ONTOLOGY GATEWAY

Enabling Interoperability between FIPA Complaint Agents

and OWL Web Services

Sabih Ur Rehman, Maruf Pasha, Arshad Ali

NUST Institute of Information Technology, 166-A, Street 9, Chaklala Scheme 3, Rawal - pindi, Pakistan

Farooq Ahmed, Hiroki Suguri

Communication Technologies, 2-15-28 Omachi Aoba-ku, Sendai 980-0804, Japan

Keywords: Semantics, Agents, Grid Computing.

Abstract: The introduction of Ontology Web Language (OWL) which is a W3C standard for providing explicit

semantics for establishing and sharing ontologies on the World Wide Web, has made it easier to embed

semantics with web data. Similarly FIPA Semantic Language is the core of the agent platforms due to its

high expressive power. Ontology plays an important role in the knowledge representation, reuse and

communication between web services. Similarly in an Multi-agent system ontology also plays an important

role, where the messages exchanged between agents should conform to an Ontology so that they could be

understood. In this paper we will introduce a technology enabling bidirectional interoperability between

FIPA compliant software agents and the Web services published in OWL. This is an extension of previous

work in which we proposed the development of semantic translations in such a way that the agents can

communicate with web services in an efficient manner. We will also describe and discuss the

implementation in which a FIPA complaint software agent will invoke and use a web service published in

OWL. Our goal for this paper is to show how a FIPA complaint agent can invoke and use the web services

published on the World Wide Web by the help of ontological transformations.

1 INTRODUCTION

The semantic web (T. Berners-Lee, 2004) field

within the W3C has grown rapidly during the last

few years. The semantic web is aimed to unite the

existing web information in order to make it easier

for the third parties to share and reuse. Formal

taxonomies, ontologies and the available web

standards such as RDF, XML, OWL, DAML-S, etc.

allow the web application developers to specify the

information in a standard way, to enable reusability

and sharing. The semantic web aims at providing

more intelligent access and management of the Web

information and more semantically richer modeling

of applications and their users. The web can reach its

full potential if the information available on the web

can be processed both by the automated tools as well

as the people.

Ontology (D. Fensel, 2003) is a specification of a

conceptualization (formal taxonomy). It provides a

vocabulary that describes a domain of interest and

the meaning to the terms being used. Through

ontologies we can produce information that is

human as well as machine understandable. In fact

the ontologies from the core of the semantic web and

enable automated interoperation and cooperation.

Computer can just read the information currently

available on the World Wide Web and cannot

interpret it. The growth of the semantic web will

give rise to such information that is structured and

understandable by the computers. It will be an

extension of the existing World Wide Web in which

the information will be given a well defined

meaning and better-enabled computers

(T. Berners-

Lee, 2004

). The semantic web will enable efficient

discovery and better utilization of the resources on

the web. W3C and several other research groups are

309

Ur Rehman S., Pasha M., Ali A., Ahmed F. and Suguri H. (2007).

ONTOLOGY GATEWAY - Enabling Interoperability between FIPA Complaint Agents and OWL Web Services.

In Proceedings of the Ninth International Conference on Enterprise Information Systems - SAIC, pages 309-314

DOI: 10.5220/0002410203090314

 SciTePress

working to provide a formal language for defining

ontologies and explicit semantics for the web

resources. Therefore Ontology Web Language

(OWL)

(M. Dean, 2004) has been proposed by the

W3C. OWL can be used to describe the classes and

relations between them. It is capable of defining and

instantiating Web ontologies and has overcome

many of the deficiencies that are present in other

ontological web languages.(overhead)

On the other hand Grid computing (IBM Grid

Computing Homepage) has emerged as an important

field; it provides a hardware and software

infrastructure that enables dependable, consistent,

persuasive and inexpensive access to computational

capabilities and coordinated resource sharing. The

semantic grid is the extension of the current Grid,

with the exception that in the semantic grid

information and services are given well defined

meanings and enables autonomous interoperation

between entities (M. Wooldrige, 1997). It is similar

to the Semantic Web in a way that it uses metadata

to describe information in the grid. Turning

information into more than just simple information,

enabling computers not only to read the data but also

understand the information presented. The key idea

behind semantic grid is to enable software agents to

dynamically discover, invoke and monitor the

semantic enabled web services.

