SCHEDULING DECISION-MAKING USING WEB SERVICE

Maria Leonilde Rocha Varela, Sílvio do Carmo Silva

Department of Production and Systems, University of Minho, Azurém

/ Gualtar

Campus ,Guimarães

/ Braga

, Portugal

Joaquim Nunes Aparício

Department of Computer Science, New University of Lisbon,Caparica, Portugal

Keywords: Production Scheduling, XML Modeling, Web Service and remote methods invocation.

Abstract: In this paper we make a contribution for scheduling problems solving through the web, by using web

service. The XML-based modeling and communication is applied to the production scheduling activity.

Therefore, scheduling concepts, like manufacturing problems and solving methods, including corresponding

inputs and outputs are modeled using XML. This kind of data modeling is used for building our web-based

scheduling decision support system, which works as a web service under the XML-RPC protocol. This

protocol is used for invoking the implemented methods, for solving problems defined by the user, which are

local or remotely available through the Internet. New methods can be continuously incorporated in the

system’s distributed repository in an easy and interactive way.

1 INTRODUCTION

The scheduling activity in an organization seeks to

optimize the use of available production means or

resources, ensuring short time to complete jobs and,

in addition, to satisfy other important organization

objectives. Thus, it can highly contribute to good

service to customers and to high profitability of an

organization.

Production Scheduling may be defined as the

activity of allocating tasks to production resources,

or vice versa, over time. The result of this is usually

expressed in a production schedule.

With this work we make a contribution for the

better resolution process of scheduling problems by

means of a web-based decision support system. This

system has been designed and implemented as a web

service using the XML-RPC protocol and requires,

first of all, the specification and identification of

each problem to be solved, followed by the access to

resolution methods, which are available for solving

them. When there are different methods available we

can obtain alternative solutions, which should be

evaluated against specified criteria or objectives to

be reached. Thus, we are able to properly solve a

problem through the execution of one or more

scheduling methods and, subsequently, select de best

solution provided by them. These methods can either

be local or remotely available and accessible through

the Internet.

In this paper a classification framework for

scheduling problems and related concepts is used,

which is represented through XML (extensible

markup language). This kind of data modeling

allows, for instance, specifying scheduling problems

and identifying methods for their resolution and to

establish the necessary communication for the

execution of the implemented scheduling methods

through the web.

This paper is organized as follows. The next

section describes the nature of scheduling problems

and the classification model used. Section 3 starts

with a general outline of the web system architecture

then goes on dealing with the main system

functionalities, which are related to methods

searching, methods execution and new methods

insertion. Section 4 briefly refers to some related

work and finally, section 5 presents some

conclusions.

169

Leonilde Rocha Varela M., do Carmo Silva S. and Nunes Aparício J. (2004).

SCHEDULING DECISION-MAKING USING WEB SERVICE.

In Proceedings of the First International Conference on E-Business and Telecommunication Networks, pages 169-176

DOI: 10.5220/0001399601690176

 SciTePress

2 SCHEDULING PROBLEMS

2.1 Characterization Nomenclature

Due to the existence of many different scheduling

problems there is a need for a formal and systematic

manner of problem representation that can serve as a

basis for their classification. A framework for

achieving this was developed by Varela et al. (2002a

and 2002b), based on published work by Conway

(1967), Graham et al (1979), Brucker (1995),

Blazewicz (1996), and Jordan (1996), as well as on

other information presented by (Morton, 1993) and

by other authors namely (Artiba, 1997), and (Pinedo,

1995). This framework allows identifying the

underlying characteristics of each problem to be

solved, and is used as a basis for the XML-based

problem specification model used in this work.

The referred framework for problem

representation includes three classes of notation

parameters for each corresponding class of problem

characteristics. The first class of characteristics, the

α class, is related with the environment where the

production is carried out. It specifies the production

system type (α

) and, eventually, the number of

machines that exist in the system (α

). The other

classes allow specifying the interrelated

characteristics and constraints of jobs and

production resources, which is expressed by the

class β (β

… β

) of parameters, and also the

performance criterion represented by class γ. Some

important processing constraints are imposed by the

need for auxiliary resources, like robots and

transportation devices and/or the existence of

buffers, among others factors. The evaluation

criteria, the third class of parameters, may include

any kind of performance measure, including multi-

criteria measures. More information about this three-

field problem classification nomenclature is referred,

for example, in Varela et al. (2002a, 2002b and

2002c).

