IMPLEMENTING TRADING AGENTS FOR ADAPTABLE AND

EVOLUTIVE UI-COTS COMPONENTS ARCHITECTURES

Jos

e Andr

es Asensio, Luis Iribarne, Nicol

as Padilla

Applied Computing Group, University of Almeria, Spain

Rosa Ayala

Computers and Environmental Group, University of Almeria, Spain

Keywords:

Trading agents, COTS User Interfaces, Model Transformation, Model-Driven Engineering, Cooperative Sys-

tems, e-Business Systems Modelling.

Abstract:

Most of the complex (e-Business) information systems need to accomplish with the use of open standards.

Environmental Management Systems (EMS), for instance, state the international regulations of the ISO 14000

family, which establish the requirements to be fulﬁlled by an EMS in order to be accepted as such. One

of these requirements concerns the User Interfaces Development. Because of the variety of ﬁnal users that

interact in this sort of complex information system (politicians, technicians, administrators, and so on) and

due to a great deal of information (some critical and conﬁdential), it is important to have real and practical

scientiﬁc/technical proposals in order to build fast and efﬁcient information exploitation systems. The human-

computer interaction (HCI) of these systems need user interfaces that adapt to the users proﬁles’ habits, and

with intelligent software agents that mediate by the users in the search processes, exploitation and decision-

making tasks. In this work we present a part of the SOLERES-HCI, a framework of the Soleres Project

for developing COTS user interfaces by using trading agents. Our studies are being applied for developing

advanced EMS and approaching Model-driven engineering techniques to the UI-COTS development.

1 INTRODUCTION

Organizations and administrations that work with en-

vironmental information for speciﬁc actions such as

territory management or planning and organization of

natural resources (among others), need to have inno-

vative quality information systems in order to guaran-

tee the success of their everyday activities. There-

fore, for the experts who operate the system not only

it is important to have reliable and updated informa-

tion that helps to make the most appropriate deci-

sions, as it happens in critical complex systems. Due

to the variety of ﬁnal users that cooperate with each

other and interact with the system for decision mak-

ing (for instance, politicians, technicians, administra-

tors, etc.), it is also important to have: (a) exploita-

tion information systems (environmental, in this case)

to facilitate the human-human and human-computer

interaction and coordination; (b) intelligent user in-

terfaces that adapt to the users proﬁles’ habits and;

half of the users and facilitate the information inter-

pretation tasks, the decision-making tasks and predic-

tion/prevention tasks (which are the most important).

In this work, we present a part of the framework

Soleres-HCI that supports all the human-computer

interaction issues of a complex environmental ma-

nagement system (EMS). This portion of the frame-

work concerns the developing of adaptable and evo-

lutive user interfaces of the system by using: (a) tra-

ding agents that intercede between user agents and

the information, (b) developing of traders that fol-

low model-driven engineering perspectives, and (c)

specialized Commercial Off-The Self (COTS) com-

ponents for real-time user interface architectures.

The rest of the paper is structured as follows. Sec-

tion 2 describes some issues for modelling user inter-

faces from UI-COTS. Section 3 continues with our

SOLERES-HCI proposal, a human-computer inter-

action perspective based on trading (ISO/IEC, 1997)

and software agents. Finally, Section 4 explains some

of our ongoing research in the SOLERES project.

259

Andrés Asensio J., Iribarne L., Padilla N. and Ayala R. (2008).

IMPLEMENTING TRADING AGENTS FOR ADAPTABLE AND EVOLUTIVE UI-COTS COMPONENTS ARCHITECTURES.

In Proceedings of the International Conference on e-Business, pages 259-262

DOI: 10.5220/0001912902590262

 SciTePress

2 MODELLING UI-COTS

In recent years, different approaches for the design

of user interfaces have been presented, most of them

following model-driven UI designs, for instance:

(a) IDEAS (Interface Development Environment

within OASIS) (Lozano et al., 2000): A methodo-

logy of UI development based on UML models.