An agent platform is an infrastructure through

which agents can be deployed. A typical agent

platform such as Foundation for Intelligent Physical

Agents (FIPA) complaint agent platform, Scalable

fault tolerant Agent Grooming Environment (SAGE)

(Abdul Ghafoor, 2004) provide services like

creating a multi agent application that uses services

provided by the agent platform itself and services on

the Grid (Ferber, 1999).

Figure 1: Agent Communication Language.

For agents to communicate with each other they

need a set of instructions and standards on how

agents will carry out effective and meaningful

conversation. For this FIPA has proposed ACL. That

provides semantics and precise syntax descriptions

for messages to be exchanged between the agents.

In cases where the service or resource required by an

agent to accomplish a task is not available on the

Grid on which the agent platform is running, in such

cases the agent will have to leave the work

incomplete. We propose an Ontology Gateway that

will enable an agent to consume services available

on the web. In cases described above if an agent is

unable to perform a task due to shortage or

unavailability of a service it can then search for that

service on the World Wide Web. The agent can then

consume or make use of a Web service published in

OWL to complete its task. We will enable such

communication through ontological transformations

in the Ontology Gateway. The architecture of the

proposed Ontology Gateway will be explained in

section 2.

2 PROPOSED ARCHITECTURE

The architecture we proposed in (Maruf Pasha,

2006) provided the semantic operability in

distributed environments where technologies like

agent applications and grid systems are combined

and reused to provide a Semantic Grid and thus

providing a service oriented framework.

In it the OWL that is a W3C standard for

specifying the services on the web will be used as a

content language. The communication infrastructure

specified by FIPA permits agents to communicate

using any mutually comprehensive content language

as far as it fulfils a few minimal criteria laid down

by the FIPA.

The key idea behind the proposed architecture

was to enable software agents to communicate with

the web services.

We will carry forward the previous architecture

but will enhance it in such a way that it will be able

to cater more OWL and ACL operators converting

them into their appropriate OWL/ACL operators. In

the enhanced Ontology Gateway we have also added

a Natural Language Process (NLP) module. The

NLP module will take the user query as input in the

form of natural language. Then it will convert the

query into an ACL query, which will then be passed

to the ontology gateway that will interpret the ACL

query and will locate the appropriate web service.

The NLP module implements a text mining (J.

Dorre, 1999) algorithm. Text mining is similar to

data mining; data mining seeks to extract data and

meaningful patterns from the data

(Eibe Frank,

ICEIS 2007 - International Conference on Enterprise Information Systems

310

1999

), where as text mining is about looking for

meaningful patterns or data in natural language text.

This module will help the agent in determining the

semantic meaning of a user query given in simple

English and will enable users that are not familiar

with the syntax of FIPA-ACL to generate queries

using simple English, which will then be converted

to an appropriate ACL query.

Figure 2 shows an abstract architecture of the

proposed system. The addition of the NLP module

will also enable users having little knowledge of the

Multi Agent Systems to use the Ontology Gateway

and generate ACL queries. The key idea behind the

proposed architecture is to show how software

agents communicate with the OWL based web

services and to enable users that are not aware of

FIPA syntax to use and generate queries in English

that would then be converted to appropriate ACL

query through the NLP module.

Figure 2: The Proposed Architecture.

3 IN DEPTH ANALYSIS OF FIPA

AND OWL

The main reason for these conversions is that both

FIPA SL and OWL have different principles in

terms of syntax, semantics and implementation

constraints. Therefore devising translations for such

a system is quite demanding and challenging, as

both the languages are quite different from each

other. This section will present some details on how

exactly the conversions in the Ontology Gateway

will take place. In this section we will also present

analysis of ontologies of FIPA

Table 1 shows the mappings of OWL class.

