AN AGENT BASED INFORMATION SYSTEM FOR

COMMUNITIES MEDIATION

Aluizio Haendchen Filho

Famesul – Uniasselvi

Hércules Antonio do Prado

Embrapa Food Technology, Catholic University of Brasília

Míriam Sayão

PUC-RS

Fénelon do Nascimento Neto

Embrapa Food Technology

Keywords: Enterprise Information Systems, Communities Mediation, Agent Based Systems.

Abstract: The adoption of the Multi-Agent System paradigm in the context of Enterprise Information Systems has

been accelerated by the technology brought by Internet. The importance of MAS applications increases as

the ubiquity of Internet, with its distributed and interconnected elements, becomes a de facto reality. How-

ever, the development of MAS is not trivial; agents-based systems are typically complex and difficult to de-

velop due to the features required, some of them hard to implement. In this paper we briefly describe MI-

DAS, a service oriented (SOA) framework built on a reusable, adaptable and loosely coupled architecture,

that aims to help in the development of MAS applications. An application in the domain of expert/customer

mediation is presented to evidence the advantages of the framework. After that, the advantages of applying

the SOA standard over the traditional message-based approach for MAS development are discussed.

1 INTRODUCTION

Due to the recent availability of Web-based

development technology, the application domain of

Multi-Agent Systems (MAS) has spread out. This

kind of system has been applied in many areas in the

Internet context: e-Commerce, Web Services

(Finin, 1994), Knowledge Management (Kendall et

al., 1999), Semantic Web (Decker et al., 2000), and

Information Systems in general (Adam et al., 2004

and Jennings et al., 1996). The adoption of MAS in

the application development over Internet has

enabled interesting solutions to B2B, e-Business

(Boughaci et al., 2005), and also applications that

requires interoperability based on knowledge about

applications and business processes. Klusch (1999)

identifies information intelligent agents as one of the

most promising areas for applying agents

technology. Information agents acts in fields like

collaborative systems over Internet, knowledge

discovery from heterogeneous sources, systems for

intelligent management of information for intranet

of Internet, among others.

The importance of Internet as a common and ubiqui-

tous systems environment has led to efforts like

MIX (Mediation of Information using XML) (Baru

et al., 1999), that uses agents to integrate informa-

tion distributed in disparate sources. The Web also

can be seen as a big distributed database having

XML (and its extensions or modifications) as an

underlying data model. Agents can naturally perform

the roles required in a mediation of information

process. In this paper, we apply MIDAS (Haendchen

Filho, 2006), a platform to develop MAS, to build an

295

Haendchen Filho A., Antonio do Prado H., Sayão M. and do Nascimento Neto F. (2007).

AN AGENT BASED INFORMATION SYSTEM FOR COMMUNITIES MEDIATION.

In Proceedings of the Ninth International Conference on Enterprise Information Systems - SAIC, pages 295-298

DOI: 10.5220/0002388302950298

 SciTePress

application to mediate the communication among a

community of experts in food technology and cus-

tomers in need for qualified information about food

products in supermarkets. For this propose, we de-

fine a mediator agent that perform the information

interchange between these communities, involving

tasks like collecting questions, searching for a con-

sent among experts, providing translations (with the

support of communication specialists), and distribut-

ing the answers.

2 MEDIATION AMONG

COMMUNITTIES

The dialogue among communities with different

jargon can be seen as a sub problem of the cross

lingual dialogue problem, a research field that looks

for highly accurate communication of semantically

complex content (Piwek and Power, 2006). The

‘Saint Graal’ for researchers in this field is to build

an automatic process to translate one language to

other, being able to cope with the different semantic

levels of each specific language. A solution to this

problem is much more important if you consider the

challenge to human-to-human communication

through a browser that has been set by the Internet.

To build an effective automatic translator for this

scenery is a dream far to be realized. However, the

necessity of communication among communities is a

reality that requires, if not the best solution, at least

one viable solution. The solution we present uses the

agents technology to facilitate the negotiation among

the communities members.

3 MIDAS DESCRIPTION

MIDAS (Middleware for Intelligent and Distributed

Agent-based Systems) (Haendchen Filho, 2006) is a

platform that provides an environment to run agents

and a WSA-compliant framework to ease its devel-

opment. WSA aims to provide a common Web Ser-

vice definition and its location inside a wider archi-

tecture in order to guide service implementers, au-

thors of services specification and Web application

developers. WSA represents the natural evolution

from traditional applications to SOA ones

(Odell, 2005).