An example of use of this notation is

“F2|n|Cmax” which reads as: “Scheduling of non-

preemptable and independent tasks of arbitrary

processing time lengths, arriving to the system at

time zero, on a pure flow shop, with two machines,

to minimize the maximum completion time or

makespan (Cmax).

2.2 Class Model Description

Figure 1 below shows a general UML (Unified

Modeling Language) class diagram about the

scheduling problems model.

Figure 1: Class diagram of scheduling problems

model.

There are eight main classes, related to the

problem itself, as well as to the system, where the

manufacturing process is carried out, and its

resources. These include the processors or machines

and other auxiliary resources. Two additional classes

representing each task related to jobs and its

operations are also shown. In this diagram, the

default values are usually set to zero. Therefore, for

example, for the SystemType attribute, in the

System class, the default value zero corresponds to

the single machine production system that is defined

in TSystemType data type. This also includes

attributes for other systems, such as parallel

machines, flow shops, open shops, job shops and

mixed shops, under the problem classification factor

, as previously described in section 2.1.

3 PROBLEMS SOLVING

3.1 System Architecture

Figure 2 below illustrates a general outline of the

web system architecture. This web system integrates

a distributed knowledge base for supporting the

manufacturing scheduling decision-making process.

Therefore, through a web interface each methods’

server owns his particular knowledge base

component, which enables searching information

about scheduling problems and corresponding

solving methods. These methods can be local or

remotely available and accessible through the

Internet, by any of those available methods’ servers,

which are organized in a P2P (peer-to-peer)

ICETE 2004 - GLOBAL COMMUNICATION INFORMATION SYSTEMS AND SERVICES

170

network, forming a distributed knowledge repository

(Papazoglou, 2003).

Moreover, the web system enables new

knowledge insertion, including validation, and

transformation of manufacturing scheduling data

about problems and methods that can be put forward

in a user-friendly way. This interface is mainly

controlled by DTD (document type definition) and

XSL (extensible stylesheet language) documents

stored in the repository of each peer of the network.

The scheduling information is also stored in XML

documents and these documents are verified using

DTDs, before being included in the corresponding

XML repository.

User Browser

Client 1

Methods

Server 1

Methods

Server2

...

User Browser

Client 2

...

ASP

Java

Legacy

System

C++

Figure 2: Web-system architecture.

Scheduling problems as well as related concepts,

namely methods and its implementations, including

the definition of its inputs and outputs are specified

by a DTD. Elements on the problem DTD precisely

characterize a scheduling problem, meaning that in

order to interact with the system a problem must be

described according to that grammar (2002a, 2002b

and 2002c).

Some interesting XML applications, which are

more or less related with this work, are PDML

(Product Data Markup Language), RDF (Resource

Description Format) and STEPml (Harper, 2001).

Other XML specifications devoted to manufacturing

processes are JDF (Job Definition Format), PSL

(Process Specification Language), PIX-ML (Product

Information Exchange), PIF (Process Interchange

Format) and XML-based workflow (Abiteboul et al,

2000).

This web system encompasses several main

functionalities, which include knowledge insertion,

about scheduling problems and resolution methods,

and correspondent information searching. Users can

make requests for visualizing scheduling problem

classes and methods information or even browse

information about other concepts presented by the

system. The data can be shown in different views,

using existing XSL documents, adequate for each

specific visualization request. Another important

functionality is the execution of scheduling methods,

given the manufacturing scheduling problem

definition. The selection of one or more specific

methods is made through a searching process on the

distributed knowledge base of scheduling methods.

The system also enables problem results

presentation and storage.

3.2 Methods Updating

Many implementations of a given method may be

accessible through the Internet. From the point of

view of the web system two implementations of the

same method may differ if, for example, they differ

on its outputs. Moreover, not all implementations

work in the same way. Therefore, for the system to

be able to use such implementations in a

programmatic way, they must also be described

within the system. This description must include,

among other things, the address to the running

method or program and its signature, which includes

the definition of the parameters that are necessary

for its execution, i.e. the inputs, and its output

format. All this information is described in a

corresponding DTD file (Varela et al., 2002a and

2002b). Figure 3 illustrates the system interface for

defining the method’s signature for an

implementation of the branch and bound method

proposed by Ignall and Schrage in 1965 for a flow

shop problem.