Every class in the OWL world is a member of the

class owl:thing, therefore each user defined class is

by default a sub class of owl:thing, where as in FIPA

Ontologies every class is by default a subclass of the

universal class :THING. Every OWL class provides

an abstraction to group together resources with

similar characteristics therefore they are mapped to

frames in FIPA Ontology, which represents an

entity.

OWL provides a set of complex operator for

defining classes such as owl:intersectionOf,

owl:unionOf, owl:oneOf etc. which are used to

provide explicit semantics with the classes. Our key

focus during the transformations will be to map the

operators in such a way that the underlying

semantics are preserved as much as possible. See

Table 1 for FIPA implementation of the OWL

operators and tags.

Table 1: OWL Class Mappings.

OWL FIPA

Basic owl:Thing :THING

Frame

Class owl:Class

:CLASS

Class rdfs:subClassOf subclass-of

Equivalent

Axioms owl:equivalentClass

:SAME-

VALUES

Individual Individual

Rdfs:domain :DOMAINS

rdfs:range

:SLOT-VALUE-

TYPE

Class owl:restriction Content

Descrip

tion

owl:minCardinality

:MINIMUM-

CARDINALITY

owl:maxCardinality

:MAXIMUM-

CARDINALITY

owl:DatatypeProperty :VALUE-TYPE

owl:intersectionOf

BinaryLogicalOp

= "and"

owl:unionOf

BinaryLogicalOp

= "or"

owl:complementOf

UnaryLogicalOp

= "not"

owl:oneOf set-of

owl: disjointWith

:NOT-SAME-

VALUES

Table 2 shows the Property mapping of Ontology

Web Language. The

owl:objectProperty tag is

used to define relations between instances of two

classes, for example :

Ontology Gateway

With the ability to

Translate and interpret

More operators/tags

Owl To

FIPA SL

To Owl

OWL based

Web

Services

Ontologies

FIPA

Compliant

Multi Agent

System

(SAGE)

Ontologies

Ontology

Gateway agent

NLP

Module

User

Queries

ONTOLOGY GATEWAY - Enabling Interoperability between FIPA complaint Agents and OWL Web Services

311

<owl:ObjectProperty

rdf:ID="fatherOf">

<rdfs:domain

rdf:resource="#Man"/>

<rdfs:range

rdf:resource="#Children"/>

</owl:ObjectProperty>

where as in FIPA frames (classes) has

slots(properties) and each slot of a frame has

associated with it a set of facets, and each facet of a

slot has a set of entities associated(classes) with it,

therefore while translating an OWL Ontology into a

FIPA Ontology

owl:objectProperty will be

mapped onto facets. The data type properties are

basically instances of classes and RDF schema and

XML schema data types. Therefore,

owl:datatypeProperty is used to define a data

type for the property.

Table 2: OWL Property Mappings.

OWL FIPA

Property rdfs:domain :DOMAINS

Associat

ions

rdfs:range

:SLOT-VALUE-

TYPE

owl:equivalentPropert

:SLOT-SAME-

VALUES

Property owl:inverseOf

:SLOT-

INVERSE

Relation

ships

rdfs:subPropertyOf

:SLOT-SUBSET-

OF-VALUES

rdfs:Datatype

:SLOT-VALUE-

TYPE

owl:cardinality

:SLOT-

CARDINALITY

Property owl:maxCardinality

:SLOT-

MAXIMUM-

CARDINALITY

Restricti

ons

owl:minCardinality

:SLOT-

MINIMUM-

CARDINALITY

OWL FIPA

owl:allValuesFrom

:SLOT-SAME-

VALUES

Restricti

ons

owl:someValuesFrom

:SLOT-SOME-

VALUES

owl:hasValue

:SLOT-SUBSET-

OF-VALUES

Whereas in FIPA Ontology :SLOT-VALUE-

TYPE is used to define a data type for a slot. Hence

owl:DatatypeProperty will be transformed to

:SLOT-VALUE-TYPE.