(b) OVID (Object, View and Interaction Design)

(Roberts et al., 1998): A methodology for UI de-

sign directed towards objects developed by IBM.

eration by using ConcurTaskTrees (CTT).

(d) WISDOM (Whitewater Interactive System Deve-

lopment with Object Models) (Nunes, 1998): A

methodological proposal for UML-based UI.

(e) UMLi (Pinheiro, 2002): An extension of UML

notation for the UI design.

In the SOLERES team there are also current works on

UI modelling that use and extend the UML diagrams

(Almendros and Iribarne, 2005) (Almendros and Irib-

arne, 2007). None of these works deal in depth with

trading UI-COTS modelling.

On the other hand, our research aims to study

UI following the approach of Component-based Soft-

ware Development (CBSD) specialized for COTS

components (Commercial Off-the-Shelf), (Meyers and

Oberndorf, 2001). There are really few works show-

ing realistic cases of IS development following the

COTS paradigm or using multi-component UI-COTS.

In (Iribarne et al., 2004) we developed an experiment

COTS composition in Geographical Information Sys-

tems (GIS). However this approach doesn’t solve a

dynamic and evolutive UI-COTS perspective.

Furthermore, the tendency during the last few

years in CBSD is to facilitate the automatic integra-

tion of commercial components by means of the com-

position (assembly) of their parts. The advances in

trading services (or traders) have played an impor-

tant role for it. Inspired on the trading model for

COTS components in open-distributed systems (Irib-

arne et al., 2004), (Iribarne et al., 2005) we develop a

trading agent service for UI-COTS in EMS systems.

3 TRADING IN SOLERES-HCI

Soleres HCI is the framework of the SOLERES en-

vironmental management system, specialized in the

human-computer interaction. This level of informa-

tion system follows the paradigm of Computer Sup-

ported Cooperative Work (CSCW) and uses techno-

logy of agents and multiagent architectures.

– 10/15 –

recognizing his/her preferences and definitely, evolving with time. This IU Agent will mediate

between the user and (a) the rest of the system’s users (who will have their own IU Agent), (b)

the searching of information system (next layer) directly, or (c) an environmental software

agent (also environmental research agent, MA).

Este último, se refiere a un asesor o interventor virtual que cooperará con otros agentes

dentro de una arquitectura multiagente preestablecida, respetando un modelo de organización

y de cooperación (objetivos del proyecto). El propósito de esta cooperación, y por tanto la

labor de cada agente MA, será de facilitar las tareas de explotación de la información:

interaccionando estos con el Sistema de búsqueda de información (siguiente nivel en el sistema; más

adelante ofrecemos detalles de este nivel) y filtrando la información irrelevante. La utilización

de este tipo de sistemas, es de gran utilidad por su capacidad natural para representar e

implementar aspectos organizacionales y sociales que pueden ayudar a identificar y/o resolver

conflictos.

The latter refers to a virtual consultant or supervisor who will cooperate with other agents into

the pre-established multi-agent architecture and will respect an organization and cooperation

model (this project’s aim). The purpose of this cooperation and therefore, each MA’s job, will

be to facilitate the tasks of information exploitation: they will interact with the searching

information system (next level in the system, later we will offer details of this level) and they

will filter the irrelevant information. The use of this sort of systems is rather useful because of

its natural capacity for representing and implementing organizational and social aspects that

can help to identify and/or solve conflicts.

Agent IU

Agent MA

Agent IU

Agent MA

. . .

EPU

Agent

. . .

Searching Information System

Cooperative system for the taking decision in group

(Trading-based multiagent architecture)

User User

Trading Service

(Trader)

New

information

SOLERES-HCI

(nota para Isa: traduce también el texto dentro de la figura)

Figura 3. Arquitectura general del sistema de información SOLERES.

Cooperative system for the decision taking in group

Searching information system

Figure 1: General architecture of SOLERES project.

