Improving Your Web Services Thorough Semantic Web Techniques

Improving Your Web Services Thorough Semantic Web Techniques

J.P. Liyanage, G.N. Wikramanayake
University of Colombo School of Computing, Colombo-7
Telephone: +94 0114 920562
Email: janakasoft@yahoo.com, gnw@ucsc.cmb.ac.lk

Abstract services. Jena API was used to manipulate OWL data
and the Pellet engine was used for the reasoning.
Most of the data in the internet are markup for text
and graphics in HTML format and are driven by syntax.
Keywords: Semantic Web, OWL-S, Semantic Web
These data are displayed to the user but the computers
Services, Service Brokerage
are concern there is no meaning of these data. If we
semantically annotate data in the internet, then the
computers can process these data in meaningful ways
and increase the usability of the data. In this project a 1. Introduction
possible architecture for a semantics based Web services have been in the forefront of the web
middleware agent (broker) will be looked at, which has technologies in the recent past and allow much more
the capabilities to serve web service requests by rich interaction between the client and server,
searching through a larger space of web services. The compared to the surfing of the HTML (Hyper Text
matching capabilities of a semantics based broker are Markup Language) pages scattered around the web.
much higher than any of the syntax based middleware Since web services allows intelligent agents and/or
frameworks in the market today like UDDI. applications in the web to access much more specific
In this project, we will be using two of the current data compared to screen scraping methods used in
semantic web technologies OWL (Ontology Web conventional applications, it allows much more usable
Language) and OWL-Services. Once the broker and targeted information to be forwarded to the end
receives a client request that contains the capabilities user. However, finding a suitable web service to satisfy
and input/output specification of a web service client a particular requirement had always been challenging.
wishes to invoke, it will read OWL-S descriptions of Registries containing information about many different
the services registered to the broker and try to find a types of web services are common in the World Wide
direct match. If it is unable to find one the broker uses Web. UDDI (Universal Description, Discovery, and
its powerful inference capabilities (not available in Integration) is the most popular standard of describing
conventional syntax based middleware) to dynamically such type of Web service description registries [1].
compose a set of available services together into a new However all of these registries are syntax based,
service. which mean that they are usually inflexible and
Our IntelliBroker works as a true broker in the inadequate. This also means that the data in these
sense that client may even not know what the invoked registries cannot be used to infer any other useful
services are. In executing the created composite relationships between the web services than those
service, the broker has the ability to exploit any explicitly defined in the registry. For example if the
parallelism between the child services and return supplier A publishes a service to sell “Published
results to the client faster. The broker will convert Documents” and a user B wants to buy “Books” and do
OWL types to SOAP messages and vice versa using a search on “Books” he will not find the A’s web site.
XSLT (eXtensible Style Language Transformations). This is simply because there was no way to define the
This paper focuses on the design and the architectural relationship between the “Published Documents” and
of our system. “Books” in the syntax based Registries or more
We implemented the system with Java language and generally in the syntax based Web.
used Axis framework for publishing and executing web Semantic Web [11] attempts to solve these issues
but is still in its infancy. The idea behind semantic web

