MODELLING ONTOLOGICAL AGENTS WITH

GAIA METHODOLOGY

María Auxilio Medina N.

Universidad Tecnológica de la Mixteca, Huajuapan de León, Oaxaca, México

Alfredo Sánchez, Ma. Lourdes Fernández

Universidad De Las Américas -Puebla,

Cholula, Puebla, México

Keywords: Ontologies, digital libraries, agents

Abstract:

Multi-agent systems have been successfully applied in information retrieval tasks, especially in

environments whose sources o

f information are distributed and highly heterogeneous. They can be

perceived as an alternative to face problems that traditional search engines are not able to solve yet. On the

other hand, ontologies have shown their efficiency to management different sources of information. We

present the model of some software agents that use ontologies to improve information retrieval tasks in a set

of federated digital libraries. Gaia methodology is used for this purpose and the paper highlights some of its

main advantages. It also shows that this methodology can be easily used in similar environments to avoid ad

hoc construction of agent-based systems.

1 INTRODUCTION

The use of the World Wide Web as a space widely

used to publish documents and the expansion of

Internet are relevant factors that have contributed to

the increase of sources of information. In this

document we are interested in digital libraries mainly

because they are invaluable repositories of reliable

and structured information.

In order to retrieve relevant resources from

sev

ral digital libraries, users require knowing their

allocation as well as features of their interfaces to

express their information requests appropriately.

Once search mechanisms of each digital library are

activated, users are in charged of collecting the results

of each source and afterwards they choose a set of

sources of information. Although a wide range of

technological tools have supported digital libraries

operation, nowadays this task is a very time

consuming one and it is often that it is delegated to

software or human agents.

This problem is not new and it has been analyzed

areas such as distributed systems, interoperability

systems, federated information systems and multi-

agent systems. Due to the problem has to deal with

different levels of semantic heterogeneity; ontologies

emerge naturally as appropriate tools [Pomerantz &

Silverstein 2001].

This paper proposes the use of ontologies by a

t-based system to retrieve information from a set

of federated digital libraries. Ontologies are used to

support the disambiguation of keywords in user

queries as well as to describe the repositories or

sources of information. In order to avoid ad hoc

development of agent-based software, the design of

these agents is based on Gaia methodology.

The paper is organized as follows: Section 2

bri

fly describes previous work with the purpose of

establish the antecedents of ontological agents.

Section 3 presents the design of ontological agents

using Gaia methodology. Related work is presented

in Section 4. Conclusions and future works are

presented in Section 5.

556

Auxilio Medina N. M., Sánchez A. and Lourdes Fernández M. (2004).

MODELLING ONTOLOGICAL AGENTS WITH GAIA METHODOLOGY.

In Proceedings of the Sixth International Conference on Enterprise Information Systems, pages 556-560

DOI: 10.5220/0002645205560560

 SciTePress

2 PREVIOUS WORK

In order to understand the design of ontological

agents, it is necessary to know the context in which

these agents are going to be integrated. This is the

Virtual Reference System [Sánchez et al. 2001]. In

the further, the term Vref is used to make reference to

this system.

VRef is a virtual space in which users post their

information needs or queries expressed in natural

language and a staff of reference librarians as well as

a set of software agents called reference agents are in

charged of searching relevant digital or physical

resources from the UDLAP

digital library. A

detailed description of this kind of agents can be

found in [Medina et al. 2002] and [Medina et al.

2003]. One of the most relevant characteristics of

Vref is that it has a knowledge base that is maintained

and enriched with the successful interactions of users.

The use of this knowledge base allows a reference

librarian make use of the work generated by other

reference librarians or by reference agents. Likewise,

a reference agent can also have access with the same

objective. It is also in charged of implement a

mechanism to associate a similarity measure between

queries.

Vref has been operating for almost a year and a

half. At present, we are considering to use it as the

interface to have access to a set of federated digital

libraries. In this paper, we took the definition of a

federated system proposed by [Busse et al 1999], this

is a kind of system which offers read and write

operations to data management such as any traditional

database system. It is commonly built through an

integration technique of schema. It is formed by

structured components that can be accessed by query

languages.

The particular federation of digital libraries we are

thinking in makes use of the Open Archives Initiative

Protocol for Metadata Harvesting, (abbreviated OAI-

PMH). It can be perceived as a basis for supporting

interoperability in digital libraries. The OAI-PMH

protocol includes a model consisting of two main

parts: data providers, which expose metadata, and

service providers, which harvest and process data

automatically. Although this protocol is relatively

new, (less than four years old), it has been

incorporated into the development of many important

research projects. [Van de Sompel & Lagoze 2001]

attribute its growing popularity to the availability of

tools to build OAI-PMH repositories and harvesters.

