HARMONY - A FRAMEWORK FOR AUTOMATIC WEB

SERVICE COMPOSITION

Viorica R. Chifu, Ioan Salomie

Department of Computer Science, Technical University of Cluj-Napoca, Romania

Emil Şt. Chifu, Constantin Pârţac

Department of Computer Science, Technical University of Cluj-Napoca, Romania

Keywords: Web service, Web service composition, semantic Web, ontology.

Abstract: Web services are software components that were designed to improve interoperability and integration of

applications developed on different platforms. Web Service composition offers the facility to create new

services out of the existing services satisfying a complex functionality. This paper presents HARMONY, a

framework for automatic Web service composition. Our approach for automatic Web service composition is

based on the GraphPlan algorithm. In HARMONY we use ontologies for the semantic annotation of Web

services, so that the automatic service discovery, composition and execution can be realized based on

ontology inference.

1 INTRODUCTION

Web services provide a standard way to ensure the

interoperability among different software

applications running on a variety of platforms. The

current standard technologies for Web services

provide descriptions only at the syntactic level of

their functionality, without any formal description of

their semantics. This drawback prevents the use of

Web services in complex business contexts, where

the automation of these business processes is

necessary. Semantic Web Services (Akkiraju, 2005)

(Lausen, 2005) enhance WS standards by annotating

services with semantic descriptions provided by

ontologies.

This paper presents HARMONY, a framework

for automatic Web service composition and

execution based on ontologies. Our approach for

automatic Web service composition is based on the

GraphPlan (Blum and Furst, 1995) algorithm. The

paper is organized as follows. Section 2 presents the

ontology model. The framework architecture and

implementation is briefly described in section 3,

Conclusions and future directions are presented in

section 4.

2 ONTOLOGY

Our ontology model contains classes, individuals

and properties. The classes are concrete

representations of domain concepts. A property

either defines a relation between concepts or a

restriction. There are two types of relations in our

ontology model: hierarchical relations and non-

hierarchical relations. The hierarchical relations are

taxonomic relations while the non-hierarchical

relations are relating concepts across the hierarchical

structure. Restrictions describe a class of individuals

and possibly a number of relationships that they

participate in. In our ontology model we have

defined existential restrictions. An existential

restriction describes the class of individuals that are

in relationship with at least one individual member

of another class.

Three main generic classes can be identified as

the core of our model: WebService, Message and

WebServiceRestriction. WebService class is the root

class of the service classification tree. The Message

tree has two generic classes of concepts: Request

and Response, which are classifications of the inputs

and outputs of the services respectively. Finally, the

WebServiceRestriction tree is a classification of the

effects and preconditions of the services. The

240

R. Chifu V., Salomie I., ¸St. Chifu E. and Pâr¸tac C. (2008).

HARMONY - A FRAMEWORK FOR AUTOMATIC WEB SERVICE COMPOSITION.

In Proceedings of the Fourth International Conference on Web Information Systems and Technologies, pages 240-243

DOI: 10.5220/0001520402400243

 SciTePress

generic properties of the Web services which are

taken into account by our model are the following:

(i) the endpoint as a data type property indicating

the address at which the service can be invoked; (ii)

the input as an object property representing a

request message as an output from another service;

(iii) the output as an object property representing a

response message; (iv) precondition and effect as

object properties representing conditions on the

information space before and after the services are

executed; (v) description as a data type property

representing the description of a Web service. Based

on this ontology model we have developed an

ontology for an online bookstore using the Protégé

OWL editor (Horridge, 2004) (Figure 1). This

ontology is used for semantic annotation of web

services, as a vocabulary for the graphical user

interface and for a “smart” planner. In our work, we

use SAWSDL (Verma, 2007) for the semantic

annotation of Web services. Because at this moment

there isn’t any final decision about representing the

preconditions and effects in SAWSDL, we have

chosen to extend SAWSDL with WSDL-S

(Akkiraju, 2005) schema, which provides a way of

representing the preconditions and effects.

Figure 1: Main OWL ontology classes.

3 FRAMEWORK

ARCHITECTURE

HARMONY is an experimental framework for

automatic Web service composition created to ease

the process of composition, thereby reducing the

complexity and the development time of a composite

Web service. HARMONY components were

designed to be independent of each other. An

overview of the HARMONY architecture is

presented in Figure 4, where the arrow connections

represent the data flow.

Figure 2: Framework architecture.

