Integration and exchange of information across the Internet is a universally recognized need, in a wide variety of domains. New promising application fields such as Semantic Web Services can improve the integration process in the Web, dealing with the major difficulty of the semantic heterogeneity in the resources, by means of domain ontologies. In this research, we propose an integration system, combining domain ontologies and Semantic Web Services, to provide an integrated access to the information provided by different Web portals in two application domains: biomedicine and geology. Semantic Web Services were designed to encapsulate Web portals in both domains, allowing the system to accomplish the integration in each domain, separately.
Strong efforts are being applied by the Artificial Intelligence community dedicated to explore potential approaches to improve the process of data discovery and integration. That happens mainly because Web users demand integrated access to all information resources available through the Internet. New promising application fields such as the Semantic Web and Semantic Web Services are being recognized as the adequate support for integrating data sources and services in the Web.
Nowadays, it is possible to find independent and
heterogeneous data resources, accessible through the
Internet, covering information related to any kind of
domain. Information resources include not only data on
the Web but also Web portals which offer services to the
users, such as search engines, providing integrated access
to dynamic content from a variety of sources. However,
only a few Web portals offer their information in the form
of Web Services (such as the portal of Amazon.com). The
power of the Web Service technology lies in the fact that
it establishes a common platform for integrating
distributed computing applications, in intranets as well as
the Internet at large [
The major difficulty when integrating different data sources is the semantic heterogeneity of information sources. Besides being widely distributed across de Internet, and being found in a variety of storage formats, different information sources can use the same object to represent different concepts, or different objects to represent the same concept. In order to solve this problem, applications make use of domain ontologies, which provide a common model of the concepts that are relevant and their allowable relations. The idea of this project is to go beyond this scope and use Semantic Web Services, which make use of ontologies.
The aim of Semantic Web Services is to provide
solutions to the challenges associated with automated
discovery, dynamic composition, enactment, and other
tasks associated with managing and using service-based
systems. Several researches explore the use of Web
Services for information integration, like [
Moreover, according to [2], current integration proposals do not allow users to take advantage of the services offered by Web portals. This happens because traditional integration techniques are focused just on data integration, ignoring the services provided by the portals.
The present research intends to achieve the integration of data and processes (search engine services), which are offered through different Web portals. The portals offer a
variety of functionalities and also permit accessing many data related to a specific domain. The approach is to encapsulate these data sources through Web Services that are semantically annotated based on a global domain ontology, which is specific to each application domain.
A system based on Semantic Web Services and on global domain ontologies was designed to encapsulate the offer of web site services and to allow an integrated access to that information. Once the Web Services are developed and semantically annotated, the system is able to access their functionalities in an automatic way. The main goal is to offer an integrated access to the different functionalities provided by Web portals in specific domains.
The remainder of this paper is organized as follows. An overview of Web Services, the Semantic Web, and Semantic Web Services is given in Section 2. Section 3 presents the proposal of the integration system, giving an overview of the complete project. Section 4 describes the case study realized, using two application domains: biomedicine and geology. The design of a general integration application based on Semantic Web Services is described in section 5. Conclusions are pointed out in section 6.
The Semantic Web is a promising evolution of today’s
Web, providing ways to make the automation process
effective and easy. One of its fundamental components
will be the markup of Web Services to make them
computer-interpretable, use-apparent, and agent-ready
[
Web Services are well-defined and reusable software
components that perform specific encapsulated tasks via
standardized Web-oriented mechanisms [
Web Services were intended to bring a new level of
integration to the computing industry and its networked
communities [
Currently, a Web Service is a software system
identified by a URI, whose public interfaces and bindings
are defined and described using XML [
A Web Service interaction involves two or more
software agents (clients and providers) exchanging
information in the form of messages. First, the provider
registers the service it is offering, using WSDL. The
registry is based on the UDDI specification (Universal
Description Discovery and Integration), which defines
data structures and APIs for publishing (describing) and
discovering Web Services [
WSDL is the W3C recommended language for
describing the service interface, in an XML notation. It
enables specification of data types, operation signatures,
port types, message format and transport protocol details,
network addresses of different ports, and grouping of
different ports into a Web Service [
Currently, there are only few specifications available
for Web Services that are able to provide some kind of
formal definition to what the syntactic descriptions of
their functionalities might mean. Besides OWL-S [
The Semantic Web provides a common framework
that allows data to be shared and reused across
application, enterprise, and community boundaries [
The Semantic Web is an extension of the current Web
where information has well-defined meaning [
A large proportion of today’s data on the Web are
