=Paper=
{{Paper
|id=Vol-440/paper-14
|storemode=property
|title=Model Driven Ontology: A New Methodology for Ontology Development
|pdfUrl=https://ceur-ws.org/Vol-440/paper14.pdf
|volume=Vol-440
|dblpUrl=https://dblp.org/rec/conf/oic/KeshkC08
}}
==Model Driven Ontology: A New Methodology for Ontology Development==
https://ceur-ws.org/Vol-440/paper14.pdf
Model Driven Ontology:
A New Methodology for Ontology Development
Mohamed Keshk
Sally Chambless
Raytheon Company
Largo, Florida
Mohamed.Keshk@raytheon.com
Sally.Chambless@raytheon.com
Abstract Model Driven Ontology uses a UML object model artifact
Semantic technology is becoming a preferable alternative as a starting point to build an ontology data model. This
for enterprise-wide applications intertwined with model as a common ground for all team members is then
interoperable information sharing, due to the distributed systematically transformed to a formal ontology,
nature of this technology. Ontology is the cornerstone of facilitating the development of enterprise-wide
semantic technology; therefore, a major challenge for the information exchange and sharing, which can be
project team is to build a complete and consistent uniformly developed, centrally maintained, and efficiently
ontology data model that represents the correct business reused [6]. This would lead to more efficient and
domain. Effective collaboration among customer and team inexpensive information sharing between different
members is essential for the creation of the correct information systems, cost effective development and
ontology model. Equally necessary is a mechanism to deployment of information systems, and better quality
automatically transform this model into ontology script. decision making as a result of more timely, accurate, and
complete information.
Within today’s leading organizations using semantic
technology, a significant factor in business success rests Introduction
solely in the hands of the ontologists. It is they alone who
The development of large-scale enterprise applications has
are responsible for building the correct ontology data
become increasingly complex due to the massive growth
model. Having no other members on the project team
of enterprise data and the constant changing of
capable of verifying and validating the created ontologies
requirements. Semantic technology has been seen as a
may put the entire business at risk.
crucial alternative for managing this complexity by
providing a solid and flexible infrastructure for
This paper describes a new methodology, “Model Driven
information exchange, retrieval, sharing, and discovery.
Ontology,” in which using a standard modeling activity as
a key process for building ontology would effectively and
As ontologies play a central role in facilitating semantic
efficiently enhance collaborations between different
technology solutions, it is essential for business to
parties of the project team. This would lead to a consistent
standardize the ways ontologies are developed. The phases
ontology model validated and approved by all members of
of ontology development include analysis, design, coding,
the project team (business experts, intel analysts, DB
validation, execution, and maintenance. Moreover, it is
admins, architects, ontologists, etc.).
vital for businesses to keep all key players (business
1
experts, intel analysts, architects, DB admins, ontologists, In general, there is no one correct methodology for
etc.) closely involved in the development phases. developing ontologies, since there is no one correct way to
Organizations using semantic technology, including those model a domain [2]. Ontology itself is a data model based
in both governmental and private sectors, frequently hand on formal logic and greatly overlaps with a UML object
ontology development tasks solely to the ontologists. In model, as both share many basic concepts. While a UML
most cases, the consequence is a dilemma, since no other object model has the concepts of classes, properties,
team member is capable of validating the ontology script associations, constraints, and instances, ontology has the
created by the ontologists, and the business might be at same concepts named classes, datatype properties, object
great risk if the script does not reflect the correct business properties, restrictions, and individuals, respectively.
model. Therefore, a methodology to standardize the way Providing a single data model for all parties of project
ontology is developed is badly needed. team will increasingly eliminate design ambiguity, reduce
the complexity of the enterprise data model, and speed up
This paper sheds light on the value of modeling in the the overall development.
context of ontology development for enterprise
applications. It shows how modeling can be an effective Therefore, a UML object model can be seen as a common
way to manage the complexity of ontology development model for ontologists and software architects, as it
[5], as it fosters better communication by overlooking enhances communication between both camps and brings
implementation details that are not relevant to the overall other parties to the table. It also aligns the effort of
system, and delivers robust design and assessment of building a consistent data model that is accessible and
requirements and architectures. Despite these virtues, usable not only by ontologists, but also by other team
mainstream ontologists have yet to take advantage of members.
modeling in everyday practice [8].
Model Driven Ontology Methodology
Our approach uses a Unified Modeling Language (UML)
object model as the common means for expressing In this section, we will discuss in detail the Model Driven
ontology models. As an industry standard, UML graphic Ontology approach with a simple, yet complete, example
models provide a common ground for team members to [1]. The following diagram (Fig. 1) shows a UML class
better understand the business data models and elevate the diagram of a purchase order example in terms of classes,
level of collaboration. The result is a consistent data attributes, enumerations, and relationships including
model, validated and approved by all team members, inheritance, composition, aggregation and associations
which leads to a more accurate ontology script. with constraints represented as cardinalities.
