondary information is left out since the modeler takes it for granted.
• A common understanding is difficult (implications, causal chains, etc.).
– Strategy content is barely amenable to automation, e.g. a faceted search or
automated comparison.
3 Ontology-based knowledge structures to manage
strategy content
To overcome these limitations, we looked for new ways to organize strategy
content and to improve the strategy content modeling process. As ontologies
can be used to build formal models of reality and the ontology knowledge model
supports the storage of unstructured knowledge and, at the same time, provides
means to structure it using taxonomies and restrictions, the use of ontologies
was considered promising. As ontology representation language we decided to
use OWL due to its role as a building block for the semantic web, its large
user community, and good tooling support. For ontology engineering Protégé-
OWL [7] was chosen, primarily because of its free availability, active and vivid
community, extensibility, and an already large pool of extensions for diverse
needs. One of these extensions, the Collaboration plugin [8], was especially useful
for our purposes.
3.1 Towards strategy knowledge models
A Description logic can be decomposed into TBox (taxonomy and additional
predicates) and ABox (instances). We decided to use the TBox as a guiding
model to structure the strategy content stored in the ABox and call it strat-
egy knowledge model (or SKM). Admittedly, a general overarching model to
model a detailed strategy plan is not reasonable in our opinion due to the multi-
tude of methods and contexts in strategy planning and the varying granularity
needed at different organizational units. In the future, we seek to provide a
meta-methodology or strategy knowledge-engineering methodology (SKEM) re-
spectively, which is able to elicit the requirements on a SKM that is usable at
a specific step in the strategy planning process (i.e. supporting the gathering
of strategy content generated by a specific strategy management tool such as a
value chain analysis).
3.2 Strategy process support
The actual strategy content is then provided by the users/strategists in terms
of ABox instances. To provide an improved interface to the user (additional
visualization and editing capabilities) which enables them to efficiently and col-
lectively manage the strategy content they generate during a specific planning
process step and at the same time help ensuring the adherence to the SKM
specific to the process step, we developed a Protégé extension (i.e. a slot-widget
plugin prototype) called OWL-Diagramming.
4 The OWL-Diagramming plugin
The OWL-Diagramming plugin builds on top of the Collaboration plugin and
allows for a visual representation and manipulation of ontology instances and
their relations. It is inspired by common modeling tools used for instance in
the domain of Business Process Management for workflow modeling and by the
graph-widget plugin that ships with Protégé. Its most important properties are:
– It provides user definable views on selected parts (concepts) of the ontology.
– It can be used to instantiate instances, delete, and relate them by simple
drag and drop (“d’n’d”) and point and click operations.
– It can be used in conjunction with Protégé server in collaborative mode.
Changes made to one model are reflected in others as soon as the server
propagates these changes.
– It makes context specific suggestions and enforces modeling restrictions ac-
cording to the underlying knowledge model (TBox). This is achieved by
programmatically interpreting cardinality restrictions to enforce cardinality
constraints and by interpreting the range and domain of a property as nor-
mative. This is in contrast to the common purpose of restrictions, property
domain, and property range in OWL for inference and is discussed in Section
7.
– The underlying graphical framework (JGraph X) [9] is available with sources
under a LGPL license. It will probably be finally released as JGraph 6 under
a BSD style license.
To show how strategy content management could be improved by the use of
ontologies and appropriate tooling, and to present the plugin, we will introduce
a use case scenario meeting a typical setting in global companies.
5 A strategy planning use case scenario
We regard a small, shared unit of a global company. This unit does consulting
in the area of knowledge management solutions as an internal technology service
and solution provider. Its two branches, besides its headquarters in Europe, are
located in the U.S. and Asia.
