ENTERPRISE ONTOLOGY MANAGEMENT
An Approach based on Information Architecture
Leonardo Azevedo, Sean Siqueira, Fernanda Araujo Baião
Jairo Souza, Mauro Lopes, Claudia Cappelli and Flavia Maria Santoro
NP2Tec – Research and Practice Group in Information Technology
Federal University of the State of Rio de Janeiro (UNIRIO), Rio de Janeiro, Brazil
Keywords: Ontology management, Knowledge management, Information architecture.
Abstract: Ontologies have gained popularity, but its promises of being a key point to the solution of real-world
problems and mitigating interoperability problems at a large scale have not yet been accomplished.
Ontology management is at the kernel of this evolution, and there is a lack of adequate strategies and
mechanisms for handling it in such a way to contribute to a better alignment between business and IT. This
work proposes an approach for enterprise ontology management as part of an Information Architecture
initiative. This approach provides a more complete foundation of the ontology lifecycle while guiding the
enterprise in this management, by defining a set of processes, roles and competencies required for ontology
management. It was applied to a big enterprise in Brazil at the Data Integration department.
1 INTRODUCTION
Gruber (2008) defined ontology as “a set of
representational primitives with which to model a
domain of knowledge or discourse”. There are many
other definitions in the literature, but all these
definitions converge to consider an ontology as a
common representation of a domain of discourse.
Once built, ontologies can be used in several areas
such as (Damjanovic et al., 2004; McGuinness,
2005): data integration; business process modeling;
database design; information retrieval and
extraction; knowledge management; qualitative
modeling; language engineering; e-commerce; and
configuration support. Many of these scenarios pose
challenges and opportunities for any enterprise.
However, building an ontology is only a part of
the way when pursuing to effectively use it within a
corporate environment. There is a need to envision
ontology construction and maintenance as part of an
initiative towards information governance, which
includes not only technical but also management
activities. Ontology management is the task of
producing and maintaining consistency between
formal and real-world semantics (Fensel, 2008).
This work extends this definition including a set of
processes to achieve this consistency.
The proposed set of processes is based on an
Information Architecture viewpoint. Enterprise
architecture is a set of principles, methods and
models that help bridge the communication gap
between IT architects and stakeholders (Lankhorst,
2005). In this context, ontologies may be considered
as conceptual data models that compose the set of
artifacts to build the enterprise information view.
This paper is divided as follows. Section 2
presents an overview of ontology management.
Section 3 presents in details our proposal, and finally
section 4 concludes this work.
2 OVERVIEW OF ONTOLOGY
MANAGEMENT
Corcho et al. (2003) explain that in order to build a
new ontology, several basic questions are related to
the methodologies, tools and languages to be used in
the development process. In the survey conducted to
base our approach, a set of disciplines that are
important for ontology management were studied,
such as: methodologies for constructing ontologies,
representation languages, query languages, ontology
reuse/compatibility, assessment of ontology quality,
ontology evolution, and supporting tools. The
studied authors do not, however, present a set of
activities, roles and competencies in order to apply
these concepts in an enterprise environment. This
243
Azevedo L., Siqueira S., Araujo Baião F., Souza J., Lopes M., Cappelli C. and Maria Santoro F. (2009).
ENTERPRISE ONTOLOGY MANAGEMENT - An Approach based on Information Architecture.
In Proceedings of the 11th International Conference on Enterprise Information Systems - Information Systems Analysis and Specification, pages
243-248
DOI: 10.5220/0002009802430248
Copyright
c
 SciTePress

section complements the concepts presented in
(Corcho et al., 2003).
There are several methodologies for ontology
construction in the literature and they can be
compared regarding several different characteristics,
such as creation strategy, application dependence,
and existence of validation and evolution phases.
In order to build the ontology it is necessary to
abstract reality according to some conceptualization.
When represented as a concrete artifact, a model can
support communication, learning and analysis of
relevant aspects of the domain. To make it possible,
an ontology must be expressed in a language.