2
Fig-1
Our approach starts with the UML class diagram which
represents the data model of a particular domain. Once Our transformation platform is EMF which is part of the
the team comes up with the correct UML data model, Model Driven Architecture (MDA) and is the
validated and approved by all parties, we generate a UML implementation of a subset of the MDA in Eclipse
version that is encoded in XMI by exporting the model platform [1]. An EMF model is essentially the class
using a standard UML tool such as RSA. We then apply diagram subset of UML. EMF is originally based on MOF
transformation rules by parsing the XMI file into ontology (Meta Object Facility) by OMG (Object Management
script. Group). EMF uses XMI (XML Metadata Interchange) as
its canonical form of a model definition. EMF has its own
This parsing is done through Eclipse Modeling meta-metamodel called Ecore. Ecore is considered the
Framework (EMF) API provided by UML2 plugin [3], metamodel for UML in addition to some other
which is an EMF-based implementation of the UML 2.x metamodels, such as XSD, WSDL, BPEL, etc. Ecore is
metamodel for the Eclipse platform. The objective of this located at the M3 layer of MDA paradigm and defines all
plugin is to provide a useable implementation of the UML kinds of metamodels located at M2, including UML.
metamodel to support the development of model Ecore, itself, is very similar to EMOF (Essential MOF),
processing tools, a common XMI schema to facilitate but has Eclipse as a runtime environment.
interchange of semantic models, test cases as a means of
validating the specification, and validation rules as a EMF lets you define a data model in one of three formats:
means of defining and enforcing levels of compliance [3]. Java interface, XML schema, or UML class diagram, then
3
allows you to generate the other two formats. The most- experts. The expertise of these teams helps generate the
likely scenario is to start with a UML model and generate correct script corresponding to the data model. For the
the corresponding Java interfaces and XML schema. Our purpose of illustrating the transformation mechanism, we
approach extends this capability by generating RDF/OWL have isolated a subset of the diagram (Fig. 2). The
script from the same UML model. complete generated OWL script is too lengthy to include
in this paper.
Building transformation rules is a joint effort between the
architecture team, the ontology team and business domain
Fig-2
The following is XMI script for “Supplier” class:
4
And its corresponding OWL script is the following: ontologists over ontology development, and save a
significant amount of development effort.
Supplier
References
[1] Frank Budinsky. Dave Steinberg, Ed Merks, Ray
Ellersick, and Timothy J. Grose, "Eclipse Modeling
Framework", Addison-Wesley Professional, August
2003.
[2] Dragan Gasevic, Dragan Djuric, and Vladan Devedzic,
"Model Driven Architecture and Ontology
Development", Springer, 1st edition, July 2006.
[3] Eclipse UML2 project,
http://www.eclipse.org/modeling/mdt/?project=uml2
[4] D. S. Frankel, Model Driven architecture: Applying
MDA to Enterprise Computing, OMG Press, ISBN:
0471319201, January 2003.
[5] Nešić, S., Jazayeri, M., Jovanović, J., Gašević, D.,
"Ontology-based content model for scalable content
reuse", In Proceedings of the 4th ACM International
Conference on Knowledge Capture, Whistler, BC,
Canada, 2007, pp. 195-196.
Conclusion [6] "National Information Exchange Model - NIEM",
http://www.niem.gov/
This paper explains the benefits and values that
modeling practice can offer for ontology-based [7] dos Santos, E.S., Ralha, C.G., Carvalho, H.S., Gašević,
applications, by treating modeling as a first class artifact, D., "MDA-based Ontology Development: A Study
rather than documentation. In addition to presenting a Case," In Proceedings of the 19th International
single common data model that all team members can Conference on Software Engineering and Knowledge
share, a mechanism is presented to generate the ontology Engineering, Boston, USA, 2007.
script directly from the UML model once it is validated
and proofed. In this case, the model is used not only as a [8] Stephen Cranefield, Jin Pan, "Bridging the Gap Between
diagram or a blueprint, but also as a primary artifact the Model-Driven Architecture and Ontology
from which efficient script is generated by applying Engineering", The Information Science Discussion Paper
transformation rules. Series, Number 2005/12, December 2005, ISSN 1172-
6024.
We argue that the use of Model Driven Ontology would
increasingly boost productivity, eliminate mistakes due
to human misunderstanding, break the monopoly of
5