5.1 An example strategy planning process
To maintain and expand its strategic position, it conducts a lightweight, qualita-
tive strategy planning process on a regular basis involving the department head
and the two branch managers, which we call strategists in the following. They
need to accomplish the following tasks:
– Describe and monitor the current strategic environment, i.e. putting issues
on the strategic agenda that become apparent and are considered poten-
tially strategically relevant at present or in the future. Detailed information
sources motivating the consideration of those issues and providing additional
information should be referenced in all cases.
– (Re-)assessing issues that have manifested or changed since the last strategy
planning. Each issue shall be subjected to an analysis to find out how it has
to be judged, using a classical SWOT (cf. [10]) analysis.
– Setting goals for the next fiscal year, considering and referencing the findings.
– Framing ways for goal achievement, including proposals for concrete actions.
5.2 A strategy knowledge model
From the description of the strategy process described above, we derived the
following concepts and properties (uppercase) to structure and constitute the
strategy content in a SKM. It exemplifies a possible outcome of the SKEM (cf.
Sec. 3.1) that we eventually strive for:
– Influencer: The Influencer2 concept is used to describe issues of strategic
relevance. It has at least one Information Source property for further expla-
nations, which also used to motivates its incorporation in and relevance for
the strategy planning.
– Strength, Weakness, Opportunity, and Threat: These concepts depict judg-
ments resulting from an Influencer’s SWOT analysis.
– Goal: Describes a future state that the unit aims to achieve. It has an op-
tional Objective property which could be used to quantify the Goal. It is
decomposable into at most three sub Goals and is supported by at most
three Strategies. This restriction was made to help ensuring the strategic fo-
cus of the unit, i.e. to focus on few but therefore more important Strategies
for the achievement of the Goal.
– Strategy: Describes a course of action to channel Goal achievement. In the
SKM used, a Strategy supports at least one and at most three Goals. This
constraint was introduced to help enforcing the focus on less but more specific
Strategies. The Strategy’s implementation details can be further specified
using its Implementation Details property. (see Fig. 5 and 4 for examples)
– Resource: Represents a resource (or capability) that the unit may deploy.
All above concepts have a mandatory Description property. Mandatory and car-
dinality restricted properties are realized using OWL-restrictions. An additional
View concept is introduced, which is not used to conceptualize strategy content
but acts as a container for the different views that can be defined (cf. Fig. 1).
5.3 Supporting the strategy planning process using the
OWL-Diagramming plugin
In our use case scenario, a Protégé server session serves as “strategy content
repository” hosting the SKM and strategy content ontology. On the appointed
date and time, the strategists log in from their respective local offices. The unit
head instantiates a new View and labels it “Assessment 2009/Q1”. It shows an
empty canvas and a palette populated by the concepts previously defined in the
SKM. In the initial phase of the planning process, influencers are explicated (cf.
Sect. 5.1). For instance, the “Journal of Knowledge Management” recently pub-
lished an article about the management of unstructured data and concluded that
“the amount of unstructured data will increase significantly”. This information is
important in the context of assessing knowledge management technologies and
2
a concept borrowed from [11] to describe issues that have “the capability to produce
an effect without apparent exertion of tangible force or direct exercise of command”.
Fig. 1. The OWL-Diagramming Plugin showing a recently instantiated Influ-
encer concept
thus introduced in the strategy content model by instantiating the Influencer
concept named “More unstructured data”. This is achieved e.g. by simply drag-
ging and dropping an Influencer node from the palette to the canvas (see Fig. 1).
The description “Amount of unstructured data in project/system business will
increase” is added by double-clicking the newly introduced Influencer instance
and filling out the form in the individual editor that pops up. The information
source that motivates the Influencer (“Journal of Knowledge Management”) is
added as well. The change made to the model of one client is promptly prop-
agated to all clients connected to the server and their canvases are updated
accordingly. Meanwhile, the Collaboration plugin keeps track of changes to the
model (e.g. time and initiator). After introducing the Influencer, a discussion
starts among the strategists (using the chat tab of the Collaboration plugin)
about its potential impact to the current strategic setting of the unit. They con-
clude that this Influencer represents an Opportunity, since the unit possesses ma-
jor expertise with unstructured data. Therefore, an Opportunity concept named
“Exploit expertise with unstructured data” is instantiated and related to the
Influencer by dragging and dropping from the Influencer to the empty canvas.
