COMPOSITION OF WEB SERVICES IN THE ICS

ARCHITECTURE

C. Roberto Baluz, S. Labidi, R. F. Tomaz, B. Wanghon and Nathália R. S. Oliveira

Eletrical Engineering Department, Federal University of Maranhão, Campus of Bacanga, São Luís-MA, Brazil

Keywords: Web Services, Semantic Web Services, Web Service Composition, Semantic Web, Ontologies.

Abstract: This paper proposes the use of the Web Services Composition to enhance the matchmaking process

acctually in use within the ICS (Intelligent Commerce System), a Business-to-Business e-commerce system.

The actual matchmaking process used in the ICS considers only single services and may return a high

number of false-negative results. The new approach aims to reduce the number of false-negative results

through the composition of existing single services to obtain new functionality.

1 INTRODUCTION

Web Services are transforming the web from a

collection of static pages to a web of dynamic

service providers that automatically discover

information that we seek, negotiate on our behalf for

goods we intend to purchase, gather information

from different sources, and fuse it into coherent

forms. Nowadays, Web Services are discovered and

invoked manually by human users (hard code

approuch), limiting their ability to take advantage of

opportunities that may exist. This investigation

represents an effort to overcome these limitations.

In this paper, we show the Web Services

composition process in the ICS context as an

alternative to reduce the number of negative results,

returned by the Matchmaker agent, responsible for

the matching of negotiating agents in the ICS.

Actually the Matchmaker agent ignores the

possibility of the salers agents combine theirs

capacities to attend a solicitation from a purchaser.

As a result of this limitation, even if there is the

possibility of attending the service by the agents

present in the negotiation environment, the

Matchmaker agent returns a denial of service (false

negative) to the requesting user.

The main contribution of our work is to improve

the matching process of agents within the ICS,

increasing the number of matching results returned

by the Matchmaker agent.

In the next section, we present an overview of

the ICS architecture. Section 3 describes three

motivating scenarios for our work. Section 4 shows

the actual matchmaking procedure used in the ICS

and emphasizes our contribution to enrich the

matchmaking process. Finally, section 5 presents our

conclusion.

2 OVERVIEW OF THE ICS

ARCHITECTURE

Aiming to describe the context of our work, we

present a brief high level vision of the ICS project.

The ICS is a Agent-Oriented implementation of

a Business-to-Business e-commerce system, where

the trading agents, representing purchasers and

salers, work in an open environment as the Internet,

moving through the network to meet at common

Figure 1: The ICS Enviroment

435

Roberto Baluz C., Labidi S., F. Tomaz R., Wanghon B. and R. S. Oliveira N. (2004).

COMPOSITION OF WEB SERVICES IN THE ICS ARCHITECTURE.

In Proceedings of the Sixth International Conference on Enterprise Information Systems, pages 435-438

DOI: 10.5220/0002645004350438

 SciTePress

negotiation areas. (Labidi et al., 2003)(Tomaz, 2003)

The main ICS components are: Virtual

Marketplaces (VMP), Regions, Matchmaker Agent,

Mediator Agent, Modeling Agent, Ontologies

Repository and Stereotype and Advertisement

databases as it showed figure 1.

The Virtual Marketplace (VMP) is the place

where the agents carry through the negotiation and

the closing of the contract. A Region is an

abstraction of higher level than the VMP that

provides a vision of grouping of virtual markets that

are operating on the same domain ontology, that is,

acting in the same area of business (vehicles,

remedies, books etc).

An Ontologies Repository is a semi-structured

database that allows the storing of the domain’s

ontologies. The ICS is not limited to operate in a

specific business domain. The stereotype database is

a semi-structured database that allows the storing of

the users profile. It is used by the traders agents to

deliberate on the preferences of the companies they

are representing. The advertisement database is a

semi-structured database that allows the storing of

the salers’ advertisements.

The Matchmaker Agent objectives to approach

trader agents with complementary objectives, that is,

it is the responsible for the composition of the VMP.

The Mediator Agent operates as an arbitrator

inside of a VMP, it follows each carried through

transaction and intervenes when necessary trying to

decide problems of negotiation, formation and

execution of contracts. For each Virtual Market

there is an instance of the Mediator Agent.

A Traders Agents can be a purchaser or a

salesman. These agents are triggered for the using

companies of the ICS from its WEB interface.

The Modeling Agent objectives to inform the

Mediator Agent of the preferences of each Trader

Agent so that the Mediator Agent can interact in a

particular way with each Trader Agent.

This section presented a high level vision of the

ICS architecture. In the next section we present two

motivating scenarios for the development of our

work.