14

has been to make machines understand the meanings WSDL
(semantics) of the contents in the web and thus Web Service Description Language (WSDL) [2]
allowing them to infer any other implicit relationships provides a common XML grammar for describing
between the data. Semantic Web proposes a language services and a platform for automatically integrating
called OWL (Ontology Web Language) [18] to those services; thus it is a mandatory component for
annotate semantically all the data published to the web. any web service. Using WSDL, a client can locate a
Applying semantic web techniques to the web service and invoke any of its publicly available
conventional web services has resulted in what are functions [2]. Since machines can understand and
known as semantic web services [10] and the aim of process WSDL specification, it is possible to automate
this project is to design a middleware agent (a broker) the process of integrating with a new service (e.g.
[20] for semantic web services. This brokerage system generation of a proxy class in Microsoft .NET [13]).
(IntelliBroker) process user requests for web services,
RDF
and map them into one or more web services, which
are capable of serving the request. Resource Description Framework (RDF) [12] is used
as the foundation for representing metadata about Web
In addition to this, we have proposed a distributed
resources such as the title, author, and modification
architecture for a Broker. In this distributed
date of a Web page, copyright and licensing
architecture, rather than a single centralized broker
information about a Web document [12]. It provides
handling all the requests, a set of loosely coupled
syntax for expressing simple statements about
brokers will share the knowledge between each other
resources, where each statement consists of a subject, a
in serving the requests of clients. The brokers interact
predicate, and an object (triples). In RDF resources are
with each other using a semantics based protocol
identified by URIs (Universal Resource Indicator) and
(called DUBIP) in the semantic web. Thus, the
thus can be used to describe web based resources
IntelliBroker is able to match more web services than a
effectively.
conventional broker based on syntax would.
The rest of the paper is organized as follows. A RDFS
background on web services and semantic web is One limitation of RDF is that it does not have any
introduced along with related technologies. Then our facilities for defining structures or hierarchies [8].
methodology is explained. Our proposed distributed RDFS (RDF Schema) was thus introduced as an
architecture is described next. The implementation of extension to RDF, to complement RDF with a type
our ideas and the results are then presented. Finally system. It provides the facilities needed to specify
conclusions and future work is highlighted. classes and properties (defined as a directed binary
relation) in RDF. In RDFS, a set of new terms has been
2. Background introduced to define classes, subclasses and properties
applicable to them.
2.1. Web Services
2.2. Semantic Web
Web services are built mainly upon XML and
HTTP [20]. HTTP (Hyper Text Transfer Protocol) is a First introduced by Tim Berners-Lee [11], it is about
ubiquitous protocol, running everywhere on the a new form of web content that would be meaningful to
Internet while XML (eXtensible Markup Language) computers and thus allow computers to infer
provides a meta-language using which you can write meaningful relationships between data. Semantic Web
specialized languages to express complex interactions technologies are supposed to be a solution to the
between clients and services or between components of problem of allowing the data in the internet available to
a composite service. Using XML, other technologies a much broader range of consumers (either human or
like WSDL [2] and SOAP [20] are built, which further machines) preferably through automated agents.
defines the platform elements of web services. Semantic web is an extension to the current Web,
where information is given a well-defined meaning and
Simple Object Access Protocol (SOAP) is a
thus machines can process and “understand” the data
protocol specification that defines a uniform way of
rather than merely displaying them like in the current
passing XML-encoded data. It also defines a way to
Web [11].
perform Remote Procedure Calls (RPCs) using HTTP
as the underlying communication protocol [20]. The Semantic Web principles are implemented in
layers of Web technologies and standards [Figure 1].
The Unicode and URI layers make sure the use of
international characters sets and provide the method of

15

identifying objects in Semantic Web. The XML layer Classes and Properties
with namespace and schema definitions assures the OWL extends the notion of classes defined in RDFS,
integration of Semantic Web definitions with the other with its own construct. OWL also extends the notion
XML based standards. With RDF and RDFS, it is RDFS property by introducing Data-type and Object
possible to make statements about objects with URIs properties. OWL also introduces a set of property
and define vocabularies that can be referred to by URIs. characteristics, which results in considerable inference
This is the layer where we can give types to resources capabilities [7]. For example, Colombo region can be
and links (properties). The Ontology layer supports the located in the Asia region if given that Sri Lanka is
evolution of vocabularies as it can define relations located in Asia region and Colombo located in the
between the different concepts. With the Digital region Sri Lanka.
Signature layer, agents or end users can detect any
alterations to documents. 2.4. RDQL
Resource Description Query Language (RDQL) is
the most commonly used query language to obtain
information from databases containing RDF data like
OWL or OWL Services. The work of RDQL is similar
to the work of SQL in relation to the relational
databases [17].

2.5. OWL-S
Figure 1: Layers in the semantic web OWL-S (OWL Services) is a set of ontologies
designed in OWL to describe web services in semantic
The top three layers (Logic, Proof and Trust), are web. By using this common set of ontologies, the web
currently being researched under W3C and simple services will be universally available to any client that
application demonstrations are being constructed. The can understand OWL. Actually there is another
Logic layer enables the writing of rules while the Proof framework called WSMO (Web Service Modelling
layer executes these rules. Trust layer provides a Ontology), aiming to describe semantic web services
mechanism to determine whether to trust a given proof [15]. However since OWL-S is the most widely
or not [10]. Proof and Trust are very important accepted choice for semantic web services and is the
concepts in Semantic Web since if one person says that recommended one by W3C, we used OWL-S for
X is blue and another says that X is not blue, we need a service description in this project.
way to determine which is true [14].
Service Profiles, Models, and Groundings
2.3. OWL The top level of the OWL-S ontology is the Service
class, which corresponds to a web service. One needs
The Semantic Web needs a support of ontologies, to know at least three things about a Service: what it
which is defined as explicit specification of a does, how it works and how one might access it. The
conceptualization [5]. Ontologies define the concepts relationships between these classes are depicted in
and relationships used to describe and represent an area Figure 2.
of knowledge. OWL has been designed and introduced
by W3C to be able to describe ontologies. OWL
(Ontology Web Language) is a layer on top of RDFS,
which defines an additional set of terms to describe the
relationships between the resources in a much richer
fashion.
Sub languages
To compromise between rich semantics for
meaningful applications (expressive power) and
feasibility/implementability the OWL support three
different sublanguages Lite, DL and Full [6, 18].

