A FRAMEWORK FOR MANAGING MULTIPLE ONTOLOGIES:
THE FUNCTION-ORIENTED PERSPECTIVE
Baowen Xu
1, 2
, Peng Wang
1
, Jianjiang Lu
1, 2, 3
, Dazhou Kang
1
, Yanhui Li
1
1
Department of Computer Science and Engineering, Southeast University, Nanjing 210096, China
2
Jiangsu Institute of Software Quality, Nanjing 210096, China
3
PLA University of Science and Technology, Nanjing, 210007, China
Keywords: Semantic Web, Ontology management, Multiple ontologies
Abstract: Ontologies are now ubiquitous in Semantic Web and knowledge representation areas. Managing multiple
ontologies is a challenging issue including comparing existing ontologies, reusing ontologies, maintaining
different versions, and so on. However, most previous multiple ontologies management work focused on
ontologies maintenance, evolutions, and versioning. They ignored the very important point: exploiting the
functions of multiple ontologies provide. This paper proposed a new framework for managing multiple
ontologies based on the function-oriented perspective, and its goal is to bring multiple ontologies together to
provide more powerful capabilities for the practical applications. The new multiple ontologies management
architecture is more feasible and robust in the dynamic and distributed Semantic Web environment.
1 INTRODUCTION
The Semantic Web (Berners-Lee, 2001) envisions a
world-wide distribute architecture where data and
computational resources will easily interoperate
based on semantic marking up of web resources
using ontologies. Ontology is a formal, explicit
specification of a shared conceptualization (Gruber,
1993), and is also the core of knowledge
representation in the Semantic Web. The Semantic
Web researches have improved the popularity of
ontologies greatly. Although large amount of
endeavors have been done for the ontology building
that produced many methodologies, tools and
criteria for helping the ontology development (Noy,
2001, Corcho, 2003, Ding, 2002), we had to accept
the fact that developing ontologies is a labour
intensive work. Today, researchers have already
accepted such a common viewpoint that due to one
cannot expect a single ontology to describe the vast
amounts of data on the web, We believe the
Semantic Web should be built on many small
ontologies (Roussset, 2004, Mena, 2000). We can
easily obtain various ontologies through many ways,
such as from the web ontology library (DAML
Ontology Library) or someone’s papers. Frequently,
using multiple existing ontologies not only can avoid
or reduce the work for building new ontologies, but
also can describing the wider knowledge and satisfying
the requirements of a varied community of users.
Due to the intrinsic syntactic and semantic
heterogeneities between different ontologies,
managing multiple ontologies will face many
challenges and problems (Noy, Musen, et al., 2004,
Ding, 2001, Wendt, 2002). Obviously, we should
develop the ontology maintenance methods
(Stojanovic, 2003) and tools (Noy, 2004) for
multiple ontologies management. Similar to the
single ontology, multiple ontologies’ evolutions and
versioning could be considered in the management
too. However, these aspects are not the ultimate goal
for managing multiple ontologies, and we believe
the real target of managing multiple ontologies is
exploiting their powerful function for the practical
applications. To bring multiple ontologies together,
several ontology management frameworks have
been proposed (Noy, Musen, et al., 2004, Cui, 2000,
Maedche, 2003, Maedche, 2002, Das, 2001).
However, these work focused on the multiple
ontologies maintenance or ontology evolutions. Few
work discussed the methods of managing multiple
ontologies for providing more powerful ability for
the applications such as semantic querying across
multiple ontologies and extracting reasonable sub-
ontologies from the multiple ontologies according
the users’ requirements.
300
Xu B., Wang P., Lu J., Kang D. and Li Y. (2005).
A FRAMEWORK FOR MANAGING MULTIPLE ONTOLOGIES: THE FUNCTION-ORIENTED PERSPECTIVE.
In Proceedings of the Seventh International Conference on Enterprise Information Systems, pages 300-305
DOI: 10.5220/0002538803000305
Copyright
c
SciTePress
This paper presents a new framework to deal
with the multiple ontologies management, and our
idea is based on the multiple ontologies’ functions.
To provide more powerful ability for the practical
applications, we design the five-layer multiple
ontologies management architecture, and investigate
the problems of normalizing multiple ontologies,
expressing the complex relations between ontologies
using bridge ontology, and how to provide functions
for different practical requirements.
The rest of this paper is organized as follows.
Section 2 argues the reasons of using multiple
ontologies, and proposes the multiple ontologies
management tasks we focus here. Section 3 presents
the framework of managing multiple ontologies and
describes the detail of each layer in the framework.
Section 4 discusses the related work of managing
multiple ontologies. Finally, conclusions are
presented in Section 5.