The UDDI Publisher takes the WSDL as input and

creates its corresponding tModel, bindingTemplates

and businessServices into the UDDI server. As

UDDI server we use jUDDI provided by the Apache

Foundation (jUDDI, 2007). The UDDI Inquirer

creates a cache of all the registry entries. This way,

inside the UDDI Inquirer cache, we create the

mappings of the inputs, outputs, preconditions and

effects of Web services to ontology classes. The

Ontology Manger consists of the Jena API

framework (Carroll, 2004) and a cache used to store

the ontology model. The GUI interacts with the Web

Service Composer and Invoker modules. The user

only needs to select by the GUI the inputs, outputs,

preconditions and effects (which are concept classes

from the ontology) of the desired services and to

HARMONY - A FRAMEWORK FOR AUTOMATIC WEB SERVICE COMPOSITION

241

invoke Web Service Composer. After the planner

finds the solution, it should be evaluated and

validated, and then Web Service Invoker is called in

order to execute the newly composed Web Service.

A more detailed description of each component is

presented in what follows.

3.1 Ontology Manager

The Ontology Manger uses the Jena Ontology API

(Carroll, 2004). The main drawback of Jena resides

in its speed inefficiency. In order to improve the

performance, our Ontology Manger has a caching

mechanism which stores the ontology model

between Jena interface calls. It allows for a

significant speed improvement mainly because the

same model is used by all of the framework modules

interacting with the ontology. If a new rule needs to

be inferred, the average loading time of the ontology

is between 2 and 5 seconds. By using the Jena

caching mechanism, a retrieval of the ontology

model takes approximatively 250 ms, but by using

our Ontology Manager caching mechanism it takes

up to 3 ms.

3.2 UDDI Publisher

A simple method for publishing Web services was

implemented in order to add a high degree of

automation to the publishing process. In our

approach we publish a service by simply providing

the SAWSDL files to our UDDI Publisher. The

UDDI Publisher consists of two modules: Interface

Publisher and Service Publisher. Interface Publisher

is responsible for publishing Service interfaces

which are mapped into UDDI tModels. The Service

Publisher is responsible for publishing the Service

implementation which is mapped into

businessServices and bindingTemplates.

3.3 UDDI Inquirer

There are remarkable implementation efforts in the

area of service composition frameworks in general

and in UDDI query in particular. The main

disadvantage of previous approaches (Châtel, 2006

and Verma, 2006) is that they are constrained to

using UDDI defined inquiry facilities. We took an

alternative approach by defining a separate UDDI

inquirer tool. A cache entry is defined for each

bindingTemplate, which corresponds to a Web

Service operation. In order to generate the cache

entry, the SAWSDL corresponding to a

bindingTemplate is parsed and the URIs of the

ontology classes representing the input, output,

preconditions and effects are extracted. Then, a call

to Ontology Manager is made in order to include the

actual ontology classes in the cache.

3.4 Web Service Composer

Our Web Service Composer implements the

GraphPlan algorithm, which takes into account the

inputs, outputs, preconditions and effects. The

GraphPlan algorithm takes into consideration only

the viable solutions. The GraphPlan algorithm finds

such viable solutions by evaluating at each state

whether the preconditions necessary to run a service

are met and whether all the inputs are present. The

java class implementing the GraphPlan algorithm

finds all the goals situated on the same depth in the

graph, but it can be configured to find the solution

until a certain depth is achieved or until it finds a

certain number of solutions. In order to be able to

reuse the solutions, they must be saved in an easily

serializable / deserializable format. The framework

saves the solutions provided by GraphPlanner in

XML format in a repository of reusable complex

services.

3.5 Invoker

The composite service invoker takes as input the list

of Web services to be invoked and the values to be

passed as inputs and generates an Axis engine client

for each call. The invoker parses the SAWSDL files

in order to configure the clients and assigns a

serializer for each input and a deserializer for each

output. Then, it constructs the java bean classes

corresponding to each input of the first service. The

result of the first service is automatically converted

by the Axis engine to a java bean. Then, the second

service is invoked, and so on, until the last one. In

this execution scenario, a java bean must be

generated for each concept class described in the

ontology.

3.6 GUI

The graphical user interface is composed out of two

functional parts: the composer GUI and the invoker

GUI. In the former, the user selects the ontology

whose concepts he wants to use in the automatic

composition of Web Services, followed by selecting

the inputs, outputs, preconditions and effects of the

desired service. The inputs, outputs, preconditions

and effects are chosen from the previously selected

ontology classes. After invoking the service

WEBIST 2008 - International Conference on Web Information Systems and Technologies

242

composer, the invocation interface is presented,

where the available solutions are shown, allowing

the user to select the input values for the composed

service.