Planner

Processes for the

generation of

environmental

information maps

Parser

EIM / EID

Environmental

Info Maps

templates

( EIM )

Original satellite

images

Agent

NVIRONMENTAL

ROCESS

NIT

NVIRONMENTAL

NFORMATION

APS

(EIM)

Comes

from the

Environmental

trading service

Environmental

Information

Documents

( EID )

RADING SERVICE

RADER

)

Activate agent

Figure 2: Some details of an EPU object.

For the process of data exploitation we identify and

structure the type of queries and the sort of informa-

tion suitable to be consulted, by using techniques of

hierarchical decomposition (i.e., trees, cut and prun-

ing) and neural networks.

Figure 1 shows the (short) architecture of the

SOLERES information system in our project frame-

work. At the user layer side (top), the system is de-

signed in order to be used for environmental decision-

making tasks and in cooperation among different peo-

ple (system’s users) organized following different or-

ganized models (i.e. depending on their hierarchy).

This human-computer interaction and human-

human interaction is guided by a cooperative system

for the decision-taking tasks in group supported by a

multi-agent architecture (next layer). Each user of the

cooperative system has an UI agent whose function is

ICE-B 2008 - International Conference on e-Business

260

Figure 3: MDA steps of the trader views.

to adapt the UI to his/her needs by identifying his/her

interaction habits with the UI. This UI agent mediates

between the user and (a) the rest of the system’s users

(who have their own UI agent), (b) the search infor-

mation system (next layer) directly, or (c) an environ-

mental software agent (environmental agent - MA).

A virtual consultant cooperates with other agents

within a pre-established multi-agent architecture and

respects a model of organization and cooperation (that

is or project’s aim). The purpose of this cooperation

(MA) is to facilitate the tasks of information exploita-

tion: they interact with the search information system

(we call an Environmental Process Unit- EPU) and

ﬁlter the irrelevant information. The use of this sort

of systems is quite useful because of its natural ca-

pacity to represent and implement organizational and

social aspects that can help to identify and/or solve

problems. Figure 2 shows internal details of an EPU.

For implementing the trading agent, we follow a

Model-Driven Engineering (MDE) perspective based

on the classical Model-driven Architecture (MDA) of

the OMG. In this perspective, the trader model stays

the three stages of the MDA: CIM/PIM/PSM (see Fig-

ure 3). A trader meta-model (MM) is deﬁned for each

stage, which describe the way to generate a diagram-

matical model (UML) of the trader.

To translate a model into views (stages), we use

model-transformation techniques. In our case, we

use ATL for implementing the transformations of the

trader. The language ATL (ATLAS transformation

language) (Jouault and Kurtev, 2006) provides declar-

ative and imperative constructs. The declarative part

of ATL is based on rules. Such rules consist of a

source pattern matched over source models and of

a target pattern that creates target models for each

match. Figure 4 shows a piece of the ATL code for

the transformation CIM/PIM of one of the ﬁve inter-

faces of a trader agent: the Lookup interface. In the

code, we can observe how associations, properties or

operations of the Lookup class are created.

rule Lookup {

from f01:in MM!Lookup

using {

O Lookup: out MM!Operation = ’null’;

O get lookup if: out MM!Operation = ’null’;

...

}

to t01:out MM!Class( name<-’Lookup’,visibility<-#public )

do { ...

O Lookup<-thisModule.create Operation

(’Lookup’, #public, ”, false, false, 1, 1);

O get lookup if<-thisModule.create Operation

(’get lookup if’, #public, ’Lookup’, false, false, 1, );

O get register if<-thisModule.create Operation

(’get register if’, #public, ’Register’, false, false, 1, 1);

A trader C Lookup C Trader<-thisModule.create Association

(’trader C Lookup C Trader’, #public, false);

Pr trader C Trader<-thisModule.create Property

(’trader’,#private,’Trader’,false, false, 1, 1, false, ”, false, false);

...