Diverse representation languages of real world
conceptualizations exist, with distinct purposes (e.g.,
LINGO, ER, OWL, OntoUML). The OWL is by far
the most adopted language for ontology
representation, since it is a W3C standard.
In addition to representation languages it is also
required to consider the query/retrieval mechanisms.
Query languages in ontologies are fairly new and
different proposals can be found in the literature.
They can be based on database query languages (i.e.,
SQL, OQL), based on rule languages (i.e., Prolog,
Lisp) and based on path expressions (specially those
languages for querying XML documents such as
XPath and XQuery). Examples of query languages
are: nRQL, OWL-QL and SPARQL.
One application of ontologies is for semantics
interoperability between different information
sources. Ontologies compatibilization allows reuse
of knowledge structure in different applications. The
main mechanisms are: ontology matching (Noy and
Musen, 2001); ontology merging and alignment
(Noy and Musen, 1999); ontology integration (Pinto
et al., 1999); ontology mapping (Noy and Musen,
2003); and others can be found in (Klein, 2001).
Brank et al. (2005) highlights that ontology
analysis is a process of value an ontology according
to some criteria, in order to define which one of
many ontologies are most appropriate for a purpose.
There are many proposals for ontology
validation, and they can be divided in approaches for
lexical validation (Maedche and Staab, 2002),
objective validation (Gangemi et al., 2006), and for
semantic validation (Guarino and Welty, 2002).
Ontologies changes according to updates,
insertions, deletions, and structure reviews, since a
better understanding of existing concepts or an
ontology scenario change. Besides it is possible to
change ontology formalism or the domain of
representation (Klein and Fesel, 2001). One change
in an ontology item can produce inconsistencies in
other parts of it. The bigger is the ontology, the more
complex is to understand completely the extension
and meaning of changes (Stojanovic et al., 2002).
The implementation of changes must be propagated
to other dependent ontologies and applications.
Ontologies’ instances must be changed to preserve
consistency.
Gómes-Pérez et al. (2002) show that many tools
appeared in this decade for ontology development
(construction, annotation, integration etc) and for
ontology use in applications. A detailed list of tools
is presented by Gómes-Pérez et al. (2002), Welty
(2004) and Cardoso (2007).
3 THE PROPOSED APPROACH
Spewak and Hill (1992) proposed a set of processes
of an information technology architecture (ITA):
Building the Current and the Future Architectures,
Maintaining Current and Future Architectures,
Defining Policies and Standards, Prospecting
Technology, Participating on Committees,
Evaluating the IT Quality, and Monitoring and
Measuring Activities.
Considering ontologies as conceptual models,
thus playing an important role within ITA, we claim
that those set of processes may be specialized in
order to drive enterprise ontology management
(Figure 1). All these processes were detailed in
macro-processes, Event-Driven Process Chains
(EPC) were modelled as well as function trees
(Sharp and McDermott, 2001).
Figure 1: Manage the ontology environment macro-
processes.
The first were used to detail critical event flows
to the ontology management in those processes in
which the activity sequence is known. The function
Manage the ontology
environment
Build the ontology
environment
Maintain the ontology
environment
Define policies and
standards of the ontology
environment
Prospect artifacts to the
ontology environment
Monitor Manage
activities in the ontology
environment
Participate on
committees
ICEIS 2009 - International Conference on Enterprise Information Systems
244

trees were used to present a set of activities that are
necessary to the processes to which the execution
sequence is not yet known, at least not exactly. In
some cases, the execution sequence of activities is
not critical to the purpose of providing an
understanding of the ontology management proposal
while in other cases it is extremely dependent of the
business practices of the organization.
3.1 Build the Ontology Environment
The process “Build the ontology environment” has a
sub-process to define roles, responsibilities and
competences related to the ontology management in
the organization as well as to attribute them to
specific professionals.
Another sub-process is to define the
infrastructure of ontology management environment.
