AN ARCHITECTURE FOR INTEROPERABILITY OF
ENTERPRISE INFORMATION SYSTEMS BASED ON
SOA AND SEMANTIC WEB TECHNOLOGIES
Fuqi Song, Gregory Zacharewicz and David Chen
IMS CNRS UMR 5218 / GRAI, Université de Bordeaux, 351, cours de la Liberation, 33405 Talence Cedex, France
Keywords: Enterprise information systems interoperability, Semantic interoperability, Semantic web, SOA, Ontology.
Abstract: Enterprises interoperability becomes a necessary and critical task for enterprises as the information and
complexity of systems increase exponentially. To adapt existing information systems to new requirements
and build bridges between them, a conceptual vision and understanding of interoperability can facilitate the
technical development to reduce the conflicts and gaps between heterogeneous systems. Now the focus
shifts from the syntactic concern to semantic issues. Semantic heterogeneity currently becomes the next
barrier and challenge to face in enterprise interoperability, it promises to play a major role to enable
interoperability of enterprise information systems. The approach can be transposed at enterprise decision
level, where heterogeneous business processes also need to be interoperable. SOA helps to explore a loosely
coupled architecture with many advantages, especially for enhancing interoperability among enterprises. To
obtain business process interoperability, service orchestration and choreography could coordinate and
compose the services and processes effectively. This paper, first gives a brief introduction of main issues in
this domain, and then recalls a framework for enterprise interoperability. Based on these methodologies and
the interoperability framework, a new architecture for enterprise interoperability of information systems
using SOA ideology and semantic web technologies is proposed, whose final objective is to enable and
enhance enterprise interoperability and then create business values.
1 INTRODUCTION
The development of information systems is a key
issue which influences the progress of enterprises
nowadays. Since the information systems have been
applied for decades in enterprises, numerous
information systems had been developped in various
domains and industries, so they are coexisting at the
same time independently. However isolated
information systems cannot adapt the rapid business
changes, such as the globalization of production,
economic crisis, marketing changes, etc. Thereby
new demands that came up requires multiple
information systems work together to undertake the
tasks that one single system could not complete.
This is the problem of interoperability, whose goal is
to fulfill the gap between new business requirements
and existing information systems in enterprises.
The main basic barrier, which prevents to
achieve interoperability, is the heterogeneity, both
structural and semantic. Data integration is a
solution to solve the problem of information system
interoperability before. It mainly solves the
structural heterogeneity, because these systems were
relatively simple. However, as the amount and
complexity of information systems increase, the
structural heterogeneity becomes a less important
issue, instead, the semantic heterogeneity becomes
new challenge and focus.
Besides the semantics issues, the issue of
business processs among different information
systems is crutial for enterprises. Because the
ultimate goal of enabling information system
interoperability is to make business interoperable.
The objective of business process interoperability is
to coordinate and compose the services into process
to fulfill new and more complex business
requirements.
Thus semantic and business process issues are
two significant points for constructing a flexible and
effective architecture for information systems
interoperability. This paper aims to propose an
operational architecture to facilitate interoperability
431
Song F., Zacharewicz G. and Chen D..
AN ARCHITECTURE FOR INTEROPERABILITY OF ENTERPRISE INFORMATION SYSTEMS BASED ON SOA AND SEMANTIC WEB TECHNOLO-
GIES.
DOI: 10.5220/0003499604310437
In Proceedings of the 13th International Conference on Enterprise Information Systems (ICEIS-2011), pages 431-437
ISBN: 978-989-8425-56-0
Copyright
c
 2011 SCITEPRESS (Science and Technology Publications, Lda.)

of enterprise information systems from IT and
enterprise modeling perspectives. On on hand, with
applying SOA ideology and methodology it provides
a loosely coupled architecture and enhance the
business process integration. On the other hand, the
use of semantic web technogloies allows solving
semantic heterogeneity.
1.1 Framework for Enterprise
Interoperability (FEI)
Enterprise interoperability is the ability for two
systems to understand one another and to use
functionality of one another. In enterprises, solutions
need to be implemented from multiple aspects to
facilitate interoperability. In (Chen, 2006, 2008) a
three dimensional framework (see figure 1) has been
proposed to represent the ability of interactions
between enterprises systems. First the framework
identifies the barriers which are obstacles for
achieving interoperability among enterprises:
conceptual, technological and organizational. Then
four concern points are defined: data, service,
process and business, these points should be taken
into account when architectures and solutions are
built. Generic approaches used for enterprise
interoperability are categorized into: integrated,
unified and federated.