Figure 2: Top-level class relationships in OWL-S

16

2.6. Web Service Brokerage CMU OWL-Broker
A web service broker finds suitable web services for This is a prototype system developed by the W3C
the client based on the capabilities requested by the designers of the OWL-S specification[16]. They have
client. The tasks performed by a broker are twofold: based their implementation on a generic OWL-S
processor called OWL-S VM. The broker’s activities
Search: Finding the best service(s) matching to the are divided into two parts namely the advertisement
requesters query. The broker may have to compose protocol and mediation protocol. In the advertisement
existing services together in order to derive new protocol the service providers’ service metadata are
services, which will satisfy the query. collected in internal broker registries. In the mediation
Mediation: Converting back and forth between the protocol following activities is performed:
data formats used by requester and provider.
The requester query the broker using OWL Query
Matching capability of two brokers can be Language (OWL-QL)
compared using the number of matches given for a The broker searches for a matching provider using
particular query. If the number of matches given is the advertisements in its registry
higher for Broker A than for Broker B then broker A is After selecting the best provider, the query is
said to be better than B. This number depends both on mapped into the inputs expected by that provider.
the search and mediation capabilities of the broker. Upon receiving the reply from provider, it is
That is the amount of the services searched and the mapped back into the outputs expected by
ability to mitigate the differences between the requester.
consumer and provider. Finally, the output is sent back to the requester.
In addition to these main functionalities, a broker
should be secure (in the sense only authorized parties This broker does not have the ability to create (by
can access it). A broker can also provide services like composing) new web services and matchmaking is
web service promotion, negotiation and maintaining limited to the services registered to the broker. It also
QoS (Quality of Service) metrics of registered services. does not take the QoS metrics into consideration when
finding the matches.
2.7. Current Brokerage Systems OWLS-MX
Complete brokerage systems that are capable of This is a hybrid semantic web service discovery tool in
doing all of the tasks of discovery (matchmaking), the sense it uses both semantic reasoning and syntactic
service composition, mediation and execution based similarity metrics to obtain the best of both
automatically without human intervention are not worlds [19]. The service descriptions must be in the
available to date. There are partial solutions, which OWL-S format and the system is implemented in Java.
implement parts of the above functionalities with the It uses the OWL-DL reasoner called “Pellet” for
human intervention developed using OWL-S. We will semantic reasoning. This is a very recent system
explain three of the most popular systems, namely: developed in 2006. However, this has only the
XPlan, CMU OWL-Broker and OWLS-MX. discovery capability and human intervention is
required throughout the process of selecting services to
XPlan testing whether the matched services are correct.
This tool is a composition-planning engine for However, the program has a very easy to use GUI,
semantic web services, which are described in OWL-S. which allows us to visualize the whole process of
It converts OWL-S 1.1 services to equivalent problem matchmaking [3].
and domain descriptions that are specified in the
planning domain description language (PDDL), and 3. Methodology
invokes an AI planner (XPlan) to generate a service
composition plan sequence that satisfies a given goal In order to derive a model to implement the required
[9]. However, this tool is intended only for expert users functionalities, we design a system called DUBIP that
and provides a nice GUI to manipulate the composed has three main components:
services manually. It only allows compositions with The client agent or service requester
sequences of child services. Additionally the tool has The web service broker
the ability to consider QoS metrics in selecting suitable
The server or service provider
candidate services for the composition [4].
The high-level interactions among these three
components are given in the Figure 3 below.

17

4. Architecture
In the following sections, we have given a brief
description of the architecture of the three main
components of this brokerage system.