}

Figure 4: A piece of the ATL transformation

Figure 5 shows the PIM trader model: a partial

class diagram. In this framework we have also used

the GMF/Eclipse to develop a tool for documenting

CIM trader models. We use the OCL language of

UML/OMG for describing the semantical restrictions

IMPLEMENTING TRADING AGENTS FOR ADAPTABLE AND EVOLUTIVE UI-COTS COMPONENTS

ARCHITECTURES

261

Figure 5: Informal class diagram model of the trader.

of a trader. At the end, java code generation from

PSM trader models can be obtained by using a parser

written in MOFScript.

For the implementation of the described frame-

work we have used the Eclipse platform and JADE

(http://jade.tilab.com/) for agents.

For space reasons, we have included here a part

of the Soleres-HCI framework. A more complete

version of the SOLERES project and details about

the model transformations and implementations of

the trading agent (p.e., Java code) are available at

http://www.ual.es/acg/soleres.

4 ONGOING RESEARCH

The following are the speciﬁc objectives of our ongo-

ing line of research.

Firstly, we plan to study the design of intelligent

user interfaces. We will look into how to adapt the UI

to different user’s proﬁles.

Our scientiﬁc objective (complement to previous

one) is to study algorithms of dynamic-services com-

position in UI-COTS components architectures. We

try to analyze this kind of component and its mar-

ketplace (i.e., how many exist, which types, how to

deﬁne them, what kind of existing repositories, etc.).

Furthermore, starting from previous works (Irib-

arne et al., 2004), we plan to implement trading ser-

vices of UI-COTS components and their applications

in Cooperative Systems (speciﬁcally in environmental

management systems.

ACKNOWLEDGEMENTS

This work has been partially supported by the EU

(FEDER) and the Spanish MEC under grant of the

project I+D TIN2007-61497 (SOLERES. A Spatio-

Temporal Environmental Management System based

on Neural-Networks, Agents and Soft. Components).

REFERENCES

Almendros, J. and Iribarne, L. (2005). Designing GUI com-

ponents from UML use cases. In IEEE Int. Conf. on

the Engineering of Computer-Based Systems, pages

210–217. IEEE Computer Society.

Almendros, J. and Iribarne, L. (2007). Visual Languages

for Interactive Computing: Deﬁnitions and Formal-

ization, chapter User Interaction and Interface Design

with UML. Idea Group Inc. Hersey, U.S.A.

Iribarne, L., Troya, J., and Vallecillo, A. (2004). A tra-

ding service for cots components. Computer Journal,

4(3):342–357.

Iribarne, L., Troya, J., and Vallecillo, A. (2005). The De-

velopment of Component-Based Information Systems,

chapter Trading for COTS Components to Fulﬁl Ar-

chitectural Requirements. M.E. Sharpe, Inc.

ISO/IEC (1997). Information technology - open distributed

processing - trading function: Speciﬁcation. Technical

report, ISO/IEC 13235-1, ITU-T X.950.

Jouault, F. and Kurtev, I. (2006). On the architectural align-

ment of ATL and QVT. In SAC ’06: Proc. of the 2006

ACM Symp. on Applied Computing, pages 1188–1195,

New York, NY, USA. ACM.

Lozano, M., Ramos, I., and Gonz

alez, P. (2000). User in-

terface speciﬁcation and modeling in a object oriented

environment for automatic software development. In

IEEE 34th Int. Conf. on Tech. of OO Lang. and Sys-

tems, pages 373–381. IEEE Comp. SP.

Meyers, B. and Oberndorf, P. (2001). Managing Soft-

ware Acquisition. Open Systems and COTS Products.

Addison-Wesley.

Nunes, N. (1998). Object modeling for user-centered de-

velopment and user interface design: The wisdom ap-

proach. PhD thesis, Univ. de Madeira.

Paterno, F. (1999). Model-Based Design and Evaluation of

Interactive Applications. Springer.

Pinheiro, P. (2002). Object Modelling of Interactive Sys-

tems: The UMLi Approach. PhD thesis, University of

Manchester.

Roberts, D., Berry, D., Isensee, S., and Mullaly, J. (1998).

Designing for the User with OVID: Bridging User In-

terface Design and Software Engineering. New Riders

Publishing.

ICE-B 2008 - International Conference on e-Business

262