Figure 1: Framework for enterprise interoperability.
1.2 Service-Oriented Architecture
(SOA)
Service-oriented architecture (SOA) is a flexible set
of design principles used during the phases of
systems development and integration in computing.
A system based on a SOA will package functionality
as a suite of interoperable services that can be used
within multiple separate systems from several
business domains. Traditional methods are highly
coupled, this doesn't facilitate the enterprise
interoperability, and is not flexible for scalability,
extendibility and maintenance. Service-orientation
requires loose coupling of services with operating
systems, and other technologies that underlie
applications. SOA separates functions into distinct
units, or services, which developers make accessible
over a network in order to allow users to combine
and reuse them in the production of applications
(Bell, 2008).
A service is usually defined as the realization of
business functionality via software that anyone can
use, anywhere, to compose new applications by
using their services in the context of new or
modified processes (Izza, 2008). The goal of the web
services effort is to achieve interoperability between
applications by using Web standards. Web services
use a loosely coupled integration model to allow
flexible interaction of heterogeneous systems in a
variety of domains including business-to-consumer,
business-to-business and enterprise application.
Service orchestration and choreography are two
terms used to describe the composition and
collaboration of services. A service is a unit of
functionality, which could be regarded as atomic to
provide users certain functions. Most of time, these
services are independent with each other, this may
cause the waste and redundancy, because one service
may be reused for different business processes.
Service orchestration refers to an executable
business process that may interact with both internal
and external web services, whereas service
choreograph means more collaborative in nature,
where each party involved in the process describes
the part they play in the interaction. Orchestration
differs from choreography in that it describes a
process flow between services, controlled by a
single part. More collaborative in nature (see figure
2), choreography tracks the sequence of messages
involving multiple parties, where no one party truly
“owns” the conversation (Peltz, 2003).
Figure 2: Web services orchestration and choreography.
The following of the paper is organized as;
section 2 analyzes the semantic issues in information
systems interoperability. Section 3 describes the
proposed architecture and elaborates each layer. In
section 4, we discuss the proposed architecture
compared with some existing approaches. Last
Business
Process
Service
Data
Federated
Unified
Integrated
Conceptual
Technological
Organizational
Interoperability
concerns
Interoperability
barriers
Interoperability
approaches
WS1
WS2 WS3
WS1
WS2
WS3
Service orchestration
Service choreography
ICEIS 2011 - 13th International Conference on Enterprise Information Systems
432

section presents some future work that will be done
to this architecture.
2 SEMANTIC ISSUES
The methods for data interoperability have a trend
shifting from structural to semantic (Ziegler, 2004).
Because the volume of information increase
exponentially and more systems get involved, to find
useful information and retrieve them from the huge
amount of data becomes necessary. The traditional
structure-oriented approaches have limitations for
retrieving “meaningful” information, because of
insufficient emphasis on semantic aspects.
2.1 Semantic Heterogeneity
(Buccella, 2003) classify heterogeneity into four
categories: (1) structural heterogeneity, involving
different data models; (2) syntactical heterogeneity
presents different languages and data representations;
(3) systemic heterogeneity, involving hardware and
operating systems; and (4) semantics heterogeneity
involves different concepts and their interpretations.
In general, the semantic heterogeneity deals with
three types of concepts: the semantically equivalent
concepts, the semantically unrelated concepts, and
the semantically related concepts. In the first case,
different models use different terms to refer the same
concept, for example, one system may use “teacher”
whereas the other uses “professor”. In the second
case, the same term may be used by different
systems to denote completely different concepts,
such as “dear” may refer to “expensive” or address a
person; and in the last case, different classifications
may be performed, for example one system
classifies “student” as “male” and “female” and
other system as “bachelor” and “master”.
2.2 Ontology
Ontology is regarded as the main approach to solve
the semantic problems. A classical definition
considers ontology as a specification of a
conceptualization (Gruber, 1993). Ontology can
represent the meaning of vocabularies themselves
and their relations. Ontology is frequently used in
data integration for solving the content explication.