Figure 3: Method signature definition.

The inputs include the definition of a parameter

n for number of jobs to be processed, a parameter m

SCHEDULING DECISION-MAKING USING WEB SERVICE

171

for number of machines in the production system,

and a set of three items in a matrix structure, which

represent the job name, the machine name and an

additional parameter p that corresponds to job

processing time. There is also the definition for the

method’s output following the same lines. This

information, once validated using the corresponding

DTD, is subsequently inserted in an XML document

in order to enable repeated executions of methods,

information retrieval and automatic generation of

interfaces for the implemented methods’ inputs and

outputs.

3.3 Methods Searching

Scheduling problems belong to a much broader class

of combinatorial optimization problems, which, in

many cases, are hard to solve, i. e. are NP-hard

problems (Ceponkus, 1999; Jordan, 1996;

Blazewicz, 1996; Brucker, 1995). In presence of

NP-hard problems we may try to relax some

constraints imposed on the original problem and

then solve the relaxed problem. The solution of the

latter may be a good approximation to the solution

of the original one. Many times we do not have a

choice and have to draw upon what we may

generally call approximation methods (French,

1982). These include both, those that we know how

near their solutions may be from optimum ones and

also a variety of heuristic methods, including those

based on meta-heuristics, which are likely to achieve

good solutions.

Examples of this kind of methods are

approximate dynamic programming or branch and

bound methods. Other approaches to obtain good or

at least satisfactory solutions, in acceptable time, are

based on the nowadays widely used local or

neighborhood search techniques, such as Genetic

Algorithms (GA), Simulated Annealing (SA), and

Tabu Search (TS), which are also known as meta-

heuristics and extended neighborhood search

techniques (Osman, 1996; Arts, 1997). There are

also other interesting types of scheduling approaches

widely used like simulation-based approaches or

bottleneck methods or even neural network or

petrinet-based approaches, among others. Methods

based on these heuristic or approximate approaches

tend to provide good results in the available time to

make decisions.

The system can be used for searching about suitable

methods for solving particular scheduling problems,

once classified through the previously described

characterization nomenclature. In the distributed

methods knowledge base the system records for

which problem class a certain method may be used.

If this information is not available the user is

encouraged to relax some of the problem constraints

in order to discover some other methods for solving

it. Figure 4 shows a small sample of methods

assigned to solve closely related scheduling problem

classes and Figure 5 presents a list of available

methods for solving some selected problem types,

which are generally characterized by occurring in

flow shop systems and by having as performance

measure the maximum completion time or

makespan.

Figure 4: Problem classes selection.

Figure 5: Methods’ implementations for solving

selected problem classes

Many scheduling methods may be more or less

adequate to solve a given class of problems and the

system gives some detailed information in order to

assist the scheduling decision-making process.

The system is able to quickly assign methods to

problems that occur in real world manufacturing

environments and solve them through the execution

of one or more appropriate solving methods, whose

implementations are local or remotely available and

accessible through the Internet, by using a certain

communication protocol for remote methods

invocation.

3.4 Methods Execution

The system here described has been designed and

implemented as a web service (http://www.w3.org)

using the XML-RPC protocol (Laurent et al., 2001)

for remote methods invocation (Varela et al., 2003a

and 2003b).

The primary motivation for this work is to be

able to provide a service that, for a given set of

parameters describing an actual scheduling problem,

i.e. a problem instance, returns a set of output values

which may be accepted as a solution for the problem

introduced, Figure 6.

To solve this task some intermediate steps need to be

accomplished. In addition to information already

illustrated and explained in the previous section we

ICETE 2004 - GLOBAL COMMUNICATION INFORMATION SYSTEMS AND SERVICES

172

provide users with access to intermediate knowledge

about methods and its implementations to better

support the scheduling decision-making process.

Figure 6: Web service.

A web service promises easy access to remote

content and application functionality using standard

mechanisms, without any dependency on the

provider's platform, the location, the service

implementation, or the data format. In a web service

a certain method accepts as input a problem

definition and returns a result in some particular

form.