4.1. Client (requester)
The agent’s query should contain two different types of
information:
Figure 3: High-level components and Interactions
Here, first the client send a query stating the A description of the capabilities of the service
requirements for the service he wants to invoke and the This is related to the metadata of the service we are
data required to invoke the service. The broker will looking for. It will be a set of OWL classes, which will
search through the registered services and invoke a describe the preconditions and effects of the service we
service that has capabilities to satisfy the client request. are searching.
Service providers register the services, by submitting Input data for the invocation of the service
the descriptions of their available services. In addition, This is a set of OWL class instances, which are
DUBIP introduces other brokers into the system, so supplied as a possible set of inputs for the service, the
that those will be referred if a matching service is not client is expecting to execute.
found in the current broker.
Here the broker will return an error code to the
We have used the OWL-S specification version 1.1 client if a service with the requested capabilities are not
to describe the provider services to the broker. These found. It is up to the client agent to handle these errors
semantic descriptions made available to the broker, in their own ways.
will make it possible to find matching services to the
requester query and to compose new services to satisfy To handle complex message handling with the
the query. In order to find matches, the broker needs to broker, a client-side component is designed. This
do reasoning, which requires that the broker can access component will simplify the usage of the broker by the
all the ontologies referred by the service descriptions. user applications, which does not need to know about
An example plan generated by the broker to satisfy a semantic markup or existence of multiple brokers.
client query of booking a cheapest flight from With respect to the DUBIP this client-side component
Colombo to New York is given in Figure 4. Here four is required to handle the following tasks:
web services has been used sequentially and iteratively Selecting the nearest or fastest broker
to create a new service. Selecting the next available broker if the current
The agent at the client side needs to know the one fails
semantic markup in order to formulate a query to be The main processes of the client-side component are
sent to the broker and to process the results sent back depicted in the Figure 5. The client-side component
by the broker. should be a platform dependent system, which needs to
interact with the user applications very closely.

Client Component

Available
Handle
brokers
DUBIP

Broker
Client
Application
Invoke Broker

Figure 4: A composition plan for a query

Figure 5: Processes in the Client-side component

18

4.2. The Broker custom ontologies will be stored as files in the file
system of the broker.
This will be the core component of the semantic
brokerage system, which will handle the main tasks of Providers can register either Atomic services, which
registration, discovery, composition and mediation. can be executed directly or Composite services, which
While handling these tasks the broker must be aware of can only be executed after matching atomic parts are
the other brokers and co-exist with them. To found. An Atomic service is registered by submitting
implement the complex set of requirements to be an OWL-S document containing a Service, which is
fulfilled by the broker, it is divided into a set of tightly DescribedBy a single AtomicProcess. Similarly, a
integrated set of modules. These modules and the Composite service is registered by submitting an
interactions among them are depicted in Figure 6. OWL-S document containing a Service that is
DescribedBy a single CompositeProcess.

Service
Upon registration the broker will need to update
provider QoS information for each of the invocation on the
service. That is if the service execution was not
Actual Web
service
successful for some request then the Quality metric of
Web that particular service will be reduced. If the Quality
Interface metric of the service is reduced below a specified
Broker threshold, the service will be unregistered from the
broker and the relevant service provider will be
OWL-S notified using the contact details in the database.
Web service
registry OWL-S description
registration
Service Composition
groundings
The query from the requester is received by this
Discovery Web Service component and will be handled mainly by this
execution component before passing onto other components.
Service tree
Upon receiving the query from the requester, the
Search results
actions taken by this Composition component are
Output
depicted in Figure 7.
Web service
Composition The Service Composition component works as the
Search results
mediator between the activities of all other components
Search other Other in the Broker, as can be seen from the Flow chart in
brokers Brokers
Figure 7. The processes 1, 2 and 5 of Figure 7 are
executed by the “Service Discovery” component. The
processes 4 and 6 are parts of the “Service Execution”
component while the process 3 is a part of the “Search
Other Brokers
other Brokers” component.
Client agent or Client-side
component
An important thing to notice here is that we always
application
prefer Atomic services to Composite services in
searching for matches. The composite processes
Figure 6: Main modules inside the brokerage mentioned here are of two types:
engine The skeleton services registered by the providers
Service Registration Services dynamically generated by the
New web services will be registered by submitting composition engine
the OWL-S description of the service with a service
provider. A provider can register more than one service
and before the service is registered, the provider is
registered with the broker.
The OWL-S description of a service may refer to
custom ontologies designed by the provider. In this
case, the referred ontologies can either be made
available through an URL or submitted as a part of the
registration request. The OWL-S description and the