In (Wache, 2001), many approaches for information
integration based on ontology are surveyed, and
three possible modes are concluded: single ontology,
multiple ontologies and hybrid approaches. Each
mode has its constraints and benefits, single
ontology is easy to implement but is not adaptive
when adding or removing a data source, whereas
multiple ontology implement costly but has more
flexibility with data source changes. Hybrid mode
facilitates comparison among different sources due
to the shared vocabulary but hard to maintain.
2.3 Semantic Web
The semantic web provides a common framework
that allows data to be shared and reused across
application, enterprise, and community boundaries.
The ultimate goal of the Web of data is to enable
computers to do more useful work and to develop
systems that can support trusted interactions over the
network (W3C, 2010). Semantic web is a group of
methods and technologies to enable the construction
of a connected and reusable data network. (Bratt,
2007) presented a stack to illustrate the main
semantic web technologies.
3 A SOA AND SEMANTIC WEB
TECHNOLOGIES BASED
APPROACH
The objective of our approach is to propose a loosely
coupled architecture for information systems
interoperability. To obtain high interoperability, we
explore the techniques and methods that could
facilitate and benefit it, thereby SOA and semantic
web technologies are applied to design this
architecture. The idea is not developing a new
dependant and separate information system, but
establishing a transparent layer among enterprises
information systems so that they can share and
exchange information without semantic barriers.
3.1 Adaption to the Framework
Based on the enterprise interoperability framework
and the approaches used in (Zacharewicz, 2009), we
focus in the framework to a specific scope:
information system interoperability. We consider all
the barriers and three concern points: data, service
and process. A federated approach is applied due to
the setup of a semantic layer. The mapping is shown
in figure 3 to situate the proposed architecture in the
Framework.
AN ARCHITECTURE FOR INTEROPERABILITY OF ENTERPRISE INFORMATION SYSTEMS BASED ON SOA
AND SEMANTIC WEB TECHNOLOGIES
433

Figure 3: Position of the proposed architecture in the
framework.
3.2 Overview
The architecture can be applied in one enterprise or
among several enterprises (ex: partner relation,
collaborations) to enable business corporation
among them via their own information systems, and
keep their own applications independent. The
architecture includes five layers as illustrated in
figure 4. Each layer is independent to the others,
which communicate through defined interfaces and
protocol. The lower levels are more concerned with
data, in the contrast, the higher level focuses more
on business and users.
Figure 4: Overview of the architecture.
In short, the data interoperability part is carried
out in the two lower layers, and the service and
business interoperability part is done in the two
upper levels, the highest level deals with user
interaction. The architecture covers three
interoperability concerns: data, service and
processes. A semantic layer upon data sources is
designed to emphasize the semantic heterogeneity of
information, not only the information semantics, the
semantic web services are also built in the service
layer to tackle conceptual interoperability barriers.
In addition, the SOA-based methods are used for
solving the technological barriers and providing a
federated architecture. The business coordination
and collaboration take place at process layer, and get
end users are involved via the user interface layer, to
remove the organizational barriers.
First, a semantic layer is built upon data sources,
which is equivalent to the information in the data
sources, through a rule-based processor to map
between the semantic layer and data sources. It does
not aim to transform the data from the data sources
to the ontology server, but a mapping connection
built between them. When retrieving or updating the
data, the real action needs to be effected on the data
sources through the semantic layer. And a monitor
component is set up to keep semantic adaptive and
evolutionary when the information and semantic
change.
Then the upper level is service layer, in which
the services are regarded as atomic. The services
layer is core part of the architecture, because it
provides functionalities and connects information
and business processes. Namely, it retrieves
information from lower level to implement the
services, and then provides the atomic services to
compose business process. Compared with the single
services in this layer, the business process layer is
linked to the real business process performed in
enterprises. The processes are composed by the
atomic services in services layer. In order to
integrate the services into process, services
orchestration and choreography are applied for
coordinating or composing services. In particular,
the description languages WS-BPEL (Web Services
Business Process Execution Language) and WS-
CDL (Web Services Choreography Description
Language) are used for service orchestration and
choreography respectively.