For other properties such as

owl:equivalentProperty, owl:inverseOf,

etc and property restrictions such as

owl:cardinality, owl:allValuesFrom,

owl:hasValue, etc there are appropriate

equivalent operators in FIPA, which also perform

the same functionality as that of OWL tags. For

further mappings see Table 2.

4 TEST BED AND RESULTS

First Machine has Tomcat web server, Eclipse

version 2.1 installed in it and Jena 2.4 to enable web

server to create and deploy OWL web services.

On the second machine, the above mentioned

system is installed to provide required

transformations.

The third machine contains FIPA compliant agent

platform SAGE

(Abdul Ghafoor, 2004), on which

agent providing some services have been deployed.

The successful ontological transformations depends

on the provided information of the web service, if

the provided information is valid then the

transformation will be successful and 100% results

can be obtained. Agents with the help of AgentWeb

Gateway (Hiroki Suguri, 2004) (Muhammad Omair

Shafiq, 2005) (H. Farooq Ahmad, 2004) searches for

the required service in the UDDI. Upon successful

search it retrieves the web service. Now the software

agent needs to know about the ontology, the

predicate schema and the concept schema therefore

it passes the OWL ontology to the Ontology

Gateway which transforms the Web Serviced

published in OWL to its equivalent FIPA Ontology.

The agent can then use the transformed ontology to

populate its knowledge base.

Figure 3: Testbed Cofiguration.

SAGE

Agent Platform

Ontology Gateway

Agent Web Gateway

OWL 2 FIPA SL

Converter

FIPA SL 2 OWL

Converter

Apache Tomcat

Web Server

JENA

Web Service

Machine 3Machine 2

Machine 1

ICEIS 2007 - International Conference on Enterprise Information Systems

312

The time required for the transformation depends

on the number of classes involved and their

corresponding properties.

The proposed system will also help the W3C

compliant OWL based web services (Grid entities)

to communicate with the software agents. The Grid

entity sends a SOAP based UDDI query which is

converted into a valid ACL based DF query with the

help of the AgentWeb Gateway. If the required

service is found it is then forwarded to the Onotlogy

Gateway which then converts the FIPA Ontology

into its equivalent OWL ontology. And the

converted Ontology is then embedded into its

original SOAP message with its name and certain

other parameters, which is then forwarded to the

Web Service client that requested for the service.

This can be established using the same testbed

configuration as described above.

5 FUTURE WORK

This paper proposes ontology gateway architecture

and detailed mappings that is our research focus for

enabling flexible, autonomous interaction between

Semantic Web services and agent services. A

prototype implementation of this architecture is

under development at NUST-Comtec lab. It is

intended to improve the proposed design so as to

cater more semantically enriched communication.

We expect that this effort of conducting

communication between agents and web services via

a Gateway service, bridging agents and WS is only a

prelude to exploring the immense potential it offers

as a means to compose, invoke, administer and

manipulate heterogeneous service populations. As

OWL is based on description logic so it

comparatively less expressive then FIPA ACL/SL

therefore we need to introduce new operators while

transforming from SL to OWL.

6 CONCLUSION

In this paper we have proposed an enhanced version

of Ontology Gateway as compared to the version

proposed in (Maruf Pasha, 2006). The system will

now allow the user to enter queries in natural

language which will then be processed by the NLP

module which extract keywords from the query and

generate an equivalent ACL query. Our proposed

system will provide mappings/translations for more

operators and tags for the interoperability of FIPA

Semantic Language with Ontology Web Language

(OWL) and will prove to be a vital step in achieving

a semantic web and ultimately the Semantic Grid.

In order to achieve this milestone, this paper

contains the key issues and the detailed comparison

of the two technologies. The paper describes the

testbed configuration and the results when OWL

ontologies are mapped/transformed using the

Ontology Gateway. We have also described the

proposed architecture that will enable us to achieve

autonomous coordination in the messages that are

exchanged between the FIPA compliant Multi agent

systems and OWL based web services. And will

allow agents to interact with OWL based Web

Services.

