SEMANTIC SERVICE DESCRIPTION TO SUPPORT

EDUCATION AND TRAINING

An Easy-to-Apply Service Description Model and Service Registry

Karsten Krutz, Tilmann Kuhn, Christophe Gnad and Sebastian Abeck

Institute for Telematics, University of Karlsruhe, Zirkel 2, Karlsruhe, Germany

Keywords: Semantic service description, service registry, education and training, semantic web.

Abstract: This work describes a semantic service description model based on and inspired by existing service

description approaches like WSDL-S and OWL-S and a service registry to support the creation of service

oriented applications in the area of education and training. Our service description is based on an OWL

ontology and combines a simple message- and state-oriented approach in order to reduce the effort for the

creation of the descriptions which is a major drawback of a lot of existing, powerful semantic service

description approaches. Furthermore, the proposed service description enables the matchmaking algorithm

of the service registry to use not only the data-oriented input and output parameters to identify suitable

services but also considers the preconditions and effects of a service. The usage of the service description

model and the search capabilities are discussed against the background of the education and training area.

1 INTRODUCTION

The importance of service-oriented architectures

increased particularly with the availability of web

services as a platform independent technology.

Especially in scenarios in which different business

units have to cooperate in order to achieve a

common purpose, the service-oriented approach is

well suited and can deliver significant benefits

(Adam et al. 2005). In the area of education and

training, scenarios of this kind are very common.

E.g. the teaching unit has to cooperate with the

course material authors to get appropriate material,

with the IT provider to get the technical equipment

needed for each lecture and with the administration

to arrange the registration of learners for

examinations and to capture the examination results.

As the teachers apply very different teaching

approaches their specific need for IT support is

much diversified. A service-oriented approach can

ease the implementation of the needed IT support by

increasing the reuse of existing functionality in the

area of education and training (Westerkamp 2006).

To support the reuse of existing services a

service registry is needed that enables the users to

define their needs and to get a list of available

services that fulfil these needs (at least partially). To

enable this search, the service registry needs to be

based on suitable service descriptions.

The two main contributions of this paper are:

A new service description model (Chapter 3) that

eliminates deficiencies of the existing approaches

(Chapter 2) and a service registry architecture and its

implementation (Chapter 4) that allows to efficiently

search for services based on our description.

2 RELATED WORK

The semantic service description should at least

cover input and output of the service and two sets of

conditions: the preconditions which have to be met

before executing the service properly and the effects

that are conditions that hold after the successful

execution of the service (Jaeger et al. 2005). These

four service description elements are often referred

to as IOPE (Input, Output, Precondition and Effect).

Current standards from the web service area like

UDDI and WSDL can form the basis for a service

description and retrieval solution but are not

sufficient. A WSDL description provides the

information that is needed to invoke a web service in

a syntactically correct manner and can therefore be

used for the (technical) integration of available

services into the IT solution. WSDL does not

433

Krutz K., Kuhn T., Gnad C. and Abeck S. (2007).

SEMANTIC SERVICE DESCRIPTION TO SUPPORT EDUCATION AND TRAINING - An Easy-to-Apply Service Description Model and Service Registry.

In Proceedings of the Third International Conference on Web Information Systems and Technologies - Web Interfaces and Applications, pages 433-436

DOI: 10.5220/0001283904330436

 SciTePress

provide a standardised way to describe precondition

and effect of a service call and the semantics of the

terms used to describe input and output of the

service can only be included in the description by

using appropriate terminology.

The explicit semantic annotation of the services

is a very important step towards the exploitation of

the full potential of the service-oriented approach

(Patil et al. 2004). Different approaches try to

provide this by adding semantics to conventional

service descriptions like WSDL-S (Akkiraju et al.

2005) or by creating completely new descriptions

like OWL-S (Martin et al. 2004) or Diane Service

Description DSD (Klein 2004). All mentioned

service descriptions include the possibility of

defining all four IOPE service description elements.

Approaches like OWL-S have the deficit, that it

is not easy for the description author to create such a

quite complex description (Patil et al. 2004). This

problem could be reduced by using less complex

description languages that have adequate

expressiveness for certain scenarios.

Service description approaches like WSDL-S

reduce this description creation problem as well, as

they contain less new description elements. It adds

the two missing service description elements

precondition and effect and enables linking to an

ontology to enrich all four service description

elements with semantics. Unfortunately WSDL-S

neither provides an ontology nor gives any

information how the referenced ontology should

look like.

Generally all four types of service description

elements can be used to identify suitable services.

But in many approaches preconditions and effects

are not considered by the matchmaking algorithm as

they are not sufficiently standardised (Jaeger et al.

2005). This leads obviously to less powerful search

capabilities.

The DSD solves this problem by treating the two

condition description elements precondition and

effect the same way as the two data description

elements input and output, thus enabling the

matchmaking algorithm to use all four IOPE

elements for the search (Klein et al. 2005). The

conditions are modelled as states that are represented

in the ontology by concepts inheriting from the top-

level concept "state" (Klein 2004). The DSD uses a

state-oriented service description, which is less

common and intuitive than a combined message-

and state-oriented service description like OWL-S.