The highest level is the user interface level,
which the user interacts with directly, in this layer
the systems are the ones existed in enterprises or the
future system to be developed. It connects to the
enterprise service bus (ESB) with business process
layer and service layer, in order to call the services
to accomplish the business processes.
3.3 Elaborations of Each Layer
In this section, we define each layer of the
architecture; the detailed illustration is shown in
figure 5.
3.3.1 Data Sources Layer
This layer represents the data sources located in
different enterprises and domains, They can be
accessed directly or through the interfaces they
provide. These sources have three characteristics:
heterogeneous, distributed and autonomous. The
details of each characteristic are as follows.
Heterogeneous: They have different structure and
format, the most common storage method is rational
Federated
Unified
Conceptual Technological
Business
Process
Service
Data
Organizational
Integrated
Enterprise A
Data Sources layer
Semantic layer
Service layer
Process layer
User interface layer
Enterprise B
ICEIS 2011 - 13th International Conference on Enterprise Information Systems
434

database, which is a well-structured format. Another
one is file. However, the formats of files vary a lot
depending on its property. For example
HTML/XML/XHTML is well-structured sources,
whereas excel sheet, word document are semi-
structured sources. As well as non-structured sources,
such as plain text document (.txt) and some other
documents with special designed format.
Distributed: The sources locate on different
domains either physically or logically. Such as in
one enterprise, the database could be deployed to
several cities, because there is one branch in each
city. Even two databases are deployed in one same
server, they belong to different database system, for
example, Oracle and DB2 respectively, this is an
instance of logically distributed.
Autonomous: Each source is logically
autonomous, each source can be regarded as one
system, and they are dependent to each other. They
are connected to the systems via interfaces or API
(Application Programming Interface).
3.3.2 Semantic Layer
Upon the structural data sources, a semantic layer is
established to represent the semantics of the
information. Not all information will be taken into
account when building the semantic layer, but only
the information which is useful and valuable to the
integrated business. Thus a rule-based processor is
constructed to map and retrieve data from the lower
data sources layer. With defined rules, the processor
could filter and control the data so as to build a
concise and effective semantic layer. The processor
can also update the data sources when a change
request is sent from the upper level.
Ontology is used for representing the semantics
of information. An ontology server is deployed as
the storage of the ontology. As we introduced in the
introduction in the recall section, the ontology server
is applied with the global schema approach due to its
advantages.
Figure 5: Five layers of the architectureAnother important issue when building semantic is to keep it adaptive and
evolutionary. Because the information in data sources changes frequently, such as adding a new database or removing some
files. The ontology needs to keep consistent with the real information layer. Even the data itself doesn’t change, the
meaning of information could change date by date. Because we are dealing with semantics, when the meaning of a word
changes, the old ones become meaningless and incorrect, so keeping evolutionary is necessary and important. To do these
tasks, a component called monitor is set up to keep the ontology update-to-date and correct.
Query Engine
Rule-based Processor
Semi-structured files
HTML
Text files
XML/XHTML
Databases
Meta-data
stora
g
e
Ontology
server
Service layer
Semantic layer
Data layer
Process layer
Enterprise Service Bus(ESB)
Process integrator
QoS, Security, Statistics
Management and Monitoring
Orders M.S.
Inventory M.S.
HR M.S.
Finance M.S.
WS-BPEL/WS-CDL
Service & Process
Presentation
Information
SOAP/UDDI/WSDL/OWL-S
User interface layer
Monitor
AN ARCHITECTURE FOR INTEROPERABILITY OF ENTERPRISE INFORMATION SYSTEMS BASED ON SOA
AND SEMANTIC WEB TECHNOLOGIES
435

A meta-data storage is designed to abstract the
global ontology vocabularies. Furthermore it could
be used to extend the global ontology schema to
multiple and hybrid schema when single mode has
too much limitations and constraints for a specific
future system.
There are many languages for representing ontology,
each one has its highlight and constraints, the
selection will depend on the specific focus and
requirements. (Pulido et al, 2006) and (Maniraj,
2010) give a review of the ontology languages,
(Corcho et al, 2002) presents the different ontology
languages with detailed comparisons and introduces
the tools for building ontologies.