UDLAP is the short name of Universidad De Las

Amèricas Puebla

The proposal of retrieving information from a set

of digital libraries federated by the OAI-PMH

protocol is summarized in the next general algorithm:

1. A reference agent receives the query

and gets the keywords.

2. An ontological agent uses a general

ontology of natural language to

present to the user some

alternatives of new words or

referred words in order to eliminate

as much ambiguity as possible from

the original keywords.

3. Once a user chose the appropriate

alternative, the ontological agent

identifies a set of digital

libraries which according with a set

of well defined ontologies

potentially have relevant resources

to satisfy the query.

4. The ontological agent sends the

query to a mobile agent, which is in

charged of visiting the sources of

information and of the retrieval

process of relevant resources. A

mobile agent is used for each

digital library identified at step

5. The partial results are sent to the

reference agent and they are

presented to the user in a

transparent way at the Vref

interface.

3 DESIGN OF ONTOLOGICAL

AGENTS BASED ON GAIA

METHODOLOGY

Gaia is a top-down methodology for agent-oriented

analysis and design. It is based on the concept of

roles. A multi-agent system is understood as a

computational organization formed by the interaction

of various roles. Gaia methodology takes into account

the macro-level and the micro-level aspects of agent-

based systems [Wooldridge et al. 2000].

At the requirements statement, the identification of

requirements enables to model and to specify in a

conceptual level of detail ontological agents. It

MODELLING ONTOLOGICAL AGENTS WITH GAIA METHODOLOGY

557

describes the semantics of a system without concern

with implementation details. We adopted the same

requirements of a federated system: scalability and

adaptability. Analysis stage can be summarized in

the next following tasks: identification of roles,

identification and documentation of protocols and

elaboration of the roles model. The roles model is

presented by the next set of tables.

Table 1: Role to process a query

Description It is in charged of getting the

keywords from a query

expressed in natural language.

Protocols and

activities:

EliminateStopWords

Permissions: Read a query

Responsibilities:

Liveness: keywords were extracted from

the query

Safety: suggest new words just when

ambiguity is detected

Table 2: Role to formulate a query

Description: This role involves ensuring to

eliminate the much ambiguity as

possible from the keywords of the

queries.

Protocols and

activities:

ShowNewWords,

FormANewQuery

Permissions: suggest other words

Responsibilities:

Liveness: keywords were extracted from the

query

Safety: suggest new words just when

ambiguity is detected

Table 3: Role to describe the sources of information

Description: This role is related with the

description of a source of

information.

Protocols and

activities:

RepresentSchema

Permissions: represent a source of information

Responsibilities:

Liveness: the source can be described

Safety: have a representation of a source

of information

Table 4: Role to visit different sources of information

Description: This involves to have access to a

source

of information

Protocols and

activities:

RetrieveResources, SendResources

Permissions: search sources

Responsibilities:

Liveness: there is a set of sources

Safety: have access to a source

Table 5: Role to collect results

Description: This role has the objective of

collecting the results of each source

of information

Protocols and

activities:

JoinResources, FiltrateResources

Permissions: AcceptsResults

Responsibilities:

Liveness: there are available resources

Safety: establish a minimum set of

resources

We have identified four protocol definitions:

KeyWordsAgree, IdentificationRelevanteResource,

RetrieveResources, PresentResources. They are

represented in the templates of Figure 3. In this

templates, the top cell has the name of the role,

middle cells are used for initiator and responder. The

bottom cell has processing attributed. Inputs and

outputs are briefly described on the right from top to

bottom respectively.

From the protocols presented above represent the

services identified by each type of agent. It is worth

to mention that was during this stage that we have

decided to define two types of ontological agents we

have termed: NLP-Agent and SourceDescriptor

Agent. The service associated to the first type of

ontological agent is related with the purpose of

eliminating as much ambiguity as possible from

keywords of user query; instead, the second type of

ontological agent has the function of representing the

description of a source of information. The direct

graph of Figure 2 represents the acquaintance model

for the agent types.

ICEIS 2004 - INFORMATION SYSTEMS ANALYSIS AND SPECIFICATION

558

4 RELATED WORK

The problem of having access to heterogeneous

sources of information has been analyzed from

diverse points of view. This section includes some of

the most relevant ones are briefly described in this

section.

FIPA (Foundation for Intelligent Physical Agents)

fosters agent-based applications. It provides

specifications to support interoperability between

agents and a special specification to manage

ontologies provided by an Ontology Agent

(abbreviated as OA). Some of the services of an OA

are: searching and accessing public ontologies,

maintaining a set of ontologies and a translation

mechanism between ontologies. It is able to answer

queries about relationships between terms. At FIPA,

it is not mandated that every OA will be able to offer

these services; the implementation of these services is

left to developers [FIPA 2001].

ure 1: The interactions model

[Saavedra 2003] proposed a method to federate a

set of documental databases based on ontologies. It

accomplishes with the requirements of scalability,

adaptability. Ontologies are used to conciliate schema

of the databases and also to build a friendly user

interface at run time. The concepts are represented in

a structure called concept trees. Users select these

trees when he or she introduces a query through the

interface. Due to XML files are used to guide

software execution; any change in the databases does

not require recompilation of any module of the

system, providing physic and logic independence of

databases.