Figure 1 shows the MIDAS generic architecture,

which is based on the coexistence of several con-

tainers, each one executing a JVM (Java Virtual

Machine). Each virtual machine provides a complete

execution environment, where agents can execute

concurrently in the same host. The architecture in-

cludes two different types of container: the Front-

End Server (FES) and the Agent Container (AC).

FES plays the integration rules of the platform, pro-

moting the synchronization services and interopera-

bility with external applications. It is similar to the

front-end server used by JADE (Adam et al., 2004).

AC is a Web container that can be inhabited by

organizations, agents and components.

Middleware and application agents are the basic

elements of the platform. The middleware agents

abstract completely generic characteristics, such as

communication, concurrency, lifecycle management,

services discovery and interoperability. They enable

the developer to focus only in the application details.

The introduction of the agent concept to play these

roles makes easier to satisfy important non-

functional requirements for Web architectures, like

flexibility, dynamic behavior, and adaptability. The

following middleware agents have been used to

perform the roles defined by the WSA reference

model,: (i) a Broker agent, playing the MOM roles;

(ii) a Proxy agent, playing the SOM roles; (iii) a

Catalog agent, playing the ROM roles and (iv) a

Manager agent, playing the MGM roles.

Figure 1: The main concepts of the MIDAS.

Application agents and components are only lo-

cated in the AC containers. AC provides a structure

composed by abstract classes (Agent and Compo-

nent) and a blackboard. The introduction of the ab-

stract agent concept extends the WSA reference

architecture specification, providing a way to group

in a super-class the common properties to all the

agents. The application agents (or components) are

implemented by extending the abstract classes,

which provide the hot-spots from which specific

ICEIS 2007 - International Conference on Enterprise Information Systems

296

applications behavior can be implemented. Applica-

tion agents and components are instantiated in Or-

ganizations, and the Blackboard agent offers a pow-

erful mechanism to support for the agents communi-

cation model.

Besides agents and components, the platform

carry resources representations, which can be stored

in XML files, relational databases or documents in

general. The General Cache Descriptor element

located in the FES represents the platform-

consolidated structure of resources, kept in cache

memory. In the AC container, the LXS element

represents the services specification described in

XML. The resources structure is also kept in cache

memory, being represented by the LCS element.

LDB represents local databases, which can be han-

dled and/or accessed by agents and components. The

Web Services specifications and registries are repre-

sented by the WSDL and UDDI elements.

4 EXPERT/CUSTOMER

MEDIATION

The problem approached in this paper can be visual-

ized in Figure 2. C

, C

, .., C

represent the custom-

ers requiring specialized information in non-

technical language. E

, E

, …, E

represent the ex-

perts that can provide technical answers to the cus-

tomers questions. The triangle represents a set of

communication specialists that are in charge of

translating the technical language from the experts to

the customer language. In order to enable its reuse,

the answers are stored in an answers server. Notice

that the information required by customers are those

not supplied by the product labels. For instance, a

customer may be interested in knowing about the

effects on human health of a cereal produced from

GMO (Genetic Modified Organism) grain, or the

difference from organic to hydroponics lettuce. In

this case, a mediator agent will be responsible for

building a consensus among the specialists ad-

dressed to provide a sound technical answer to the

customer. This answer will be translated by the

communication team and sent back to the customer

and stored in the answers server. The application

developed is named BeyondTheLabel (BL).

Figure 3 shows a partial of BL structure, as pro-

vided by one of the GUIs wizards in MIDAS. The

right side panel shows the resource representation, in

the structure mode. The BeyondTheLabel element is

the root node, from what the agents and components

are instantiated. It can be seen, in the panel, the

components wrapper AreaData, InstitutionData,

ResearcherData, that encapsulate Web Services

required by the application. The component Institu-

tionSearch encapsulate two Web services: searchIn-

stitutionByName and searchInstitutionByRegion,

that retrieve institutions by name of region. The left

side panel shows specification details of searchInsti-

tutionByName service: service name, address, pa-

rameters and description.

Figure 2: Interaction among communities.