This process is related to the definition of
appropriate tools for manipulation, usage and
storage of global ontology as well as the application
ontologies. Several categories of tools must be
provided: editors, storage and retrieval servers for
the knowledge structures, backup server, ontology
visualization, etc. This environment has also to
provide information security and allow the definition
of different access levels to the users, be scalable
and integrated to other tools of the enterprise.
After the environment infrastructure is
operational, the global (or domain) ontology is
created. Competence questions are formulated in
order to understand and determine the ontology
scope and objective. It is then validated according to
syntax and semantics. If necessary, adjustments are
applied. After the validation activities, ontology
concepts are mapped to the integrated data model.
Therefore, ontology items are associated to tables
and attributes. If database concepts are not fully
covered by the ontology, new items are added to it.
Other sub-process is related to the use of the
ontology. The ontology must be published,
highlighting benefits and possible uses. Access
mechanisms and strategic uses must be defined.
Training, specially the ontology maintenance and
use is also very important to guarantee the success.
3.2 Maintain the Ontology
Environment
It is important to maintain the ontology environment
so that ontologies have the concepts that are used in
the enterprise, especially those presented in the
integrated database.
Infrastructure Maintenance. In order to maintain
the ontology management environment it is
necessary to update tools, manage users for each tool
and solve problems in the computational
environment. Some required functions are: manage
tools licenses, manage access permissions, create
and publish tools’ manuals, guarantee tools
integration, update the environment whenever
necessary, make contracts with the suppliers and
partners, provide support to the environment etc.
Global Ontology Maintenance. Ontology changes
can be required when a new business process is
modelled, when domain concepts changes or when
the enterprise knowledge changes.
Whenever a change is required, it is semantically
evaluated and if applicable, the change is
represented. After that, the change is implemented
and the ontology is validated, capturing the changing
meaning (conceptualisation) and verifying possible
inconsistencies in other items. Finally, the new
version of the ontology is propagated to the
ontologies and dependent systems.
Global Ontology Extension. Global ontology can
be extended to cover other knowledge areas or
domains. The necessary information can come from
different sources such as business process models,
data models, domain specialists and documents.
Processes can provide good information about
ontology items that have not been considered in the
ontology or even is not really correct or complete. In
addition, mapping process models and ontologies
provides a better understanding of the processes and
activities (Cappeli et al., 2007).
The data model is also a good source of
information because some ontology items are
captured from tables (or entities), relationships,
attributes and restrictions. A document detailing the
process of building ontology from data models was
also developed: “Good practices to build an
ontology from logical or physical data models”.
Domain specialists can be interviewed and help
at capturing ontology items and are good sources of
information as well as documents about the domain.
Some examples of these documents are manuals,
data dictionaries, training documents,
conceptualization documents etc. It is also possible
to use natural language processing to automatically
capture ontology items, but manual processes were
defined to the enterprise at this moment.
For the ontology items that are to be added or
modified to the global ontology it is necessary to
follow the changing processes.
ENTERPRISE ONTOLOGY MANAGEMENT - An Approach based on Information Architecture
245

Application Ontology Creation and Maintenance.
In certain cases it is important to have application
ontologies that extend the global ontology concepts
and have concepts that are specific to a certain
application. If a need for such kind of ontology is
identified, then the application scope is observed and
competence questions must be formulated. Existing
ontologies in the same domain are analyzed to
search for reusable items.
Then, the application ontology is planned; risks
are identified and documented as well as the effort
for the development. An execution chronogram is
created. After that, the ontology is modeled. The
necessary concepts that are in the global ontology
are identified as well as those present in other
existing ontologies. These concepts that have
already been defined in other ontology are mapped
while new concepts and items are codified.
Selecting concepts from the global ontology
allows reusing these concepts that inherit
characteristics from a class of the global ontology,
for instance. Therefore it is possible to reduce
development time and to allow disseminating
concepts through the enterprise. The reuse is also
applicable when identifying existing ontologies
about the same domain. In some cases, the existing
concepts must be adapted in order to keep the
ontology consistence.