The semantic web project is focused on providing

intelligent access to heterogeneous and distributed

information. In this project, agents operate as

mediators between user needs and available

information resources. Ontologies play a key role in

this project [Fensel 2000], [Fensel 2001].

ure 2: Re

resentation of a

uaintance model

MODELLING ONTOLOGICAL AGENTS WITH GAIA METHODOLOGY

559

5 CONCLUSIONS

In this paper, we have briefly described Gaia

methodology and we have presented its application to

design a set of agents to retrieve information of a set

of federated digital libraries. This design has been

proposed as an alternative to have access to these

heterogeneous and distributed sources of information.

All the stages and the different of models were

included.

Gaia methodology was chosen due to its

adaptability characteristics in such a way that can be

applied to a wide range of agent-based systems. The

design of the agents presented in this paper with this

methodology provides us of a detailed description of

the system in such a way that the different kinds of

agents can be implemented easily; likewise, it can be

perceived as a reliable document to support system

maintenance. According to the requirements

statement, this design can be a base for the stages of

validation, verification and test.

REFERENCES

Louis, R., 1999. Software agents activities. In ICEIS’99, 1st

International Conference on Enterprise Information

Systems. ICEIS Press.

Chandrasekaran, B.; Josephson, R. What are ontologies,

and why do we need them? In IEEE Inteligent systems,

1999.

Busse S., Kutsche R., Lesser U., Weber H. 1999. Federated

information systems: concepts, terminology and

architectures. Technical Report 99-9. TU Berlin. April.

Farquhar, A., Fikes, R., Pratt, W., and Rice, J. (1995).

Collaborative ontology construction for information

integration. Technical report, Stanford University.

Fensel D., 2000. The semantic web and its languages. IEEE

Computer Society 15, 6 (November /December), 67-73

Fensel D., 2001. Ontologies: A Silver Bullet for Knowledge

Management and Electronic Commerce. Springer.FIPA

2001. Foundation for Intelligent Physical Agents.FIPA

Ontology Service Specification. March. Available on:

http://www.fipa.org/specs/fipa00086/XC00086D.html

Luck M., McBurney P, and Preist C. 2003. Agent

technology: Enabling next generation computing: A

roadmap for agent-based computing. AgentLink report,

available from

www.agentlink.org/roadmap.

Medina, M. A., Sánchez, J. A. 2002. Agents at the

reference desk: Serving information needs and

constructing knowledge for wide communities of users.

Memorias del XII Congreso Internacional de

Electrónica, Comunicaciones y Computadoras

(Conielecomp 2002, Acapulco, México, Feb.), 74-78.

Medina Ma. Auxilio, Chávez-Aragón A. Inclusion of

Reference Agents in a Virtual Reference Environment,

ENC 2003, Apizaco Tlaxcala, September 2003.

Pomerantz, J., & Silverstein, J. (2001, 13 November). The

Importance of Ontology: Standards for a Referral Tool.

Paper presented at the 3rd Annual VRD Conference:

Setting Standards and Making it Real, Orlando, FL.

Available on:

http://www.vrd.org/conferences/VRD2001

/proceedings/silverstein.shtml.

Saavedra A. 2003. Arquitectura para federación de bases

de datos documentales basada en ontologías. Disertación

doctoral. Universidade Da Coruña. Departamento de

Computación. Enero.

Sánchez, J. A., García, A. J., Proal, C., Fernández, L. 2001.

Enabling the collaborative construction and reuse of

knowledge through a virtual reference environment.

Proceedings of the Seventh International Workshop on

Groupware, (Darmstadt, Germany, Sept. 6-8). IEEE

Computer Society Press, Los Alamitos, California. 90-

97.

Van de Sompel H., Lagoze C. Notes from the

interoperability front: a progress report on the open

archives initiative. Proceedings of the 6th European

Conference on Research and Advance Technology for

Digital Libraries (ECDL 2002, Rome, Italy, September),

144-157.

Wooldridge, M., Jennings N. Kinny D. 2000. The Gaia

Methodology for Agent-Oriented Analysis and Design.

Autonomous Agents and Multi-Agent Systems, 3, 285-

312. Kluwer Academic Publishers. The Netherlands.

ICEIS 2004 - INFORMATION SYSTEMS ANALYSIS AND SPECIFICATION

560