Different implementations may provide results in

different formats, and the system must have a

description of them in order to format them

according to the problem output to be returned to the

client as the very last step of the service. The result

from running a method implementation on the given

problem instance can then be delivered to the client

as an XML file and/or can be transformed into some

expressive output, like a Gantt chart.

Web services use XML to encode both the

message wrapper and the content of the message

body. As a result, the integration is completely

independent of operating system, language or other

middleware product used by each component

participating in the service. The only fundamental

requirement is that each component has the ability to

process XML documents and that each node

connected in a distributed system supports HTTP as

a default transport layer.

The XML-RPC protocol is the sequence and

structure of requests and responses required to

invoke communications on a remote machine.

Several other protocols that could also be used exist,

namely SOAP (Simple Object Access Protocol),

UDDI (Universal Description, Discovery, and

Integration of business for the web), WSDL (Web

Services Description Language), or other well

known, like CORBA (Common Object Request

Broker Architecture), RMI (Remote Method

Invocation) or DCOM (Distributed Component

Object Model).

The eXtensible Markup Language provides a

vocabulary for describing remote procedure calls,

which are transmitted between computers using the

Hyper Text Transfer Protocol (HTTP). XML-RPC

clients make procedure requests to XML-RPC

servers, which return results to the XML-RPC

clients. XML-RPC clients use the same HTTP

facilities as web browser clients, and XML-RPC

servers use the same HTTP facilities as web servers.

XML-RPC requires a minimal number of HTTP

headers to be sent along with the XML method

request for solving a given problem instance.

For a better illustration of system’s

functionalities a problem instance belonging to the

previously referred F2|n|Cmax problem class will be

shown next.

Once the problem is correctly specified the

system provides solving methods information as

well as information related to the available methods’

implementation(s). This information includes the

link(s) for implemented methods’ execution and

other general information, for example, about each

method class and author, as previously shown in

section 3.3. The system also provides more detailed

information about the method and its

implementation(s), so that an easier selection of

adequate scheduling methods can be achieved for

solving the problem.

After having selected an implemented method

available for solving the current problem, we only

need to feed the system with the problem instance

data and run it. Possible implemented methods are

the Branch-and-Bound (B&B) method from Ignall

and Schrage (Ignall and Schrage, 1965; Conway,

1967), and the Johnson method or rule, as previously

shown in Figure 5. The first one belongs to the class

of exact mathematical programming methods using

the B&B technique and is an exponential time

complexity method and the Johnson rule is a simple

sequencing rule, with polynomial time complexity.

Therefore, let us consider the Johnson’s rule for

solving a problem instance with 4 jobs, which have

to be processed in a flow shop with 2 machines. The

objective, already known, consists on minimizing

the maximum completion time (Cmax). Table 1

shows the time required for processing each job j on

each machine i (p

Table 1: Scheduling problem data.

i / j job1 job2 job3 job4 job5

m1 2 3 2 6 2

m2 4 1 5 3 4

Listing 1 shows an example that joins the

headers and XML payload to form a complete

XML-RPC request for solving this problem instance.

POST /rpchandler HTTP/1.0

User-Agent: AcmeXMLRPC/1.0

Host:localhost:6001

Content-Type: text/xml

Content-Length: 873

SCHEDULING DECISION-MAKING USING WEB SERVICE

173

<?xml version=“1.0”?>

<methodName>ExactJohnson

</methodName>

<param>

</param>

</param>

<value><double>2</double></value>…

</data></array></value></param>

</params>

</methodCall>

Listing 1: An XML-RPC request.

Upon receiving an XML-RPC request, an XML-

RPC server must deliver a response to the client.

The response may take one of two forms: the result

of processing the method or a fault report, indicating

that something has gone wrong in handling the

request from the client. As with an XML-RPC

request, the response consists of HTTP headers and

an XML payload.

Listing 2 shows a complete response from an

XML-RPC server, for the considered problem

instance, including both the HTTP headers and the

XML payload.

HTTP/1.0 927 OK

Date: Tue, 8 Jun 2004 08:49:02 GMT

Server: MyCustomXMLRPCserver

Connection: close

Content-Type: text/xml

Content-Length: 893

<?xml version=“1.0” encoding=” UTF-8”?>

</value></param>

</value></param>

<value><double>2</double></value>…

</data></array></value>

</param>

</params>

</methodResponse>

Listing 2: An XML-RPC response.