2 MULTIPLE ONTOLOGIES
MANAGEMENT TASKS
In this section, we will discuss the reason of
managing multiple ontologies and argue the
advantages of using multiple ontologies through
comparing the multiple ontologies with single
ontology. Then we analyze the deficiencies of
former multiple ontologies management work and
provide the tasks we want to realize through
managing multiple ontologies.
2.1 Multiple Ontologies Versus Single
Ontology
Why we want to manage multiple ontologies? What
are the advantages of using them? In order to answer
these questions, we begin with the comparison
between single ontology and multiple ontologies
shown as in Table 1. There are seven criterions
listed in the table. First is the knowledge expressive
range. Except for few large-scale ontologies, most
single ontology can only express limited and specific
domain, but the multiple ontologies can express
broader and even the knowledge of crossing
domains. The reasoning range is similar to the first
criterion that multiple ontologies can reason on the
wider range than single ontology could. As far as the
usability, singe ontology is easy to use, but multiple
ontologies are difficult to use for the reason of the
user should know the relations between ontologies in
advance. As for the acquirability, it is not easy to
find a suitable ontology to meet the requirement of
specific domain because each ontology may have
some shortages for represent the knowledge of
current domain, for example it maybe too ‘big’ for
the needs. So in most time, the users have to build a
new one. But it is easy to find multiple ontologies to
overcome this problem, and we just use the suitable
part in each ontology to combine to satisfy the
requirements. The fifth criterion is about the
heterogeneous problem. Single ontology has not this
kind of problem, but it is a serious problem must be
considered in multiple ontologies. How to deal with
the heterogeneity is the key problem in our multiple
ontologies management. As far as the ontology
engineering, single ontology includes building new
ontology or extending the existing ontologies, and
multiple ontology need ontology mapping or
merging. They both are laborious work and lack of
automatic method to support. The last criterion is
flexibility. It is obvious that multiple ontologies are
more flexible than single ontology, and fit the
distributed and dynamic Semantic Web
environment.
From the comparison discussed above, both the
single ontology and the multiple ones have their
advantages and disadvantages. But for the fact that
more and more small ontologies are popular and the
difficulties of building and maintaining big
ontology, it is necessary to face to employ multiple
ontologies. Therefore, we need some feasible
methods to manage multiple ontologies and avoid
their disadvantages.
Table 1: Multiple ontologies versus single ontology
Criterion Single ontology Multiple ontologies
Expressive range Limited and specific domain Wider and cross domain
Reasoning range Narrow Wider
Usability Easy to use Difficult to use
Acquirability
Difficult to find suitable ontology
for specific domain
Easy to find multiple
ontologies for specific domain
Heterogeneity No Yes
Ontology engineering
Building new ontology; Extending
existing ontology
Ontology mapping/alignment
and merging/integration
Flexibility No Yes
A FRAMEWORK FOR MANAGING MULTIPLE ONTOLOGIES: THE FUNCTION-ORIENTED PERSPECTIVE
301
2.2 The Goals of Managing Multiple
Ontologies
There are two main advantages of using multiple
ontologies. First, using multiple ontologies may
realize the ontology reuse. Secondly, the multiple
ontologies are more robust in the dynamic Semantic
Web environment, which is important to the
Semantic Web applications. To achieve these merits,
we need an efficient management to deal with the
problems of using multiple ontologies to reconcile
multiple ontologies. The motivation of this paper is
finding a flexible and low-cost approach to manage
multiple ontologies to meet the tasks of representing
crossing domain knowledge in the dynamic
Semantic Web. Different from the previous work,
we focus on how to exploit the powerful ability of
multiple ontologies provide, and we call this is
function-oriented perspective. The following is some
goals we want to reach.
Query and retrieve across multiple ontologies
Use the mappings defined between ontologies to
support query to one ontology posed in terms of
another ontology. We should manage the useful and
complex relations between ontologies for the
querying rewrite in these applications.
Reason across multiple ontologies Use the
relationships defined between ontologies to support
inference across several ontologies.
Extract sub-ontology from multiple ontologies
Analyze dependencies and allow users to extract sets
of concepts and relations as a sub-ontology.
Interoperability of shared ontologies Specify
transformation rules between different ontologies
and versions of the same ontology; Align and map
between ontologies; Translate ontologies from one
form to another.
3 MULTIPLE ONTOLOGY
MANAGEMENT FRAMEWORK
In this section, we will present the architecture of
managing multiple ontologies. And then we
investigate each functional component in the
framework.