19

1
Search for an The boxes in thick borders denote Atomic services
atomic process
while boxes with thin borders denote Composite
4 services. Here the service A is found by the process 2
Execute the
Found a
match
Yes Atomic in Figure 7 and then the other services B-G are found
Process
by the process 6 in Figure 7 by recursively searching
No
for matching parts in all the composite services. After
2 Search for a
composite
all the Composite services are decomposed into
process 5 Atomic services the root service (e.g. service A in this
7 Search for a
Get the matching atomic case) which satisfies the query, can be executed since
Found a
match
Yes contained
processes
or composite
processes
the WSDL groundings are available to all the lower
recursively level Atomic services.
No 6

3 Search in other Execute the atomic
brokers Are all processes
processes Yes according to the
atomic data and control
flows defined by
composite process
Found a
match

No
8 Return “No
Match” to the No
requester

Figure 7: Flow chart showing control flow among
the components of broker
Figure 8: Tree of services discovered to satisfy a
For the Composition component to return a
Query
successful response to the requester it should have
In matching inputs, outputs, preconditions and
found at least one Atomic service and zero or more
effects by comparing the OWL classes of these
Composite services, which specifies the data and
instances, there are three possible options namely exact,
control flows among these Atomic services.
relaxed, and subsume. In exact matching required and
One important activity handled by this Composition available class are equal. In relaxed matching required
component which is not shown in Figure 7, is the type is a sub class of the available type and vice versa
asking for more inputs from the requester in case the in subsume matching.
inputs provided by requester are not sufficient to
invoke the selected service. The requester will be asked Service Execution
for more inputs only if the requester has capabilities to After the Tree of services satisfying the query has
provide inputs on demand (i.e., if the mode-2 is used as been found by the “Service Discovery” component, it
described previously). is forwarded to this component. Here for each of the
Atomic services, the groundings will be obtained from
Service Discovery
the OWL-S registry by reading the ServiceGrounding
The service discovery would be done by checking from each of the specific OWL-S documents. The
the ServiceProfile of the OWL-S documents stored ServiceGrounding specifies the URL of WSDL
in the OWL-S registry. The preconditions and effects document and the XSL transformations (which defines
(indicated by hasPrecondition and hasResult the conversion between OWL data types and WSDL
respectively) of the ServiceProfile will be match data types). The WSDL document specifies the
with those specified in the query to find a match or a location of the concrete web services and thus the web
set of matches. In addition, if more than one composite service can be executed.
service is found, the services with least number of parts The executions of the services in Tree hierarchy will
are preferred and thus searched for matching Atomic proceed in Bottom-Up order, feeding the Output of the
parts firstly. lower level services to the upper level service, until the
The Atomic and Composite services discovered to Output of the root level service is available. This will
satisfy a query, can be thought of as making up a tree be returned to the requester as the response to the query.
hierarchy in which Atomic services make up leaves The execution component exploits parallelism
with Composite services making up non-leaves. among the services to reduce the service time. If two
Figure 8 shows a possible hierarchy of services services are not dependent on each other (through
discovered to match the query described in Figure 4. output to input forwarding), the execution engine