REFERENCES

T. Berners-Lee, 2004; T. Berners-Lee, J. Hendler, and O.

Lassila. The semantic web. Scientific American,

284(5): 34-43.

Maruf Pasha, 2006; Maruf Pasha, Sabih ur Rehman, H.

Farooq Ahmad, Arshad Ali, and Hiroki Suguri,

“Middleware Between OWL and FIPA Ontologies in

the Semantic Grid Environment”, The 2006

International Conference on Semantic Web and Web

Services (SWWS'06), June 2006, Las Vegas, USA

D. Fensel, 2003; Ontologies: Silver Bullet for Knowledge

Management and Electronic Commerce, 2nd edition.

Springer-Verlag,Berlin,2003.

M. Dean, 2004; M. Dean and G. Schreiber, eds. OWL

Web Ontology Language Reference. 2004. W3C

Recommendation 10 February 2004.

IBM Grid Computing Homepage ; www.ibm.com/grid/

Abdul Ghafoor, 2004; Abdul Ghafoor, Mujahid ur

Rehman, Zaheer Abbas Khan, H. Farooq Ahmad,

Arshad Ali, “SAGE: Next Generation Multi-Agent

System”, PDPTA'04, pp.139- 145, Vol. 1, (2004).

M. Wooldrige, 1997; ”Agent based Software Engineering

IEE Proc. Software Engineering, vol. 144, no .1,

pp.26-37, 1997”.

Ferber, 1999; Multi Agent Systems, Addeson Wesle,

ReadingMA,1999.

Eibe Frank, 1999; Eibe Frank, Gordon W. Paynter, Ian H.

Witten, Carl Gutwin and Craig Nevill-Manning,

Domain-Specific Keyphrase Extraction. Proceedings

of the Sixteenth International Joint Conference on

Artificial Intelligence, Stockholm, Sweden. Morgan

Kaufmann Publishers, San Francisco, CA, pp. 668-

673.

J. Dorre, 1999; J. Dorre, P. Gerstl, and R. Seiffert,

Text mining: Finding nuggets in mountains of textual data.

In Proc. of the 5th ACM SIGKDD, pages 398--401,

1999.

Zaheer Abbas Khan, 2004; Zaheer Abbas Khan, H. Farooq

Ahmad, Arshad Ali, Hiroki Suguri, China, April

“Decentralized Architecture for Fault Tolerant Multi

ONTOLOGY GATEWAY - Enabling Interoperability between FIPA complaint Agents and OWL Web Services

313

Agent System”, ISADS, 04, 5-7, 2005 (2004) pp 167-

174.

Hiroki Suguri, 2004; Hiroki Suguri, H. Farooq Ahmad, M.

Omair Shafiq and Arshad Ali: Agent Web Gateway -

Enabling Service Discovery and Communication

among Software Agents and Web Services. Proc.

Third Joint Agent Workshops and Symposium

(JAWS2004), pp. 212-218, October 2004, Karuizawa,

Japan

Muhammad Omair Shafiq, 2005; Muhammad Omair

Shafiq, Hafiz Farooq Ahmad, Hiroki Suguri, Arshad

Ali, “Autonomous Semantic Grid: Principles of

Autonomous Decentralized Systems for Grid

Computing” IEICE/IEEE TRANS., VOL. E85-

A/B/C/D, No. 1 (2005).

FIPA SL Content Language Specifications;

http://www.fipa.org/specs/fipa00008/SC00008I.html#

_Toc26668863

H. Farooq Ahmad, 2004; H. Farooq Ahmad, Hiroki

Suguri, M. Omair Shafiq and Arshad Ali: Autonomous

Distributed Service System: Enabling Webs Services

Communication with Software Agents. Proc. of 16th

International Conference on computers and

Communication (ICCC 2004), pp. 1167-1173

September 2004, Beijing, China

ICEIS 2007 - International Conference on Enterprise Information Systems

314