Applying this Johnson method or rule

(ExactJohnson) an optimal solution can be reached

for the problem F2|5|Cmax. Figure 7 shows a Gantt

chart for this problem, which has a minimal

mekespan of 19 time units.

Figure 7: Gantt chart.

Gantt charts, like this one, are also automatically

generated by the system, given the outputs provided

by methods’ executions. This is easily achieved

because the methods’ output data is expressed in

XML documents, which enable an easy way of

outputs conversion into different desired problem

results presentation forms, including Gantt charts.

Listing 3 presents a DTD specification for modeling

problem results as Gantt charts.

<!ELEMENT gantt (problem*)>

<!ELEMENT problem (job+)>

<!ATTLIST problem

id CDATA #REQUIRED

class CDATA #REQUIRED>

<!ELEMENT job (machine+)>

<!ATTLIST job

id CDATA #REQUIRED

name CDATA #REQUIRED>

<!ELEMENT machine (start+,finish+)>

<!ATTLIST machine

id CDATA #REQUIRED

name CDATA #REQUIRED>

<!ELEMENT start (#PCDATA)>

<!ELEMENT finish (#PCDATA)>

Listing 3:

DTD specification for Gantt chart.

Listing 4 exemplifies the corresponding XML code

for the Gantt chart, which represents the given

problem results.

<?xml version="1.0" encoding="UTF-8"?>

<!DOCTYPE gantt SYSTEM "gantt.dtd">

<gantt>

<machine id="m001"

name=”m1”>

ICETE 2004 - GLOBAL COMMUNICATION INFORMATION SYSTEMS AND SERVICES

174

</machine>

</job>…

</problem>

</gantt>

Listing 4:

XML code for Gantt chart.

These time-based charts continue to be largely

used due to its expressive power, enabling a very

easy way of comparing results obtained from the

execution of several different implemented methods

available. Other alternatives for displaying those

outputs are available, including direct outputs

presentation through XML documents and tables.

4 RELATED WORK

Recently we notice a tendency for an increase in

scheduling systems accessible through the Internet.

The web systems that we came across usually

involve solvers or a community of solvers,

addressing each one the resolution of a restricted

range of scheduling problems, using specific

techniques or tools, such as mathematical

programming. Moreover, they are not usually

designed to easily incorporate new methods

implementations by users.

An example of a web system that can be used for

scheduling is the NEOS Server, developed under the

auspices of the Optimization Technology Center of

Northwestern University and Argonne National

Laboratory, for optimization problems solving. It

makes nearly 50 solvers available through a broad

variety of network interfaces. According to the

authors, although having evolved along with the web

and the Internet, it is limited to some degree by early

design decisions (http://www-neos.mcs.anl.gov/).

We can also refer the BBN’s Vishnu scheduling

system, a web-based optimisation scheduling system

(http://vishnu.bbn.com) and the FortMP, a

Mathematical Programming Solver from Mitra’s

Group at Brunel University

(http://www.brunel.ac.uk/depts/ma/research/com).

The e-OCEA, a portal for scheduling intends to

help identifying scheduling problems, to help

development of new algorithms and to conduct

benchmarks through the Internet. However, one

requirement of this system is to only consider

elements (algorithms, data sets, schedules and

modules) that are e-OCEA compatible

(http://www.ocea.li.univ-tours.fr/eocea).

The LekiNET, a prototype Internet scheduling

environment, by Benjamin P. C. et al. (Yen et al.,

2004), which has migrated from LEKIN, a flexible

job shop system, is a system that has some

similarities to our system, focusing more on cost

effective choice of scheduling agents for solving

problems. The authors propose a migration scheme

to transform existing standalone scheduling systems

to Internet scheduling agents that can communicate

with each other and solve problems beyond

individual capabilities. They treat each system as an

agent and build the relations between the systems.

Therefore, wrappers need to be specifically designed

for each system.

In our case, any method which is accessible through

the Internet, provided its signature and location are

specified within our web system, can be used, for

solving problems put by users. No further

requirements are necessary for being able to

remotely use available methods. The association of

scheduling problems to resolution methods is done

using the information available in the DKB, about

both problems and solving methods. We did not

come across to systems with identical architecture

and underlying approach, i.e. oriented for solving a

large variety of manufacturing scheduling problems

based on a continuously updatable distributed

knowledge base, which allows a network of peers to

provide the scheduling service to users and the

dynamic enlargement of the number of methods that

can be accessed.