3.1 Architecture
Several frameworks were proposed to manage
multiple ontologies. In the function, these work
mainly focus on the ontology evolution and
maintenance. In the architecture, they just have two-
layer architecture that is ontology repository and
ontology application. There are some disadvantages
in these frameworks. First, the two-layer architecture
is too simple to manage multiple ontologies, and we
need a more systematic architecture. Second, the
functions provide by the multiple ontologies are
embedded in the practical applications, and that
cause many repeated work.
Our idea of solving these problems is based on
two aspects. Firstly, we separate the inter-
relationships among ontologies from multi-
ontologies with bridge ontology, and try to provide
the unified extraction for relationships among
ontologies. After this step, each ontology is still
independent, but we have collected the various
relationships among multiple ontologies. It is a
Figure 1: The architecture of managing multiple ontologies.
ICEIS 2005 - INFORMATION SYSTEMS ANALYSIS AND SPECIFICATION
302
flexible way for the dynamic web environment.
Secondly, we separate multi-ontologies functions
from applications to provide unified functions for
practical applications, and also provide substitute
methods for unnecessary ontology engineering work
such as alignment and integration.
Based on these ideas, we design a five-layer
architecture for managing multiple ontologies. The
framework is composed of the ontology base,
ontology representation, bridge ontology to express
the relations between ontologies, unified functions
provided by multiple ontologies, and the practical
applications.
3.2 Ontology Base
We can obtain ontologies from different ways. Each
ontology may be built by different people in
different period, so they may be in different model
and ontology language. Some of them are built in
RDF(S), some are built in OWL. These ontologies
are our source ontologies, and we use the ontology
base to store them.
It is necessary to store all available ontologies in
the ontology base. Choosing ontologies is based on
the users’ requirements. For example, if the users are
interested in university and company, we could only
store many the two kinds of ontologies in the
ontology repository.
3.3 Ontology Representation
There are many different Ontology languages on the
Web, such as OWL, DAML+OIL and Ontolingua.
They are different in syntax and structure, and based
on different logic foundations. Besides the different
in syntax, ontology languages may be based on
different logic system, such as Frame Logic,
Description Logic (DL) and N3. Due to different
representation using different logic model and
language, their expressing abilities are different too.
Therefore, the analyzing, extracting and integrating
methods of them are different. Translating
ontologies into a unified internal representation, i.e.
a unified Ontology model, is necessary.
Generally, an ontology can be seen as a
quadruple O = (C, R, X,
), where is the set of
individuals; C is the set of concepts, which are
subsets of
; R is the set of relations, which are
subsets of
× ; X is the set of axioms. This
definition is very broad and general. This definition
is too general and is hard to operate the ontologies in
such representation.
In our framework, the information about
individuals is not concerned. We focus on the
concepts and relations. They are both fundamental
elements in most Ontology languages. It makes the
extraction difficult. We consider concepts to be the
only fundamental element, and organize them into a
hierarchy. Relations are divided into attributes
(datatype properties in OWL) and other relations
(object properties in OWL). Attributes are special
relations that between individuals and literals. Both
attributes and relations are depending on the
concepts that they linked; they are not the
fundamental element in Ontology. (The meaning of
these elements in Ontology can be found in OWL
standards of W3C)
Definition 1. We consider an ontology as a
eight-tuple:
O = (C, A
C
, R, A
R
, S(C), E(C), H, X)
where C is the set of concepts; A
c
is the set of
attributes about concept
Cc ; R is the set of
relations; A
r
is the set of attributes about relation
R
r
; S(c) and E(c) are the sets of relations that
can start and end with concept
Cc ; H represents
the concept hierarchy; and X is the set of axioms.
The concept hierarchy is the set of two-tuples of
concepts that have subsumption relations. It
organizes all the concepts into a well-formed
hierarchy.
Relations need to depend on certain concepts.
However, each relation associate to pairs of
concepts; the number of pairs that is the square
number of concepts may be too large to handle. An
alternate plan is to describe the starting and ending
of relations respectively. S(c) is the set of
relations
)}),((|{ rbacaba,Rrr
; E(c) is
the set of
relations
)}),((|{ rbacbba,Rrr
, where
a, b are individuals. An obvious fact is
that
)()()()(
212121
cEcEcScScc ; the
redundancy can be cleared based on this during
implementing.
Attributes are specific relations depending on
only one certain concept or relation; for example,
person name is a string attribute of person. Axioms
are restrictions about the concepts, relations and
attributes. Each axiom is of course depending on the
elements which it put the restrictions.
Visualization of this model is realizable. It can
be viewed as a tree-like concept hierarchy with
concepts as the nodes and relations between the
nodes; attributes are contained in certain concepts
and relations.