20

identifies that sends SOAP requests to both the can also specify the matching strength used in
services at the same time. comparing the input and output types for matches.
“Equivalent” denotes the strictest match condition
Search other Brokers
where the types should be the same for a match.
This module searches for matching services in other
brokers known to it. This is how the overall process
happens:
A limited number of brokers (n number of
brokers) known to the broker will be stored in the
“Other Brokers” registry shown in Figure 6.
The list of brokers is refreshed frequently to store
the n closest brokers to this broker.
Upon receiving the search request from
Composition component, the search request is sent
to the closest broker (broker B) to this broker
(broker A).
Broker B searches its registry for a matching
service. If one is found it would be sent to A.
Otherwise B will send the search request to its
closest broker exempting A (say C). The process
continues until a match is found or a threshold
number of brokers are searched.
Say at broker M a match is found (or threshold),
then M will directly send the reply to the original Figure 9: The client GUI for formulating requests
broker A (or no match error if threshold). to be sent to the broker
After invoking the broker, the resulting composite
5. Implementation service created by the broker is shown in the fourth tab
and the actual response, which contain the output data
We have implemented the system using Java as our
values are shown in the fifth tab.
main programming language. We have used the Jena
API to manipulate OWL descriptions associated with
services are well as to reason about the service 6. Results
descriptions and other custom ontologies. We tested the brokerage system by using number of
The Java web services were implemented using the web services from a flight-booking scenario. We
Axis framework and hosted in Apache Tomcat server. invoked the broker using various inputs, outputs and
We used the same Axis framework to execute provider effects using the prototype client component we have
services. The broker does matching based on inputs, implemented. We have also used super classes and sub
outputs and effects of web services. Matching based on classes of the input & output types with relaxed
preconditions or complex SWRL expressions are not matching strength to test the mediation capabilities of
implemented. Dynamic composition of web services the broker.
by the broker was limited to sequences of services. The system worked reliably and predicatively for all
However, providers can design abstract composite the services in our sample scenario. Figure 10 shows a
services using any of the control structures available in composite service generated and subsequently
the OWL-S specification and register them with the executed by the broker. However, the sample scenario
broker. The broker will match the slots in the contained only half a dozen of services and about a
composite service with concrete atomic services. dozen of data types. The importance of using effects to
We also developed a prototype client component to describe services will increase as the number of
send requests to the broker and display the responses. registered services rises since more IO based matches
Figure 9 shows this client GUI application. It has five would result in a large search space.
main tabs, three tabs aiding the user to formulate the
request and two tabs to display the response sent back
by the broker. “Inputs” tab allows the user to define
input types and actual values associated with these
types, while “Outputs” and “Effects” tabs let the user
define output and effect types respectively. The user

21

7.1. Future Work
The importance of OWL-S to semantic web services
is analogues to the importance of WSDL to the
syntactic (conventional) web services. The ability of
the current WS frameworks to automatically generate
WSDL for their web services is a key factor for the
popularity of the current web services. Humans
intervention will be needed only choose the OWL
types for WSDL complex types and to define
preconditions and effects. All of these can be done
through an easy to use GUI without requiring the user
Figure 10: A composite service generated to ever look at the WSDL or OWL-S files directly.
dynamically by the broker OWL-S precondition and effects (called post-
conditions as well) are currently based on SWRL
7. Conclusion expressions and hard to be understood by conventional
This project was concerned with using semantic procedural programmers (let along expecting them to
web related technologies to dynamically select and write such expressions). However even the W3C
compose web services to satisfy client requests and activity group is still not sure of what language to be
finally execute these web services to return the result to used to define preconditions and effects and it is a still
the client. Even if semantic web is out of the research wide open research area to select a suitable candidate.
labs, semantic web services are not. We used the W3C However, as soon as the research community and the
recommended semantic web services language called W3C come to a conclusion of what rule language to be
OWL-S to semantically annotate our web services. used for OWL-S preconditions and effects, and as tool
OWL-S specification is still under the R&D stages and support for manipulating rule expressions is available,
going through rigorous changes as new features are this functionality can be integrated to the IntelliBroker.
added and existing features are altered. Even if OWL-S Due to the time constraints, we did not implement
specification as it currently stands is very rich in the multiple broker interaction functionality. However,
describing the web services, reliable third party library in a distributed environment like internet (where
support for manipulating these semantic descriptions failures are common) it makes sense to have number of
are not available as of yet. Another consideration is small service brokers interacting with each other rather
that using these rich descriptions in general contexts than having one single broker creating a single point of
such as public web is still not very practical as many failure. Since the DUBIP itself is based on semantic
web-based agents are still not capable to reading these web languages, rather than syntactic fluff (like HTTP
specifications. The broker we have implemented in this or SOAP) it is inherently extendable and this
project is concerned with some core features of the introduces us to the new concept of semantic protocols.
OWL-S specification and is suitable for a specific
context or a domain, where the available core OWL-S Acknowledgement
features can be used in an intuitive manner to obtain
predictable results (a mandatory requirement for any We want to thank all of the people who have
enterprise scale application). In that sense, this project contributed with solutions to the OWL-S mailing list.
works as an ambitious effort to bring the newest We want to thank Harshana Liyanage at Zone24x7 Inc
developments in the academia through to the industry. for providing us with insights into semantic web.
Finally yet importantly, we wish to thank all who have
Another important fact is that when someone in the contributed to the progress and improvement of
industry wants to do something related to semantics semantic web technologies, which has inspired us carry
(for example in Enterprise Application Integration) on this project from start to end.
they always go for their own proprietary/custom
formats rather than going with available standard
formats, due to the complexity and generality of these References
formats and the performance considerations of 1. ARIBA. 2000. UDDI Overview. Fujitsu Limited.
implementing these full standards. We think this
project keeps an important step towards identifying a 2. CERAMI, E. 2002. WSDL Essentials. Web Services
Essentials 6. O'Reilly & Associates.
useable set of core features to be used in industry
oriented, semantic web service related applications.