The OWL-Diagramming plugin checks for valid outgoing relations (object prop-
erties) from the Influencer and which values they could take. Accordingly, it
suggests instantiating a Strength, Weakness, Opportunity, or Threat concept
(see Fig. 2). The strategists choose the Opportunity concept. The “judged as”
relation is inserted automatically, since according to the SKM it is the only
valid (see Fig. 3). Since the strategists forgot to add a Description as demanded
Fig. 2. Only concepts connectable to the In- Fig. 3. The only valid predicate
fluencer can be chosen is selected automatically
by the SKM, the Opportunity instance shows a warning sign (see Fig. 3). The
warning is resolved by adding the description “Our unit could possibly exploit
the trend towards unstructured data“. To make clear that the assessment result
is motivated by the fact that the unit possesses major expertise with unstruc-
tured data, a Resource concept is instantiated and related to the Opportunity
to provide an explicit reference to the unit’s ability (“Major expertise with un-
structured data”). Since this directly leads to a potential Goal, a Goal concept is
instantiated labeled “Exploit and extend expertise with unstructured data”. The
strategists then start brainstorming about the best means to realize the Goal.
Each member of the group puts some Strategies into play, ending up with four
potential Strategies that they relate to the Goal. Each Strategy is specified by
adding concrete implementation details (see Fig. 4). Since the SKM allows for
at most three Strategies supporting a Goal, a warning sign appears at the Goal
indicating a cardinality constraint violation. It is resolved after a discussion and
the removal of the Strategy considered least effective to support the Goal.
The above step is repeated for other Influencers and Goals as well, leading
to a lightweight strategy model which the unit seeks to align its activities in the
next fiscal year with.
Fig. 4. Strategy implementa-
tion details Fig. 5. Highlighting a modeling error
6 Related Work
Concerning the graphical visualization of ontologies (i.e., the relations between
concepts or individuals), there are a lot of tools available for which [12] pro-
vides an overview. Only very few support the interactive editing of the contents
they show, amongst them, the Graph widget-plugin [13] that already ships with
Protégé. To our knowledge, none supports the interpretation of restrictions on
properties and classes to guide the user in the ontology editing process or a
concurrent graphical editing.
For structuring strategy content, there are models available which suggest
concepts to model strategy content (e.g. BMM [11] or TOGAF [14]), strategy
management tools incorporating models (e.g. the BSC’s strategy map [4]), or
ontologies developed for that purpose (cf. the TOVE project [15]).
7 Discussion
The guidance in modeling offered by the plugin by interpreting restrictions as
constraints clearly helps the user to create a standardized and “valid” (regarding
the intentions of the TBox modeler) output. However, the programmatical ap-
proach in interpreting restrictions as constraints could be significantly improved.
It should be replaced by an efficient constraint checking mechanism supporting
the full power of OWL’s expressiveness. Recently, there has been some progress
with respect to incremental reasoning and constraints checking (interpreting
restrictions as constraints) using the pellet reasoner [16,17] which could poten-
tially be very useful in our setting. Using a “probing” pattern, that is, inserting
an instance in the ontology and checking if it violates any constraints or leads
to contradictions, would - in combination - bring us very close to this goal. It
would be even better if a reasoner could provide such functionality out of the
box (without actually having to temporarily modify the ontology).
The graphical representation of the ontology supplied by the plugin helps
non-experienced users populating and editing the ontology’s ABox. Nonetheless,
the use of the plugin still requires at least a basic understanding of ontologies.
Hiding the technical details would further increase its usability and foster its
acceptance by non ontology experts.