Representing concepts that do not exist in the
global ontology can be through a graphical tool for
representing concepts or in a textual way, indicating
its name and other characteristics. For correctly
identifying concepts, supporting documents can be
used such as “Good practices to build an ontology
from logical or physical data models” and
“Questionnaire for gathering ontological items”,
which were developed in this project. The
relationships between concepts are also identified,
presenting the concepts that have to be modelled and
the dependence of concepts from the global
ontology. The codification of the concepts and their
relationships in OWL must follow description
standards. Then the ontology is validated and the
adaptations are executed.
3.3 Define Policies and Standards
This process is responsible for creation,
maintenance, publishing and auditing policies and
standards of the ontology environment.
Creating the Policies and Standards. This process
starts with a need for standardizing some task or
item of the ontology environment. Then the item or
task is analyzed and the policies and standards are
defined and approved.
Raising the item characteristics looks for
improving the ontology quality, getting rules such as
coverage (Which areas do the ontology items cover?
the whole department or a part of the department,
the whole enterprise or a department? How can this
concept be generic to cover all the knowledge areas
of the enterprise?), correctness (How correct is this
item? Is it applicable to all cases? Is there an
exception? How to validate all the aspects?),
richness (Is it complete? Can it be more detailed? Is
it worthwhile to detail? Are there other
characteristics, properties or relationships to be
considered?), commitment (Did all the possible
areas that are going to use this item validate it
adequately? Who was the responsible for the
validation?), organization and modularity, reality
correspondence, clarity, goal consistence, capture of
the domain invariant structure, description form,
syntax correctness, etc.
Then the standards are prospected in the market
and if a proposal is found, it is discussed,
adaptations are investigated and executed. After that,
the standard is validated and approved.
Maintaining the Policies and Standards. This
process starts with a need for maintenance, which
can be a new standard in the market or new aspects
to be considered in the organization standard. A
group is created for standard revision and if changes
are identified, they are analyzed as well as the
impact of the change. If approved, the change is
implemented. The changes are then validated.
Publishing the Policies and Standards. This
process starts when a standard is created or changed.
The enterprise communication mechanisms are used
to publish the policies and standards. Training
sessions are provided.
Controlling Policies and Standards. This process
starts with a need for controlling the correct use of
these policies and standards. A sample of projects is
selected and it is verified the use of the policies and
standards. The result of the auditing tasks is
published as well as a list of needs for changes in the
standard.
3.4 Prospect Artifacts
This process is responsible for prospecting
technologies for the ontology management
environment. It is a continuous process. First,
information about tools for ontology management is
ICEIS 2009 - International Conference on Enterprise Information Systems
246

searched through the Internet, the participation on
technical and scientific events and consulting.
Then, criteria for evaluating the tools are defined
as well as the candidate tools. Criteria should cover
mandatory and desirable characteristics for the type
of tool in evaluation. For ontology management,
they can be related to ontology edition (Is it possible
to change concepts in a graphical mode? Is it
possible to save the ontology in OWL?), ontology
versioning (Is it possible to save more than one
version of the ontology? Does it allow relating
ontology versions?), etc. There are also essential
criteria to the organization such as existence of
technical support in the country, training facilities to
the team, etc. Usually these business requirements
are determinant to tool acquisition.
For each criterion it is given a weigh. Each tool
is evaluated Then the result is analyzed and the
recommendation is generated. Finally, the tool is
chosen, directives for integrating technologies are
defined and the evaluation result is published.
In addition to prospecting tools it is also
desirable to prospect domain concepts. Prospecting
can be through searches on the standardization
organisms as well as institutions with similar
domains, governmental organizations, etc. or
through participation in inter-enterprise committees.
The concepts are then evaluated according to
existing concepts of the global ontology. The
evaluation result is published, presenting a
comparative report of the concepts as well as the
recommendation of the evaluation team. This
recommendation can generate a change requirement
for the global ontology.