“understandable” only to humans or custom-developed
applications. The Semantic Web will bring structure to the
meaningful content of web pages, creating an environment
where software agents roaming from page to page can
readily carry out sophisticated tasks for users [
Technologies from the Semantic Web can make
crucial contributions to Web Service frameworks [
At present, the use of Web Services requires human
involvement (information has to be browsed and forms
need to be filled in). The existing technologies for Web
Services only provide descriptions at the syntactic level,
making it difficult for requesters and providers to interpret
or represent nontrivial statements such as the meaning of
inputs and outputs or applicable constraints [
The Semantic Web vision, as applied to Web Services,
aims at automating the discovery, invocation, composition
and monitoring of Web Services by providing
machineinterpretable descriptions of services [
Semantic annotations are the base for the automatic
processing of Web pages. In order to achieve automation,
Web Services will be augmented with rich formal
descriptions of their capabilities, such that they can be
utilized by applications or other services without human
assistance or highly constrained agreements on interfaces
and protocols [
A Semantic Web Service is defined through a service
ontology, which enables machine interpretability of its
capabilities as well as integration with a domain
knowledge. The service ontology aggregates all concept
models related to the description of a Semantic Web
Service, and constitutes the Knowledge-level of the
information describing and supporting the usage of the
service [
According to [
OWL-S is an OWL ontology with three interrelated
subontologies, known as the service profile, service model
(process model), and service grounding [
describedBy presents supports ServiceModel
ServiceGrounding
ServiceProfile
OWL-S provides the major Semantic Web Service description language for specifying the functions (preconditions and effects) of an operation and semantic types for each of the inputs and outputs of the service. It offers to extend the capability of OWL to form the required ontology for Web Services, making them machine understandable while supporting automated Web Service composition and interoperability.
A service profile in OWL-S is described through the
properties: input, output, precondition and effect. It
provides the following information: a human-readable
description of the service and its provider; a specification
of the functionalities provided by the service; and
additional information, such as expected response time
and geographic constraints [
Service models are based on the key concept of a
process, which describes a service in terms of inputs,
outputs, preconditions, effects, and its composition of
component subprocesses. A powerful feature of OWL-S is
the ability to model composite processes [
The (OWL-S/WSDL) grounding uses OWL classes as
the abstract types of message parts declared in WSDL.
The central function of the OWL-S grounding is to show
how the inputs and outputs of an atomic process are
realized as messages in transmittable format [
Since the OWL-S service ontology is public and does
not prescribe a framework implementation, it has been
used as the starting point of individual efforts towards
Semantic Web Services [
Whenever users need to look up information related to a given domain, they have to access one or more Web portals, making queries in each one, until they find the answer they are looking for. However, people want to save time, improving the quality of their work. Therefore, an integrated access to the portals of interest would facilitate the user’s task, providing them with a unique tool, where they can access all the available definitions with just one query.
In this work, we propose the development of a domain independent integration system to provide users with an integrated access to data and processes, offered by Web portals that may attend their goals. Two application domains will be used as case studies in this project: the biomedicine and the geological domain, each one providing its own domain ontology. The domain ontologies will help users expressing their needs and will provide information integration by means of Semantic Web Services.
The biomedical domain is supported by the ‘Oncogen’
ontology [
The main goal of this project is having a system, based on global domain ontologies and on Semantic Web Services, that encapsulates the offer of different Web portal services and allows an integrated access to the functionalities provided by them. We apply this system to both domains in order to evaluate the capability of information integration in a domain-independent way. In order to provide those functionalities, a complete system architecture was designed, making use of all the techniques described in the previous section. The integration system comprises all the components illustrated in Figure 2.
The user interface allows users to express their query using an ontology guided tool (based on the global domain ontology), which assists them in expressing their goals. The domain ontology is loaded through the Protégé-OWL API, and its main concepts are used to form a simple menu where the user can choose the type (class) of the object they are looking for.
Through the query component, the system searches
and selects the most appropriate Web Services by
accessing their semantic description. This is done by
matching the term from the domain ontology with the
semantic description of the services capabilities. The
matching is done by using the criteria proposed by
Paolucci et al. [
The invoker, in its turn, uses the grounding of the Web Services, which is part of the semantic description, to invoke each service selected by the query component. It returns the result of each service execution to the core component.
The main component is the core of the system, which acts as an intermediate application among the other components. Once the results are returned by the invoker, they are integrated and given to the user by the core of the system.
As the present study aims at providing a global domain independent solution to the problem of Web portals integration, the biomedicine and the geological domains will be used as case studies, each one providing its own domain ontology.