3 MOTIVATING SCENARIOS

With the objective to give a better understanding of

our investigation, we present here two scenarios that

show how the composition of services can enrich the

actual matchmaking procedure used in ICS.

Suppose three services represented by Web

Services: (i) an airplane tickets agency; (ii) a hotel

chain and (iii) a transportation firm, all operating

globally. Consider that all the services are descripted

in a services description language like DAML-S

(DAML-S,2003) and published in a UDDI directory

service (Daum, 2002), as (Paolucci, 2002) proposes.

If a client wishes to buy a full travel package, a

matching service that doesn’t use the services

composition techniques will return that there isn’t a

service that satisfies the client’s request. However, a

matching engine that is capable to use the services

composition will promote the sum of the individual

capacities of each service to satisfy the client’s

needs.

Another example of Web Services composition

is presented by (Sirin, 2003). The example deals

with the composition of the capacities of an on-line

language translator and a dictionary service. The

language translation service translates texts between

various pairs of languages, the dictionary, however,

returns the meaning of words in english. If the user

needs the service of a french dictionary, neither of

the services alone will satisfy this demand. But

together the services are capable to attend the

request – the french text is passed as input to the

language translation service, who returns as output

the text in english. This output is passed as input to

the english dictionary service to obtain synonyms

and then translated back to french.

To provide the semantic concepts as “language”,

“english”, “french”, etc. we can use ontologies,

(Guarino,1998) now used in the construction of the

Semantic Web.

In the next section, we’ll give a brief description

of the matching procedure actually used in ICS,

concentrating on how the Web Services composition

could enrich it.

4 ENHACED MATCHMAKING

PROCESS

The Matchmaker agent in the ICS restricts itself to

compute the degree of similarity (distance) among

purchasers’ requests and salers’ advertisements. As

result, it returns a binary match-pairs list (or record-

match list), which contains the possible matching

among customers and suppliers (a set of clusters).

When the Matchmaker agent is asked to make a

search for a purchaser agent, this will be done inside

an advertisements database. The matchmaker looks

for possible single saler agents that satisfy the

searches requests and returns the possible partners

salers agents.

An important requirement for the Matchmaker

agent is the search flexibility. The search for

partners should not be restricted to the syntax aspect

only. This is because equivalent terms incorporated

in more generic concepts (subsumption), as well as

ICEIS 2004 - SOFTWARE AGENTS AND INTERNET COMPUTING

436

terms having some relationships (such as

aggregation, synonymy, antinomy, etc.), must be

considered in the matchmaking process. The figure 2

displays the ICS Matchmaking process.

Purchasers and salers use DAML-S to express

requests and advertisements respectively. The

DAML-S parser extracts inputs, outputs and pre-

conditions from DAML-S requests and

Advertisements.

The Matchmaker Agent, made by the semantic

matching engine and DAML-S Parser, classifies and

crosses the requests inputs, outputs and pre-

conditions to the advertisements inputs, outputs and

pre-conditions. The ICS reasoning server,

implemented by a Description Logic reasoner

(Tomaz,2003), supports the classification of

taxonomy of the terms found in the inputs, outputs

and pre-conditions based on the application domain

ontology.

To maintain compatibility with the

advertisements repository structure, and the parser

capabitlity, a composed service must also use

DAML-S. In fact, a composed service entry is

exactly the same as the entry of a single service. The

composition plan – a description of wich ones and in

what order the single services will be grouped to

form the composed service - is stored in other

repository, the Composition Repository. With this

compatibility, no alteration is needed in this part of

the matchmaking process to make it use composed

services also.

When a composed service is chosen by the

Matchmaker Agent is that the first difference in the

actual process appear, but now under the control of

other agent, the Composer Agent. Instead of the

salers agent that form the composed service, it is the

Composer Agent that is warned when a composed

service is found adequate. The Composer Agent then

warns the salers agents, as the Matchmaker Agent

does, and then passes the composition plan to the

Mediator Agent, wich will be responsible for

realizing the business.

Figure 2: The ICS Architecture (Tomaz, 2003)

There are several approaches for the

development of the matchmaking algorithms.

Nevertheless, DAML-OIL is greatly influenced by

the description logics languages. That’s why, we

proposed, in our work, the use of algorithms based

on description logics (DL) languages.

A limitation of the actual matching procedure

used in ICS is that the matchmaker agent, which is

responsible to gather the negotiating agents, ignores

the possibility to compose the capacities of the salers

agents to satisfy the requests of the buyers that are

passed as parameters to it. The matchmaker simply

compares the buy requests of each one of the

advertisements published in the advertisements

repository, so that if none of them presents some

degree of similarity with the buying parameters, the

matchmaker returns an empty list (a void binary

match-pairs list).

ure 3: Web Com

osition on the ICS matchin

rocess.