22

3. FRIES, B., AND KLUSCH, M. 2006. OWLS-MX, Hybrid 13. MSDN. 2003. .NET Web Services. MSDN
Semantic Web Service Matchmaker, Version 1.1b. Documentation. Microsoft Cooperation.
Users Manual.
14. PALMER, S. B. 2001. The Semantic Web: An
4. GERBER, A., BUTT, S., AND KLUSCH, M. 2006. OWL-S Introduction. Internet. Retrieved from
XPlan, OWL-S Semantic Web Service Composition http://infomesh.net/2001/swintro on 2nd September
Planner. DFKI Saarbrücken, Germany. 2005.
5. GUNATILAKA, M., WIKRAMANAYAKE, G. N., AND
KARUNARATNA, D.D. 2004. Implementation of 15. PAOLUCCI, M., AND SYCARA, K. 2002. DAML-S:

Ontology Based Business Registries to Support e- Semantic Markup for Web Services. The DAML
Commerce. University of Colombo School of Services Coalition.
Computing. Proceedings of 6th International 16. PAOLUCCI, M., SOUDRY, J., SRINIVASAN, N., AND SYCARA,
Information Technology Conference, Colombo, Sri K. 2004. A Broker for OWL-S Web services. The
Lanka, 29th Nov – 1st Dec 2004, pp. 222-231. Robotics Institute, Carnegie Mellon University.
6. HERMAN, I. 2005. Tutorial on Semantic Web 17. ROMAN, D., ZAREMBA, M. 2005. Semantic Web services
Technologies. WWW Consortium. for Autonomic Computing. Second IEEE International
7. JENA 2.2 DOCUMENTATION. 2005. RDQL Examples, HP Conference on Autonomic Computing, Seattle, USA.
Research Labs. June 2005.

8. JIAN, S., SUI, X., AND WANG, B. 2004. Semantic Web 18. SMITH, M. K., WELTY, C.,AND MCGUINNESS, D. 2004.

Services. OWL Web Ontology Language Guide. W3C
Recommendation 10 February 2004.
9. KLUSCH, M., GERBER, A., AND SCHMIDT, M. 2005.
Semantic Web Service Composition Planning with 19. SYCARA, K., FRIES, B., AND KLUSCH, M. 2006.

OWLS-XPlan. Proceedings of the 1st Intl. AAAI Fall Automated Semantic Web Service Discovery with
Symposium on Agents and the Semantic Web, OWLS-MX. Proceedings of 5th International
Arlington VA, USA, AAAI Press, 2005 Conference on Autonomous Agents and Multi-Agent
Systems (AAMAS), Hakodate, Japan, ACM Press
10. KOIVUNEN, M., AND MILLER, E. 2001. W3C Semantic
Web Activity. In Proceedings of Semantic Web Kick-off 20. VASUDEVAN, V. 2001. A Web Services Primer. Internet.
Seminar 2001. WWW Consortium. Retrieved from
http://webservices.xml.com/pub/a/ws/2001/04/04/webser
11. LEE, T. B., HENDLER, J., AND LASSILA, O. 2001. vices on 01st September 2005.
Semantic Web. Scientific American (May).
12. MANOLA, F., AND MILLER, E. 2004. RDF Primer. W3C
Recommendation 10 February 2004. World Wide Web
Consortium.

23

Improving Your Web Services Thorough Semantic Web Techniques

More Related Content

What's hot (20)

Similar to Improving Your Web Services Thorough Semantic Web Techniques (20)

More from Gihan Wikramanayake (20)

Recently uploaded (20)

Improving Your Web Services Thorough Semantic Web Techniques