This model of Ontology is expressive enough to
represent ontologies in most Ontology languages. It
is possible to translate ontologies in other languages
into this model and vice versa.
A FRAMEWORK FOR MANAGING MULTIPLE ONTOLOGIES: THE FUNCTION-ORIENTED PERSPECTIVE
303
3.4 Bridge Ontology
The multiple ontologies we need are often
overlapped and have relationships between them.
However, we do not expect to involve the
difficulties of ontology integration. The complete
integration will lost the flexibility of multiple
ontologies in the dynamic web. Here, we try to
generate the relations between ontologies to reach
the functions provide by multiple ontologies. We use
a similar but more powerful method to ontology
mapping, and it is bridge ontology (Wang, 2004).
Bridge ontology can describe more refined relations
between ontologies. The bridge ontology is a
peculiar ontology, and has the ability of expressing
the complex relations between multiple ontologies.
It can be created and maintained conveniently, and is
effective in the applications based on multiple
ontologies. In bridge ontology, 12 kinds of relations
between ontologies are presented such as different
between concepts, complex is-a relationships
between concepts, and composed relations between
relations of different ontologies.
First, we generate the relations between
ontologies based on the requirement of applications.
We call these relations as semantic bridges. The
generation process of bridge ontology has semi-
automatic method to support. And all semantic
bridges are managed by bridge ontology. After the
generation of bridge relations, some semantic
redundancies and conflicts could arise. The problem
also can be solved by the algorithms in (Xu, 2004).
After the relations are generated, we just focus
on the management of these relations. Due to the
process of generating relationships is automatic or
semi-automatic, so we can delete ontology or add
new ontology into the current multiple ontologies.
Now, the ontologies in the multiple ontologies are
independent but their inter-relations are extracted.
3.5 Multiple Ontologies Functions
Some functions should be provided by the multiple
ontologies to satisfy the requirements of applications.
Firstly, all semantic bridges in the bridge ontology
could provide simple and complex ontology
mapping and alignment. Secondly, we could merge
many ontologies as a integrated one through the
semantic bridges. Third, for the bridge ontology
generates the relations across ontologies, we can
realize the knowledge inference across multiple
ontologies. And the fourth function is that we can
query across multiple ontologies because the bridge
ontology provide the transform and rewrite of
querying expressions. Finally, through the
interaction with users’ requirement, we can extract
sub-ontology with complete semantic and
independent function from the multiple ontologies
environment (Kang, 2004).
3.6 Expected Applications
Managing multiple ontologies should server for
many practical applications. The semantic
annotation base on multiple ontologies is a typical
application. We use multiple ontologies to provide
more detailed semantic data, and can avoid the
problems of finding fit ontology or building new
ontologies. The information query based multiple
ontologies is also a promising application. Many
semantic search and query involve multiple
ontologies, where the management of multiple
ontologies can provide the precise or approximate
querying transform. Extracting sub-ontology
corresponding to the requirements also is a useful
application.
4 RELATED WORK
Some frameworks of managing multiple ontologies
are proposed (Noy, Musen, et al., 2004, Cui, 2000,
Maedche, 2003, Maedche, 2002). Some research
also put forward some challenges of the ontology
management (Noy, Musen, et al., 2004, Ding, 2001,
Wendt, 2002). Some people deal with the
management of single ontology. (Noy, 2004) use
ontology tool Prompt Pplugin to manage ontology.
(Stojanovic, 2003) discussed the ontology manage
through the modification of an ontology with respect
to user' needs. These previous work give some
foundations for the management of multiple
ontologies. Different people have different
perspective for the ontology management. (Das,
2001) focus on the ontology management in e-
commerce. But we believe that it is a narrow
multiple ontologies management. Some work of
discuss the ontology management focused on the
multiple ontologies evolutions or versioning
problems. They all ignore the most important goal of
managing multiple ontologies that is using multiple
ontologies to realize the more powerful functions
which the single ontology can not provide for.
Therefore, we propose our function-oriented
perspective for managing multiple ontologies. In
other way, managing multiple ontologies is a
promising way to reuse many existing ontologies.
ICEIS 2005 - INFORMATION SYSTEMS ANALYSIS AND SPECIFICATION
304
5 CONCLUSIONS
Managing multiple ontologies is a challenging issue.
Most previous work focus on the multiple ontologies
maintenance and evolutions, and ignored the
developing the functions of multiple ontologies. This
paper proposed a new framework for managing
multiple ontologies, and its goal is to bring multiple
ontologies together to provide more powerful
capabilities for the practical applications. The
approach is not only feasible, but also robust in the
dynamic and distributed Semantic Web
environment. Some previous researches provide the
foundations for the feasibility of this framework.