The development of the complete integration system starts with the design of Semantic Web Services to encapsulate different Web portals in each domain. The idea is to select a few web portals from the ones that provide information about genes, proteins, and their relations, in the biomedicine domain; and about minerals and rocks, in the geological domain. Next step is to design a Web Service to each web portal, offering access to its information. However, to make use of a Web Service, a software agent needs a computer-interpretable description of the service and the means to access it. In order to make it possible, the developed Web Services must be semantically described, based on a domain ontology, so that the automation can be achieved.
People who work in research fields, such as biomedicine and geology, need the source of information to be available whenever they need a specific definition. Therefore, we analyzed Web portals in both domains, choosing the ones that provide search mechanisms or glossaries with definitions and descriptions of interest. From the Web portals that were analyzed, four were used as data sources for biomedicine and two were used for geology. The selected sources are described as follows.
The Gene Ontology project provides a controlled
vocabulary to describe gene and gene product attributes in
any organism. The PubMed database, available via the
NCBI Entrez information retrieval system, is a free search
engine developed by the National Center for
Biotechnology Information (NCBI) at the National
Library of Medicine (NLM). MPact - the Representation
of Interaction Data at MIPS (The Munich Information
Center for Protein Sequences) provides a common access
point to interaction resources at MIPS [
Schlumberger Oilfield Glossary6 is an instant reference that offers accurate definitions for major oilfield activities, reviewed by technical experts. The Geologynet7, through two of its online geology databases: the Mineral and the Rock databases. They have a search interface providing access to a collection of rock and
3 http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?DB=pubmed 4 http://mips.gsf.de/genre/proj/mpact/ 5 http://www.expasy.ch/cgi-bin/sprot-search-ful 6 http://www.glossary.oilfield.slb.com/default.cfm 7 http://geologynet.com/indexa.htm mineral databases that can be used for education and reference.
A domain ontology is a type of formal specification of
a shared conceptualization [
The ontology used in the biomedicine domain, called Oncogen, is a domain ontology for genetic oncology – genes implied in the development of cancer. It is an ontology of biological terminology, which provides a model of biological concepts that can be used to form a semantic framework for data storage, retrieval and analysis tasks. Oncogen was implemented in OWL, through the Protégé ontology editor.
The ontology applied in the geological domain is a refinement of an existing domain ontology, which is part of the PetroGrapher project. It is modeled as a partonomy of terms and was translated to OWL, using the Protégé ontology editor. This ontology represents the vocabulary in the sedimentary petrography area.
The domain ontology provides the best way of helping communities in agreeing on a standard definition of terminology, encoding this definition, then solving the data heterogeneity problem. Each input and output from the operations provided by the portals must be validated against a domain ontology type. This way, the system not only recognizes the parameters as strings, integers, or any other basic type, but also as concepts related to the ontology classes, having an associated meaning. The domain ontology will help users express their needs and achieve information integration by means of Semantic Web Services.
A Web Service must be developed to each data source in order to expose the functionalities provided by the selected Web portals. The implemented services are simple Web Services because they do not rely upon other Web Services nor require further interaction with the user.
Each Web Service is designed as a common class in
an object oriented programming language. Then, a WSDL
file is generated from the designed class, allowing the
specification of the syntax of input and output messages,
as well as other necessary details to the service invocation
[
<wsdl:documentation> PubMed Service </wsdl:documentation> + <wsdl:types> </wsdl:types> + <wsdl:message name="getProteinMessage"> </wsdl:message> + <wsdl:message name="getProteinResponse"> </wsdl:message> + <wsdl:message name="getGeneMessage"> </wsdl:message> + <wsdl:message name="getGeneResponse"> </wsdl:message> + <wsdl:portType name="PubMed_WebServicePortType"> </wsdl:portType> − <wsdl:binding name="PubMed_WebServiceSOAP11Binding" type="axis2:PubMed_WebServicePortType"> <soap:binding transport="http://schemas.xmlsoap.org/soap/http" style="document"/> + <wsdl:operation name="getProtein"> </wsdl:operation> + <wsdl:operation name="getGene"> </wsdl:operation> </wsdl:binding> + <wsdl:binding name="PubMed_WebServiceSOAP12Binding" type="axis2:PubMed_WebServicePortType"> </wsdl:binding> + <wsdl:binding name="PubMed_WebServiceHttpBinding" type="axis2:PubMed_WebServicePortType"> </wsdl:binding> + <wsdl:service name="PubMed_WebService"> </wsdl:service> </wsdl:definitions>
However, the WSDL description by itself does not
provide enough information to support the decision of
which service better suits the client necessities. No
software system can read and utilize the WSDL interface
without human assistance, because the WSDL
specification language provide no means of including
representations of the semantics of the defined operations
and associated messages elements [
The WSDL’s lack of semantic description associated
to the meaning of inputs and outputs makes it impossible
to develop software clients that can, without human
assistance, dynamically find and successfully invoke a
service [
The semantic annotation was done in a semiautomated way, through the OWL-S language (based on the domain ontology, which is written in OWL) using the OWL-S Editor – as a Protégé plug-in. This tool offers a development environment where the domain ontologies are well integrated with the service descriptions.