The DSD ontology language is proprietary which

reduces the reuse capabilities in other scenarios.

3 SERVICE DESCRIPTION

The service description needs to support the search

of existing services and the integration of the found

services. To support these requirements we propose

a description model that consists of two parts: A

service profile supporting the semantic search and

the service grounding supporting the technical

integration of a found service.

The description of the service profile is based on

a set of ontologies as shown in Figure 1. The

composition of ontologies has some similarities with

the one proposed by (Klein and König-Ries 2003).

Top-Level Ontology

Service

Ontology

Domain

Ontology

State

Ontology

Service

Ontology

Domain

Ontology

State

Ontology

Thing

entity

State

Entity

hasState

Value

Figure 1: Service Description Ontologies.

An excerpt of the very simple top ontology is

shown in Figure 1. The concept "state" is later used

to define a services preconditions and effects while

"entity" and "value" are used for input and output.

Thing

State

ServiceProfile

Entity

String

textDescription

Figure 2: Service Profile.

The service profile is situated in the service

ontology. Its input and output parameters have

references to concepts in the domain ontology, while

preconditions and effects are defined by references

on state concepts in the state ontology. Furthermore

the service profile contains a textual description of

the service operation that can be used by a human

reader to finally decide whether a service fits his

needs or not. The dashed boxes represent concepts

that are defined in another ontology (the top-level

ontology).

The domain ontology and the state ontology,

contain the specifics of the domain of education and

training as shown in Figure 3.

WEBIST 2007 - International Conference on Web Information Systems and Technologies

434

LearningObject

Entity

CourseMaterial

DeliverySystem

State

entity hasState

LDoc

MultiMediaObject

Published

PublishingArea

Domain Ontology State Ontology

Transmitted

Faxed

Mailed

Figure 3: Extracts of the Domain and State Ontologies.

The service which will be used as an example for

a semantic service description has the functionality

to publish a specific multimedia course material

called Living Document (LDoc) to a specific

publishing area. The service profile of the example

service "LDocPublishingService" has two input

parameters referenced as "logicalInput" and as effect

the state "LDocPublished", which is a subclass of

the state "Published" with the restriction that the

entity needs to be an "LDoc". The technical

representations of the inputs are linked using the

concept "Represented". The service ensures that a

given LDoc is published in a given publishing area

after the execution of the service.

Service Profile

LDoc

PublishingArea

effect

entity

Represented

PublishingAreaId

represented

Represented

LdocReference

represented

location

logicalInput

LdocPublished

LdocPublishingService

Figure 4: Example Service Profile.

The link from the technical representations of the

input parameters to the interface description which is

called grounding is realized using WSDL-S. In

WSDL-S the specification of the structure of

message objects are enriched by a semantic

annotation using a so called "modelReference" as

link towards the ontology.

4 SERVICE REGISTRY

In order to store and search for the service profiles

and the service groundings a service registry was

created based on existing software components.

The service registry is divided into a server and a

client part. The server contains a UDDI registry

which is used to register the WSDL-S descriptions.

Furthermore the server provides access and

reasoning functionality to the four OWL ontologies

that are needed for the service profile descriptions.

In the service registry server the UDDI

implementation jUDDI was used based on a mySQL

database and a Tomcat J2EE server. The OWL

ontologies were created using Protégé 2.1 as

ontology editor and are provided using an Apache

webserver. As reasoner the product

RacerPro 1.9 was used and the access to the

reasoning functionality was provided using DIG

Description Logic Interface 1.1.

<<device>>

:Service Registry Server

<<execution env>>

:Eclipse

:ServiceRegistry

ClientPlugIn

:ServiceRegistry

ClientPlugIn

<<device>>

:Service Registry Client

HTTP

UDDI/

SOAP

DIG

<<execution env>>

tomcat:J2EEServer

<<web app>>

juddi:UDDIServer

<<database>>

mySQL:SQLServer

top.owl

upperService.owl

serviceProfile.owl

eduDomain.owl

apacheHttp:WebServer

racerPro:DIGReasoner

<<execution env>>

tomcat:J2EEServer

<<web app>>

juddi:UDDIServer

<<execution env>>

tomcat:J2EEServer

<<web app>>

juddi:UDDIServer

<<web app>>

juddi:UDDIServer

<<database>>

mySQL:SQLServer

top.owl

upperService.owl

serviceProfile.owl

eduDomain.owl

apacheHttp:WebServer

top.owl

upperService.owl

serviceProfile.owl

eduDomain.owl

apacheHttp:WebServer

racerPro:DIGReasoner

Figure 5: Deployment Diagram of the Service Registry.

On the client side a GUI interface realized as

Eclipse plug-in enables a user to enter his queries

and get access to search results.

The retrieval process that is supported by the

service registry starts with a user's need for a

service. The user tries to represent his need using the

query syntax that is supported by the retrieval

process (Stojanovic et al. 2003). In the retrieval step,

the query is matched against the repository using a

retrieval model such as Boolean or probabilistic

model (Baeza-Yates and Ribeiro-Neto 1999). The

output of this step is a list of service instances.