Figure 3: BL container partial view provided by the Man-

ager GUI wizard.

Agents can request Web Services transparently,

as they were located inside the container. Con-

creteAgent is an agent that requests a remote Web

Service. For this purpose, it invokes the require()

method, that is inherited from the abstract class

Agent, informing as parameter the service name.

The method returns a service wrapper object from

the Manager agent. In this moment, it adds the data

in the wrapper and invokes its run() method, for a

synchronous call, or submit(),for an asynchronous

call.

Whenever a remote request is executed, the

Proxy agent detects what protocol is used (HTTP or

SOAP), and redirects the request to the Broker in-

stead of asking for the class Factory instance of a

....

Customer Community

Translator

team

Answers

serve

Expert Communit

AN AGENT BASED INFORMATION SYSTEM FOR COMMUNITIES MEDIATION

297

native entity (in case of a local call). Once the re-

quest is redirected to the FES Broker, it is forwarded

to Adapter, that converts it to the SOAP format. The

processing results are captured and send back to the

client agent in a List.

This short description shows how the agents can

perform a remote call for Web Services, in the BL

application.

5 DISCUSSION

In this paper MIDAS, a platform to support MAS

development, has been applied to the communities

mediation problem. An application to build consen-

sus among specialists over customers questions were

presented. Considering that, even for simple do-

mains, the automatic translation of different jargons

is a far to be achieved reality, the solution involving

human actors sounds to be an interesting one. The

next step for this work is to stress the BL application

in an experimental plot of three supermarkets in

order to have a robust application to be scaled for

any supermarket that desires to join the BeyondThe-

Label network.

REFERENCES

Adam, E.; Mandiau, R. Design of a MAS into a Human

Organization: Application to an Information Multi-

Agent System. Proceedings of the 5th Agent-Oriented

Information Systems, October 13, 2003 Chicago, IL,

USA. Springer Verlag, Lectures in Artificial Intelli-

gence LNAI 3030, 2004. pp. 1-15.

Baru, C.; Gupta, A.; Ludaescher, B.; Marciano, R..; Papa-

konstantinou, Y.; Velikhov, P. XML-based Informa-

tion Medidation with MIX. Proceedings of the Exhibi-

tions Program of ACM SIGMOD, 1999. pp. 25-34.

Boughaci, D.; Drias, H. An Agent-Based Approach using

the ebXML Specifications for E-business. Proceedings

of the 5th International Conference on Information

Technology: Coding and Computing (ITCC'05) - Vol-

ume II, 2005, pp. 766-767.

Decker, S.; Melnik, S.; Harmelen, F. V.; Fensel D.; Klein

M.; Broekstra, J.; Erdmann, M.; Horrocks, I.. The Se-

mantic Web: the Roles of XML and RDF. Proceedings

of the IEEE Internet Computing, September 2000.

pp.71-80.

Finin, T. KQML as an agent communication language.

Third International Conference on Information and

Knowledge Management. CIKM-94, ACM Press, nov.

1994. pp. 17-29.

Haendchen Filho, A. A Middleware Framework for Buid-

ing Multi-Agent Systems in the Internet. Ph.D. Thesis,

Pontifícia Universidade Católica do Rio de Janeiro

PUC-Rio, Rio de Janeiro, RJ, Brasil, 2005.

Jennings, N. R.; Wooldridge M. J. Applying Agent Tech-

nology. Int. Journal of Applied Artificial Intelligence,

1996. 9 (4) 351-369.

Kendall, E. et al. A Framework for Agent Systems. In:

Implementing Applications Frameworks – Object-

Oriented Framewoks at Work. M.Fayadd et al. John

Wiley & Soons, 1999.

Klusch, M. (Ed.). Preface of “Intelligent Information

Agents.” Springer Verlag, Berlin, 1999.

Odell, J. Agent-based Process Management for SOA and

WS Applications. OMG Workshop on MDA, SOA and

Web Services March 21-24, 2005, Orlando, FL, USA.

Piwek, P.; Power, R. CROCODIAL: Crosslingual Com-

puter-mediated Dialogue. In Proceedings of the 8

In-

ternational Conference on Enterprise Information Sys-

tems, Paphos, Cyprus, 2006, 75-84.

ICEIS 2007 - International Conference on Enterprise Information Systems

298