4 CONCLUSIONS

In this paper we have presented HARMONY, a

framework for automatic Web service composition

based on ontologies. In our approach, the ontology is

used in all the steps of the automatic composition

process: for the semantic annotation of Web

services, as a vocabulary for the graphical user

interface, and for implementing a “smart” planner by

using semantics. The proposed solution is rather

generic, and could be used in different contexts.

Framework components were designed to be

independent of each other, so that we can add /

replace framework components without major

modifications. The user interface is simple, but at

the same time powerful and intuitive. It drives the

user in the composition process in a friendly manner

by providing a controlled language that uses the

ontology concepts. The only task for the user when

he desires a new composed Web service is to change

the specification, i.e. the input, output, preconditions

and effects. In future work, we plan to extend our

framework with a QoS module in order to allow for

dynamic selection of the best solution from the set of

solutions generated by the planner. Another

important effort will be directed towards the addition

of heterogeneity handling. This would allow using

semantically compatible concepts from different

ontologies.

ACKNOWLEDGEMENTS

This work was supported by the PI-1020 (333/2007)

project within the framework of the “Research of

Excellence” program initiated by the Romanian

Ministry of Education and Research.

REFERENCES

Akkiraju, R., Farrell, J., Miller, J., Nagarajan, M.,

Schmidt, M.-T., Sheth, A., Verma, K., 2005,

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

Bechhofer, S., van Harmelen, F., et al., 2004. OWL web

ontology language reference. W3C recommendation,

M. Dean, G. Schreiber (eds.).

Blum, A. and Furst, M., 1995. Fast planning through

planning graph analysis. In Proceedings of the 14th

International Joint Conference on Artificial

Intelligence (IJCAI 95), pp. 1636-1642.

Cardoso, J., Sheth, A., 2003. Semantic e-Workflow

Composition. Journal of Intelligent Inform. System,

21(3): pp. 191-225.

Carroll, J., Dickinson, I., Dollin, D., Reynolds, D.,

Seaborne, A., Wilkinson, A., 2004. Jena:

Implementing the Semantic Web Recommendations.

Proceedings of the 13

World Wide Web conference

on Alternate track papers & posters, New York, NY,

USA, pp. 74-83

Châtel, P., 2006. WSDL 2.0 to UDDI mapping SAWSDL

to UDDI mapping, Thales Group.

Farshad, H., et al., 2005. Semantic Web Service

Composition in IRS-III: The Structured Approach, in

Proc. of the 7 IEEE Intl Conf on Ecommerce

Technology (CEC'05).

Haarslev, V., Möller, R., 2003. Racer: An OWL reasoning

agent for the Semantic Web. Proceedings of the

International Workshop on Applications, Products and

Services of Web-based Support Systems, pp: 91-95.

Lausen, H., Polleres, A., Roman, D., (Eds), 2005. Web

Service Modeling Ontology (WSMO). W3C Member

Submission, http://www.w3.org/Submission/WSMO/.

McIlraith, S. and Son, T., 2002. Adapting Golog for

Composition of Semantic Web Services. In the

Proceedings of the Eighth International Conference

on Knowledge Representation and Reasoning

(KR2002).

Roman, D., et. al. , 2005. Web Service Modeling

Ontology. Applied Ontology 1 (2005) pp.77–106

Sirin, E., et al., 2005. Template-based Composition of

Semantic Web Services, AAAI Fall Symposium on

Agents and the Semantic Web, Virginia.

Sirin, E., Parsia, B., Grau, B.C., Kalyanpur, A., and Katz

Y.,2006. Pellet: A Practical OWL-DL reasoner. In the

Journal of Web Semantics.

Verma, K., Gomadam, K., Sheth, A. P., Miller A.J., Wu,

Z., 2005. The METEOR-S Approach for Configuring

and Executing Dynamic Web Processes, Technical

Report.

Verma, K., 2006. Configuration and Adaptation of

Semantic Web Processes. PhD thesis, Department of

Computer Science, University of Georgia.

Verma, K., Sheth, A. P, 2007. Semantically Annotating a

Web Service, IEEE Internet Computing, pp. 83-85B.

Horridge, M., Knublauch, H., et al., 2004. A practical

guide to building OWL ontologies using the Protege-

OWL plugin and CO-ODE Tools. The University Of

Manchester.

Wu, Z., Ranabahu, A., Gomadam, K., Sheth, A.P., Miller,

J.A.,2007. Automatic Composition of Semantic Web

Services using Process and Data Mediation, LSDIS

lab, University of Georgia.

jUDDI, http://ws.apache.org/juddi/, accessed June 2007

HARMONY - A FRAMEWORK FOR AUTOMATIC WEB SERVICE COMPOSITION

243