3.5 Monitor Management Activities
This process is responsible for monitoring
management activities in the ontology environment
through quality indicators as well as communication
of results. It is necessary to establish and monitor
indicators to verify if (and how) internal activities of
the area are executed and are according to what was
expected. The results should be published.
3.6 Participate on Committees
This process is responsible for defining participation
mechanisms of the management group of the
ontology environment in institutional committees in
the organization. Through these committees it is
possible to establish relationships with other groups
and areas in the organization.
Some possible functions of such participation
can be: to analyze the contribution of using ontology
inferences; support in systems modelling; support in
data schema validation; share concepts in process
models; create mappings between process models
and ontologies; know strategies for making
knowledge explicit; know business process meta-
model; know infrastructure standards for ontology
management; know information exchange standards;
define systems that can use existing ontologies;
discuss infrastructure requirements for ontology
management; establish consensus in concept
definition; identify data that can be used to refine
concept definition; identify improvements in data
quality; identify improvement in analysis tools;
identify data integration opportunities; identify
common processes with the data management;
influence strategies for making knowledge explicit;
influence standards for information interchange;
identify improvements in the ETL process for the
informational database; interact with data
administration area; interact with informational area;
interact with transactional area; interact with
software development area; interact with knowledge
management area; interact with support and
infrastructure area; interact with concepts
standardization groups; interact with process offices.
4 CONCLUSIONS
This work proposed a set of processes for ontology
management in enterprises. The rationale for the
processes definition was based on an Information
Architecture viewpoint, where the ontologies play
the role of conceptual data models. The proposed set
of processes include: build the ontology
environment (plan, organize, define roles and
responsibilities, define and deploy infra-structure
and construct global ontology); maintain the
ontology environment (make changes on existing
ontology and extend it by new concepts, update
infra-structure, construct application ontologies,
etc); define policies and standards (define new rules
for guaranteeing ontology quality, correctness,
commitment, ontology descriptions etc); prospect
artifacts (prospect new tools and new domains);
monitor and manage activities (define metrics to
measure results and execute evaluation activity, and
analyze if the expectations are accomplished); and
participate in committees (establish relationships
with other groups and areas in the organization in
order to grow the contribution of ontology).
These proposed processes were presented and
evaluated by consultants on Ontology Management
and researchers, as well as enterprise professionals.
They were a result of a project at the data
ENTERPRISE ONTOLOGY MANAGEMENT - An Approach based on Information Architecture
247

administration department at PETROBRAS, which
is the largest and most important oil and gas
company in Brazil. The data administration
department at PETROBRAS is responsible for data
integration in the domain of oil and gas exploration
and production.
According to the three traditional abstraction
levels of database design (Elmasri and Navathe,
2005), the administration department uses ontologies
to represent the first level of abstraction (conceptual
level). Ontology is also used to help data integration
of concepts belonging to different areas. Besides, the
use of some of the proposed processes has
demonstrated good results. Different departments
are developing ontologies, which have been
integrated following the proposed activities, and
they are used to support communication, to learn and
to analyze relevant aspects of the company domains.
ACKNOWLEDGEMENTS
The authors would like to thank Petrobras, mainly
TIC/TIC-E&P/GDIEP, for supporting this project.
REFERENCES
Brank, J., Grobelnik, M., Mladenic, D., 2005. A survey of
ontology evaluation techniques. In 8th International
mulit-conference Information Society IS, pp. 166–169.
Cappelli, C., Baião, F., Santoro, F., et al., 2007. An
approach for domain ontology construction from
business process models. In Second Workshop on
Ontologies and Metamodeling in Software and Data
Engineering (WOMSDE'07), João Pessoa, Brazil.
Cardoso, J., 2007. The Semantic Web Vision: Where are
we?. IEEE Intelligent Systems, 22 (5), pp. 84-88.
Corcho, O., Gómez-Pérez, A., Guerrero-Rodríguez, D.J.,
et al., 2003. Evaluation experiment of ontology tools’
interoperability with the WebODE ontology
engineering workbench. In ISWC2003 Workshop on
Evaluation of Ontology Tools, USA.