Building a system to realize all ideas in the
framework is the next step work.
ACKNOWLEDGMENTS
This work was supported in part by the Young
Scientist's Fund of NSFC (60373066, 60303024,
60425206, 90412003), National Grand Fundamental
Research 973 Program of China (2002CB312000),
and National Research Foundation for the Doctoral
Program of Higher Education of China
(20020286004).
REFERENCES
Berners-Lee, T., Hendler, J., & Lassila, O. (2001). The
semantic web. Scientific American, 284, 34-43.
Corcho, O., Fernández-López, M., & Gómez-Pérez, A.
(2003). Methodologies, tools and languages for
building ontologies. Where is their meeting point?
Data & Knowledge Engineering, 46, 41-64.
Cui, Z., O’Brien, P. (2000). Domain ontology
management environment. In Proceedings of the 33rd
Hawaii International Conference on System Sciences.
DAML Ontology Library. from http://www.daml.org/
ontologies/
Das, A., Wu, W., & McGuinness, D.L. (2001). Industrial
Strength Ontology Management. In International
Semantic Web Working Symposium.
Ding, Y., Fensel, D., Klein, M., & Omelayenko, B. (2001).
Ontology management: survey, requirements and
directions. IST 1999-10132 Ontoknowledge Project,
from http://www.ontoknowledge.org/downl/del4.pdf.
Ding, Y., Foo, S. (2002). Ontology research and
development part 1 – a review of ontology generation.
Journal of Information Science, 28, 375-388.
Gruber, T.R. (1993). A translation approach to portable
ontology specifications. Knowledge Acquisition, 5,
199-220.
Kang, D.Z., Xu, B.W., Lu, J.J., Wang, P., & Li, Y.H.
(2004). Extracting Sub-Ontology from Multiple
Ontologies. In Proceedings of OTM 2004 Workshop
on Ontologies, Semantics and E-learning. Lecture
Notes in Computer Science (LNCS).
Maedche, A., Motik, B., & Stojanovic, L. (2000).
Managing multiple and distributed ontologies on the
Semantic Web. The VLDB Journal, 12, 286-302.
Maedche, A., Motik, B., Stojanovic, L., Studer, R., &
Volz, R. (2002). Managing multiple ontologies and
ontology evolution in ontologging. In Proceedings of
the Conference on Intelligent Information Processing,
World Computer Congress 2002.
Mena, E., Kashyap, V., Illaramendi, A., & Sheth, A.
(2000). Imprecise answers in distributed
environments: estimation of information loss for
multi-ontology based query processing. International
Journal of Cooperative Information Systems, 9, 403-
425.
Noy, N.F. (2004). Ontology Management with the
PROMPT Pplugin. In 7th international Protégé
conference.
Noy, N.F., McGuinness, D.L. (2001). Ontology
development 101: a guide to creating your first
ontology. Stanford Knowledge Systems Laboratory
Technical Report.
Noy, N.F., Musen, M.A. (2004). Ontology versioning in
an ontology-management framework. IEEE Intelligent
Systems, 19, 6-13.
Rousset M.C. (2004). Small can be beautiful in the
semantic web. In: S.A. McIlraith et al. (Eds.): ISWC
2004, Lecture Notes in Computer Science, 3298, 6-16.
Stojanovic, L., Stojanovic, N., Gonzalez, J., & Studer R.
(2003). OntoManager – A System for the Usage-based
Ontology Management. In OTM confederated
international conferences, CoopIS, DOA, and
ODBASE 2003, Lecture Notes in Computer Science,
2888, 858-875.
Wang, P., Xu, B.W., Lu, J.J., Li, Y.H., & Kang, D.Z.
(2004). Theory and Semi-automatic Generation of
Bridge Ontology in Multi-ontologies Environment, In
Proceedings of OTM 2004 Workshop on Ontologies,
Semantics and E-learning. Lecture Notes in Computer
Science (LNCS).
Wendt, O., Stockheim, T., Grolik, S., & Schwind, M.
(2002). Distributed ontology management prospects
and pitfalls on our way towards a web of ontologies.
In Dagstuhl workshop – intelligent software agents
and economical application.
Xu, B.W., Wang, P., Lu, J.J., Li, Y.H., & Kang D.Z.
(2004). Theory and semantic refinement of bridge
ontology based on multi-ontologies. In Proceedings of
the 16th IEEE International Conference on Tools with
Artificial Intelligence (ICTAI’04).
A FRAMEWORK FOR MANAGING MULTIPLE ONTOLOGIES: THE FUNCTION-ORIENTED PERSPECTIVE
305