The collaboration features provided already ease the concurrent modeling sig-
nificantly. The use of Web 2.0 technology could possibly offer a more lightweight
and convenient approach in a collaborative environment. Since the graphical li-
brary used [9] is also (commercially) available in a JavaScript flavor, the porting
to a web based version for the upcoming WebProtégé [18] is facilitated.
As an additional feature, the integration of external data sources could be
contemplated, for instance, to color a possible Goal (“customer satisfaction”)
according to up-to-date data from an external CRM system.
Finally, the applicability of reasoning techniques beyond constraints checking,
e.g. for inferring new knowledge in strategy content, is an interesting research
question.
8 Conclusion
We showed the use of our prototypical Protégé OWL-Diagramming plugin in a
simplistic strategy planning process supporting ontology based strategy content
management. The plugin is appropriate for other domains as well, merely pro-
viding an easier and more user friendly way to populate ontologies and edit the
relations between its instances graphically. Extending its application spectrum
will be part of future work. The strategy knowledge model (i.e. TBox) used was
well suited to the lightweight strategy planning process example. In the future,
we will investigate means to support more complex settings (strategy processes
and content) and aim at providing a strategy knowledge-engineering methodol-
ogy (SKEM) (cf. Sect. 3.1), able to elicit customized Strategy Knowledge Models
(SKM) appropriate for specific steps in strategy planning processes.
References
1. Shakespeare, W.: The tragedy of Hamlet, Prince of Denmark. (1603)
2. Lubatkin, M.H., Lane, P.J.: A strategic management model of agency relationships
in firm. The Blackwell handbook of strategic management (2001) 229
3. Meyer, R., Wit, B.D.: Strategy Synthesis: Resolving Strategy Paradoxes to Create
Competitive Advantage. 2. rev. edn. Thomson Learning (November 2004)
4. Kaplan, R.S., Norton, D.P.: The balanced scorecard. translating strategy into
action. (1996)
5. Kim, W.C., Mauborgne, R.: Creating new market space. HBR 77(1) (1999) 83–93
6. Porter, M.E.: How competitive forces shape strategy. HBR (1979) 86–93
7. Knublauch, H., Fergerson, R., Noy, N., Musen, M.: The protégé OWL plugin: An
open development environment for semantic web applications. In: The Semantic
Web - ISWC 2004. (2004) 243, 229
8. Tudorache, T., Noy, N.F., Tu, S., Musen, M.A.: Supporting collaborative ontology
development in protege. In: The Semantic Web - ISWC. Volume 7. (2008)
9. JGraph: JGraph X. http://www.jgraph.com/jgraphx.html (2009)
10. Friesner, T.: History of SWOT analysis. http://www.marketingteacher.com/
SWOT/history_of_swot.htm
11. OMG: Business motivation model (BMM), v1.0 (February 2008)
12. Katifori, A., Halatsis, C., Lepouras, G., Vassilakis, C., Giannopoulou, E.: Ontology
visualization methods - a survey. (2007)
13. Stanford Center for Biomedical Informatics Research: Protégé: Graph widget tu-
torial. http://protege.stanford.edu/doc/tutorial/graph_widget/index.html
14. The Open Group: The open group architectural framework version 9 (TOGAF).
http://www.opengroup.org/togaf/ (February 2009)
15. Grüninger, M., Atefi, K., Fox, M.S.: Ontologies to support process integration in
enterprise engineering. Computational & Mathematical Organization Theory 6(4)
(December 2000) 381–394
16. Sirin, E.: Integrity constraints for OWL. http://clarkparsia.com/weblog/2009/
02/11/integrity-constraints-for-owl/ (February 2009)
17. Clark & Parsia LLC: Pellet features. http://clarkparsia.com/pellet/features/
18. Tudorache, T., Vendetti, J., Noy, N.F.: Web-protege: A lightweight owl ontology
editor for the web. In: 5th OWL Experiences and Directions Workshop (OWLED
2008). (2008)