The OWL-S Editor allows the generation of a “skeletal” OWL-S description based on a preexisting WSDL file. Using this feature, parts of the OWL-S description were generated automatically, based on the inputs and outputs defined in the WSDL file that was generated from the service class. Afterwards, this simple description was enriched by the semantic part, which is the association of the parameters with the domain ontology classes, together with the definition of preconditions and effects. A piece of the OWL-S semantic description of the PubMed service is presented in Figure 4.
The objective of Semantic Web Services is to support clients that can find and correctly utilize newly discovered service without additional programming. The semantic description allows the automatic access to the services by software agents. <j.1:Service rdf:ID="getGeneService">
<j.1:supports rdf:resource="#getGeneGrounding"/> − <j.1:presents> − <j.0:Profile rdf:ID="getGeneProfile"> <j.0:serviceName
rdf:datatype="http://www.w3.org/2001/XMLSchema#string">getGene</j.0:serviceName> <j.0:textDescription rdf:datatype="http://www.w3.org/2001/XMLSchema#string">Get gene description from a given gene name or ID</j.0:textDescription> − <j.0:hasOutput> + <j.2:Output rdf:ID="description"> </j.2:Output> </j.0:hasOutput> − <j.0:hasInput>
+ <j.2:Input rdf:ID="term"> </j.2:Input> </j.0:hasInput> <j.1:presentedBy rdf:resource="#getGeneService"/> </j.0:Profile> </j.1:presents> −<j.1:describedBy> + <j.2:AtomicProcess rdf:ID="getGeneProcess"> </j.2:AtomicProcess> </j.1:describedBy> </j.1:Service>
Having the repositories with semantic descriptions related to both domains, and all the developed components of the application integrated, the system is be able to perform the integration of the sources, based on those repositories, in each domain separately. The answers to the queries are validated against the information in the Web portals. Besides returning the expected results, it is important to notice that only the services of interest are invoked. That is, when a user makes a query, the system does not invoke and execute all the services in the domain to return the expected answer. Instead, it queries the semantic descriptions, building a list with the services that must be invoked, which are only the ones that provide the answer to the user query.
The development process of an integration application based on Semantic Web Services must begin with the collection and formalization of the essential information that should be integrated in the given domain. Based on that analysis, the data design is done in two tasks: Web portals analysis and selection, and the domain ontology specification (or the reuse of an existing one). Next step is to encapsulate each selected Web portal with a Web Service, followed by the semantic description of the designed Web Services, based on the domain ontology.
The tasks described above are specific to each application domain, which means that they must be performed every time a new domain is chosen. After having the semantic repositories, the components of the integration system must be developed and integrated.
The system development comprises the architecture design, the implementation, besides the testing and evaluation of the application as a whole. The complete system must encapsulate all the service offers from the different Web portals, allowing an integrated access to that information, and answering the user’s query.
Web Services, as well as Semantic Web Services, are
applications where Artificial Intelligence techniques can
be used effectively [
The use of domain ontologies, through the OWL-S tool, made it possible to give richer meaning to inputs and outputs through class hierarchies, properties and property restrictions. The chosen Web portals in each domain could be encapsulated by Web Services, which, in their turn, could be semantically annotated using OWL-S, providing a consistent basis to automate the integration process.
The major contribution of the present work is the provision of an integration system architecture with basic main components, which could be used in more than one application domain. The system was able to provide integration both in the biomedical and in the geological domain. The central components in the system perform the integration according to the application domain specified and linked to the system by the domain ontology, and the semantic descriptions.
As future work, we plan to search and select a few more web portals to improve the testing set. Besides that, we plan on extending the case study, adding more application domains, in order to turn the system into a domain independent solution to the problem of integrating web portals functionalities through Semantic Web Services.
[1] Abel, M., L.A.L. Silva, et al. “PetroGrapher: Managing petrographic data and knowledge using an intelligent database application”, In Expert Systems with Applications 26(1 SPECISS), 2004, pp. 9-18. [2] Acuña, C.J. et al. “Toward Web Portals Integration through Semantic Web Services”, In International Conference on Next Generation Web Services Practices – NWeSP’05, 22-26 Aug, 2005, 6 pp.