Damjanovic, V., Devedžic, V., Djuric, D., et al., 2004.
Framework for Analyzing Ontology Development
Tools. AIS SIGSEMIS Bulletin, 1 (3), pp. 43-47.
Elmasri, R., Navathe, S., 2005. Fundamentals of Database
Systems, Addison-Wesley.
Fensel, D., 2008. Foreword. In: Hepp, M.; De Leenheer,
P.; de Moor, A.; Sure, Y. (Eds.). Ontology
Management: Semantic Web, Semantic Web Services,
and Business Applications. p. 295.
Gangemi, A., Catenacci, C., Ciaramita, M., et al., 2006.
Modelling Ontology Evaluation and Validation, In 3rd
European Semantic Web Conference (ESWC2006).
Gómes-Pérez, A., Fernández-López, M., Fensel, D., 2002.
Deliverable 1.3: A survey on ontology tools, Technical
Report IST-2000-29243, OntoWeb - Ontology-based
information exchange for knowledge management and
electronic commerce.
Gruber, T.R., 2008. Ontology. In: Liu, L. and Özsu, M.
(Eds.) Encyclopedia of Database Systems, Springer-
Verlag.
Guarino, N., Welty, C., 2002. Evaluating ontological
decisions with OntoClean. In Communications of the
ACM, 45 (2), pp. 61-65.
Klein, M., 2001. Combining and Relating Ontologies: An
Analysis of Problems and Solutions. In Workshop on
Ontologies and Information Sharing at the 17th
International Joint Conference on Artificial
Intelligence, pp. 53-62, Seattle, USA.
Klein, M., Fensel, D., 2001. Ontology Versioning on the
Semantic Web. In Proceedings of the International
Semantic Web Working Symposium (SWWS), pp. 75-
91, California, USA.
Lankhorst, M., 2005. Enterprise Architecture at Work:
Modelling, Communication, and Analysis, Springer.
McGuinness, D.L., 2005. Ontologies Come of Age. In:
Fensel, D., et al. (eds), Spinning the Semantic Web:
Bringing the World Wide Web to Its Full Potential,
MIT Press.
Noy, N.F., Musen, M.A., 1999. SMART: Automated
Support for Ontology Merging and Alignment. In:
12th Workshop on Knowledge acquisition, modeling
and management, v. 4, pp. 1-20, Banff, Canada.
Noy, N., Musen, M.A., 2001, Anchor-PROMPT: Using
Non-Local Context for Semantic Matching. In 17th
Workshop on Ontologies and Information Sharing at
the International Joint Conference on Artificial
Intelligence, 1, pp. 63-70, Seattle, EUA.
Noy, N., Musen, M.A., 2003. The PROMPT Suite:
Interactive Tools For Ontology Merging And
Mapping, International Journal of Human-Computer
Studies, 59 (6), pp. 983-1024.
Pinto, S.H., Gómez-Pérez, A., Martins, J.P., 1999. Some
Issues on Ontology Integration. In Workshop on
Ontologies and Problem Solving Methods: Lessons
Learned and Future Trends (IJCAI99's), 18, pp. 7-12,
Stockholm, Sweden.
Sharp, A., McDermott, P. 2001. Workflow Modeling:
Tools for Process Improvement and Application
Development, Artech House.
Spewak, S. H., Hill, S. C, 1992. Enterprise Architecture
Planning: Developing a Blueprint for Data,
Applications, and Technology, John Wiley & Sons,
Inc.
Stojanovic, L., Maedche, A., Motik, B., et al., 2002. User-
driven ontology evolution management. In
Proceedings of the 13th European Conference on
Knowledge Engineering and Knowledge Management,
Ontologies ad the Semantic Web, pp. 285-300.
Welty, C., 2004. Ontology Maintenance Support: Text,
Tools, and Theories. In Presentation at the 7th
International Protégé Conference, Bethesda, USA.
ICEIS 2009 - International Conference on Enterprise Information Systems
248