Figure 4. Composition Plan

COMPOSITION OF WEB SERVICES IN THE ICS ARCHITECTURE

437

We propose the use of the Temporal Workflow

technology (Labidi et al., 2000) to model a semi-

automatic service composition tool within ICS. This

tool has a GUI (Graphical User Interface) that

allows the user to compose a new service from the

available services in the advertisement repository.

To illustrate this, let’s take the second example

presented in section 3 – Motivating scenarios.

The advantage of the use of the ICS Composer is

that, from a restrict set of services published in the

advertisement repository, a user can create a vast

number of new services that enriches the

environment and increases the possibility of

satisfaction of the requisites given by the buyers,

what increases also the number of accomplished

business in the ICS. As it is shown in figure 4, to

make the service composition it is necessary that

there is communication between them. We propose

the use of the SOAP protocol to allow each service

to pass its output as an input parameter to other

services.

After the composition of the new service from

the previously stored services of the advertisement

repository, the user must also describe its capacities.

We propose the use of the DAML-S as the

description language of the services capacities. Each

new service represents a new entry in the

advertisement repository and also a new entry in the

composition plans repository. When a composed

service satisfy the requisites of a buyer, the

Composer Agent consults the composition plan in

the composition plans repository and executes it,

receiving and sending the parameters of each one of

the simple services that constitutes the composed

service until it fulfills its objective described in the

capacities of the composed service. As figure 3

shows.

As we can observe, the main use of the

composer agent is to guarantee the execution of the

composition plan. It is the composer agent that

communicates with the buyer, giving him the

necessary details to fulfill the business (price,

payment form, delivery date, etc.)

5 CONCLUSIONS

In this paper, we presented how the technologies that

come from the development of the Semantic Web

allied to the technologies of the Web Services can

enrich the actual tools of e-commerce. The

composition of web services enhances the

negotiation possibilities inside the ICS, increasing

the number of possible traders partners.

The coupling of the composition service to the

ICS architecture will be an easy task, once it uses the

same standards and technologies used in the ICS

prototipation. Today, we are studying the tools and

standards used in the Semantic Web and soon we

will begin the prototipation work to validate our

propositions.

ACKNOWLEDGMENTS

We are very grateful to the Faculdade Atenas

Maranhense (FAMA) for granting this research.

REFERENCES

DAML-S Semantic Matchmaker

http://www.ri.cmu.edu/projects/project_480.html

Last visited August 2003.

Daum, Bertold. Arquitetura de sistemas com XML:

conteúdo, processo e apresentação. Rio de Janeiro. Ed.

Campus. 2002.

Guarino, N. Formal Ontology and Information Systems.

in: N. Guarino, (Ed.) Formal Ontology in Information

Systems. pp. 3-15, IOS Press, Amsterdam,

Netherlands. 1998.

Hendler, J., Berners-Lee, T. and Miller, E. Integrating

Applications on the Semantic Web. Journal of the

Institute of Electrical Engineers of Japan. Vol.

122(10), p. 676-680. October, 2002.

Horrocks, I., Harmelen, F. V. Reference Description of the

DAML+OIL Ontology Markup Language. Draft

Report, 2001. Acessado em Janeiro de 2002.

Disponível na Internet por www em:

http://www.daml.org/2000/12/reference.html. 2001.

Labidi, S., Hammoudi, S. and Gannoun, L. Cooperation

and Temporal Organization in Workflow

Management. In the Proceedongs of the 2000

International Conference on Artificial Intelligence (IC-

AI'2000). World Scientific Engineering Society. Las

Vegas, USA. June 20-22, 2000.

Labidi, S., Fonseca, Luis C., Filho, Othon B. and

Nascimento, E. Intelligent B2B Commerce System. In

the textbook Techno-Legal Aspects of Information

Society and New Economy: an Overview, Formatex.

Spain. 2003.

Paolucci, M., Kawmura, T., Payne, T. and Sycara, K.

Semantic Matching of Web Servies Capabilities. In

Firts Int. Semantic Web Conf. 2002.

Sirin, E., Hendler, J. and Parsia, B. Interactive

Composition of Semantic Web Services. University of

Maryland, USA. 2003

Tomaz, Ricardo F., Labidi, S. and Wangon, B. A Semantic

Matching Method for Clustering Traders in B2B

Systems. To appear in 1st Latin American Web

Congress, Santiago, Chile. November 2003.

ICEIS 2004 - SOFTWARE AGENTS AND INTERNET COMPUTING

438