3.3.3 Services Layer
Service layer lies above the information and
semantics, these services are the existing and new
developed ones located in different information
systems and enterprises. Traditionally, these services
are independent and not related; however, they could
be reconstructed to compose into a process that
fulfills certain tasks. The upper level will perform
the service integration. In this layer, the services are
gathered to establish a semantic web services layer,
compared with the conventional web services,
semantic web services has the ability to understand
the services and the relations between them.
A component called query engine is designed for
querying and answering between the semantic layer
and the services layer. The query engine is ontology
oriented, such as, SPARQL (W3C recommendation,
2008), so that it can query the ontology in the
semantic layer. An optimizing mechanism is set up
in this query engine, for optimizing the query
request sent from application level and the answers.
3.3.4 Business Process Layer
Business process layer represents the business
activities in enterprises, these business processes
happen internally or among enterprises. These
processes are composed with the atomic services in
services layer. Service orchestration and
choreography are applied to coordinate and compose
services into business process, for example WS-
BPEL and WS-CDL are specific description
language for business process integration.
WS-CDL is an XML-based language that
describes peer-to-peer collaborations of parties by
defining, from a global viewpoint, their common and
complementary observable behavior; where ordered
message exchanges result in accomplishing a
common business goal (W3C, 2004). WS-BPEL is
an OASIS standard executable language for
specifying actions within business processes with
web services, WS-BPEL export and import
information by using web service interfaces
exclusively (OASIS, 2007).
A business integrator is designed to perform the
process integration task, and in charge of connecting
to the ESB (Enterprise Service Bus) (Chappell,
2004). The dotted line between the process and the
services in the figure 7 is logically connected. It
means a certain process is composed with several
services, but physically, the process and services are
connected via ESB.
3.3.5 User Interface Layer
This layer represents the user level, namely, the
interface the users interact with, such as the human
resources system, orders management system,
inventory management orders and etc. Via the
defined interface, the applications connect to the
ESB and explore the process and service to complete
user requirements. New modules and functionalities
could be added and developed within the current
systems. So this could keep the dependability of the
current systems, they don’t need to be changed to
adapt the new requirements, because the cost for
updating the legacy or developed systems is very
high, we try to minimize the unnecessary changes
and maximize the value-added features.
4 DISCUSSION
In this paper, we addressed the importance of
semantic and business process for enterprise
information system interoperability. Based on the
ideology of SOA, we proposed a new architecture to
integrate enterprises information systems, which
uses semantic web technologies to construct a
semantic layer upon data sources, and service
orchestration and choreography to integrate business
process. The top level is the information systems,
with which the end users interact directly. The
applications connect to the ESB through the defined
interface, and then invoke the processes and services
to fulfill business requirements.
Ziegler (2004) sums up the usual methods for
data integration by six levels of integration. They are
described from the architectural perspective of
information systems. In particular, at the user level
only “Manual” integration, which has a very low
integration degree, is available. The most often used
approach is “Uniform Data Access” to solve the data
integration problems, which provide user a unified
user interface of physically distributed data, such as
ICEIS 2011 - 13th International Conference on Enterprise Information Systems
436

portals and mediated query systems. “Common data
storage” uses a method to transfer the data sources to
a new data storage, warehouse and operational data
stores are such kind of instances.
Compared with these integration approaches,
interoperability has more the meaning of coexistence,
autonomy and federate environment, whereas
integration refers to the concepts of coordination,
coherence and standardization (Chen, 2008). The
interoperable architecture will provide added values
for enterprise information systems interoperability.
First, due to the semantic technologies used in the
architecture, the “understanding” problems can be
facilitated among different information systems.
Secondly, with SOA, the architecture is loosely
coupled, where each layer is only connected through
interface and protocols. The data sources, services
and business processes could be added or removed
without affecting the other parts of the architecture,
gaining flexibility for the enterprise. Thirdly, it
provides high scalability; this architecture could be
applied internally in one enterprise for its sub
branches or different departments, as well as among
different enterprises and partners. Involved
applications and services can be reused ones; also
new developed ones can be added easily. We believe
this approach is promising regarding web 2.0
orientation of all enterprise applications.
5 FUTURE WORK
The future work will include, first, more details to
describe the running mechanism of the architecture,
such as, the interfaces between each layer. Second,
we will elaborate and refine the semantic layer to
define how the ontology presents the semantics of
information; third, the aspects of enterprises
business level (e.g. decision level) will be taken into
account to make the architecture adapted the
complex situation in enterprises. The architecture
will be applied to a project in the following research,
to figure out the improvements and get feedbacks
from industry.