Information Retrieval

Conceptualization

Information

Retrieval

Need for a

Service

Profile

Query

Service

Profiles

Service

Grounding

Query

generation

Service Instance

Information

Retrieval

Service

Instances

Service Profile

Information

Retrieval

Figure 6: Semantic Service Retrieval Process.

SEMANTIC SERVICE DESCRIPTION TO SUPPORT EDUCATION AND TRAINING - An Easy-to-Apply Service

Description Model and Service Registry

435

The retrieval step starts with a service profile

search query created by the user. This query is

defined on a quite abstract (business-)level but can

also contain technical restrictions. The first activity

generates a list of fitting service profiles that

represent types of services that meet the needs

defined in the query. One very important feature is

the query refinement which is used to get optimized

queries for recall and precision enabling the user to

get more results that are potentially relevant to him.

In the second activity a list of service instances is

generated that provide the functionality defined in

the service profiles of the first retrieval's result list.

5 DISCUSSION

If a specific publishing service like the one in Figure

4 is needed, the corresponding query will probably

use a specific concept as input. The service

repository can find not only exact matches but will

use the hierarchic structure of the domain ontology

and might find a compatible service with a super

class as input. This usage of an inheritance hierarchy

can be applied to increase the recall as much more

potential services descriptions match these queries.

Synonymous terms in search queries and service

descriptions can lead to no search results even if a

matching service would be available. Due to the

common terminology of the domain ontology the

usage of synonymous terms can be prevented.

The possibility to specify the functionality of a

service makes it possible to prevent search results

with matching operation signature that do not

provide the needed functionality.

Compared to OWL-S the proposed service

description is much less complex which can lead to

decreased effort for the creation of service

descriptions and service queries.

The support of the business and the technical

layer enables linking these two terminologies.

6 SUMMARY

The paper has introduced an easy-to-apply semantic

service description for web services and

demonstrated its usage in the area of education and

training services. The introduced service description

is based on a combined message- and state-oriented

approach and the matchmaking algorithm can use all

four IOPE attributes to identify suitable services. As

the approach is based on a domain and a state

ontology it enables query modifications that lead to

higher recall of the search. Compared to OWL-S the

complexity of the service descriptions is low and

reduces the effort required for the creation of service

descriptions and search queries.

Apart from the service description a service

registry based on UDDI and OWL was introduced

that enables publication and search of service

descriptions created using the proposed service

description in an efficient manner.

One of our next steps to enable the description of

a broader area of services targets the refinement of

the currently available domain and state ontologies

for the area of education and training.

REFERENCES

Akkiraju, R. et al., 2005. Web Service Semantics –

WSDL-S, Version 1.0, W3C Member Submission,

[online], November 2005

available at: http://www.w3.org/Submission/WSDL-S.

Adam, O.,Hofer, A., Zang, S., Hammer, C., Jerrentrup,

M., Leinenbach, S., 2005. A Collaboration Framework

for Cross-enterprise Business Process Management. In

INTEROP-ESA '05, First International Conference on

Interoperability of Enterprise Software and

Applications. Springer

Baeza-Yates, R., Ribeiro-Neto, B., 1999. Modern

Information Retrieval, Addison Wesley.

Jaeger, M., Engel, L., Geihs, K., 2005, A Methodology for

Developing OWL-S Descriptions. In INTEROP-ESA

'05, First International Conference on Interoperability

of Enterprise Software and Applications Workshop on

Web Services and Interoperability. Springer.

Klein, M., König-Ries, B., 2003. A Process and a Tool for

Creating Service Descriptions Based on DAML-S. In

TES'03, 4th VLDB Workshop on Technologies for E-

Services. Springer.

Klein, M., König-Ries, B., Müssig, M., 2005. What is

Needed for Semantic Service Descriptions? – A

Proposal for Suitable Language Constructs. In

International Journal of Web and Grid Services 2005,

Volume 1 (Issue 3) pp. 328-364, Inderscience

Publishers.

Klein, M., 2004. Manual for the DIANE Service

Description, Technical Report TR 2004-17, ISSN

1432-7864.

Martin, D. et al., 2004. OWL-S: Semantic Markup for

Web Services, W3C Member Submission, [online],

November 2004,

available at: http://www.w3.org/Submission/OWL-S.

Patil, A., Oundhakar, S., Sheth, A., Verma, K., 2004.

METEOR-S Web service Annotation Framework. In

13th conference on World Wide Web. ACM Press.

Stojanovic, N., Studer, R., Stojanovic, L., 2003. An

Approach for the Ranking of Query Results in the

Semantic Web. In ISWC 2003, 2

International

Semantic Web Conference, Springer.

Westerkamp, P., 2006. Flexible Elearning Plattforms: A

Service-Oriented Approach, Logos Verlag, Berlin.

WEBIST 2007 - International Conference on Web Information Systems and Technologies

436