REFERENCES
Bell M. (2008). Service-Oriented Modeling: Service
Analysis, Design, and Architecture, Wiley & Sons
Bratt, S. (2007).Semantic Web, and Other Technologies to
Watch, [Online], Available: http://www.w3.org/2007/
Talks/0130-sb- W3CTechSemWeb, [Dec. 14, 2010]
Buccella, A., Cechich, A. and Brisaboa, N. (2003)
‘Ontology-based data integration methods: A
framework for comparison’.
Chappell, D. (2004). Enterprise Service Bus, O’Reilly
Chen, D., Daclin, N., (2006) ‘Framework for enterprise
interoperability’, Proc. of IFAC Workshop EI2N 2006,
Interoperability for enterprise software and
applications (ed. H. Panetto and N. Boudjlida),
Bordeaux, March 21, 2006, pp. 77-88.
Chen, D., Doumeingts, G. and Vernadat, F. (2008)
‘Architectures for enterprise integration and
interoperability: Past, present and future’, Computers
in Industry, vol. 59, no. 7, September , pp. 647-659.
Corcho O., Fernández-López M., Gómez-Pérez A. (2003).
‘Methodologies, tools and languages for building
ontologies. Where is their meeting point?’, Data &
Knowledge Engineering, vol. 46, no. 1, pp. 41-64.
Gruber, T. (1993). ‘A Translation Approach to Portable
Ontology Specifications’, Knowledge Acquisition.
Available: http://tomgruber.org/writing/ontolingua-
kaj-1993.pdf.
Izza, S., Vincent, L. and Burlat, P. (2008), ‘Exploiting
semantic web services in achieving flexible
application integration in the microelectronics field’,
Computers in industry, vol.59 , no. 7, pp. 722-74
Maniraj V., Sivakumar R. (2010). ’Ontology Languages –
A Review," International Journal of Computer Theory
and Engineering, vol. 2, no. 6, pp887-891.
Peltz, C. (2003), ‘Web service orchestration and
choreography – a look at WSCI and BPEL4WS’, Web
Service Journal. Available: http://wpage.unina.it
/rcanonic/ didattica/at /documenti/wsOrchestration.pdf
Pulidoa J.R.G., Ruizb M.A.G., Herrerac M.A.G., Cabellod
E., Legrande S., Ellimanf D. (2006). ‘Ontology
languages for the semantic web: A never completely
updated review’, Knowledge-Based Systems, vol. 19,
no. 7, pp. 489-497.
W3C recommendation, Query Language for RDF (2008),
[Online], Available: http://www.w3.org/TR/rdf-sparql-
query/ [Dec. 20, 2010]
W3C study cases (2010) ,[Online], Available:
http://www.w3.org/2001/sw/sweo/public/UseCases/
[Dec. 14, 2010]
Wache, H., Gogele, T., Visser, U., Stuckenschmidt, H.,
Schuster, G., Neumann, H. and Hubner, S. (2001).
‘Ontology-based integration of information – a survey
of existing approaches’, Wache2001 , pp. 108-117.
Web Services Business Process Execution Language (WS-
BPEL) OASIS Standard (2007), [Online], Available:
http://docs.oasis-open.org/wsbpel/2.0/OS/wsbpel-
v2.0-OS.html [Apr.11, 2007]
Zacharewicz, G., Labarthe, O., Chen, D. and Vallespir B.
(2009). ‘A Multi Agent/HLA Platform for Enterprises
Interoperability: Short-Lived Ontology Based’ in
Electronic Supply Network Coordination in Intelligent
and Dynamic Environment: Modeling and
Implementation, IGI Global
Ziegler, P. and Dittrich, K. (2004) ‘Three decades of data
integration – all problems solved?’, In 18th IFIP
World Computer Congress, Volume 12, Building the
Information Building the Information Society, vol.
2004, pp. 3-12.
AN ARCHITECTURE FOR INTEROPERABILITY OF ENTERPRISE INFORMATION SYSTEMS BASED ON SOA
AND SEMANTIC WEB TECHNOLOGIES
437