5 CONCLUSION

In manufacturing enterprises, it is important

nowadays, as a competitive strategy, to explore and

use software applications, now becoming available

through the Internet and Intranets, for solving

scheduling problems, which can be achieved in an

easily way by using web service technology.

This work is based on an XML-based

specification framework for production scheduling

concepts modeling, which is used as specification

framework for production scheduling decision-

making through a web service. Some of the

important functions include the ability to represent

scheduling problems and the identification of

appropriate available methods for solving them.

The XML-based data modeling is used in order to

make possible flexible communication among

different scheduling applications. This modeling and

specification contributes to the improvement of the

scheduling process, by allowing an easy selection of

several alternative methods available for problem

solving, as well as an easy maintenance of the

distributed knowledge base supporting the

scheduling service. This is achieved by providing a

user-friendly way of new knowledge insertion. This

knowledge primarily includes scheduling problems

and solving methods as well as its implementations,

SCHEDULING DECISION-MAKING USING WEB SERVICE

175

and corresponding inputs and outputs form. The

implemented methods are available and accessible

through the Internet. These may be implemented on

different programming languages and running on

different platforms. Such implementations are easily

accessible through the web service here presented

for solving scheduling problems, through the

execution of local or remotely available scheduling

methods. The system allows comparing different

solutions obtained by the execution of different

methods for a same scheduling problem, and to

choose the best solution found to solve the problem,

according to a given performance measure defined.

The XML based specification can be generated and

visualized by computers in appropriate and different

ways. An important issue is that the data

representation model is general, accommodating a

large variety of scheduling problems, which may

occur in different types of manufacturing

environments.

REFERENCES

Artiba, A., 1997. Elmaghraby, S., The Planning and

Scheduling of Production Systems, Chapman & Hall.

UK.

Abiteboul, S., et al., 2000. Data on the Web - From

Relations to Semistructured Data and XML, Morgan

Kaufmann Publishers. USA.

Arts, E., Lenstra, J. K., 1997. Local Search in

Combinatorial Optimization, John Wiely & Sons.

Blazewicz, J., et al., 1996. Scheduling Computer and

Manufacturing Processes, Springer-Verlag. Germany.

Brucker, P., 1995. Scheduling Algorithms, Springer-

Verlag. Germany.

Ceponkus, A., Hoodbhoy, F., 1999. Applied XML, Wiley

Computer Publishing. USA.

Conway, R. W., Maxwell, W. L., Miller, L. W., 1967.

Theory of Scheduling, Addison-Wesley Publishing

Company, Inc. England.

French, S., 1982. Sequencing and Scheduling – An

Introduction to Mathematics of the Job-Shop, John

Wiley and Sons, Inc.

Graham, R. L., Lawler, E. L., Lenstra, J. K., Rinnooy Kan,

A. H. G., 1979. Optimization and Approximation in

Deterministic Sequencing and Scheduling: A survey.

In Annals of Discrete mathematics.

Harper, F., 2001. XML Standards and Tools, eXcelon

Corporation, USA.

Ignall, E., Schrage L., 1965. Application of the Branch-

and-Bound Technique to Some Flow-Shop Problems.

In Operations Research, 13:3.

Jordan, C., 1996. Batching and Scheduling, Springer-

Verlag. Germany.

Laurent, S., et al., 2001. Programming Web Services with

XML-RPC. O’Reilly & Associates, Inc.

Morton, T., Pentico, D., 1993. Heuristic Scheduling

Systems, John Wiley & Sons Inc. USA.

Osman, I. H., Kelly, J. P., 1996. Meta-Heuristics: Theory

and Aplications, Kluwer Academic Publishers.

Papazoglou, M.P.; Krämer, B.J.; Yang, J.; 2003.

Leveraging Web-Services and Peer-to-Peer Networks,

In: Proceedings of Advanced Information Systems

Engineering, 15

International Conference, CaiSE,

Klagenfurt, Austria. June 16-18, pp. 485-501.

Pinedo, M., 1995. Scheduling Theory, Algorithms and

Systems, Prentice-Hall Inc. USA.

Varela, L., Aparício, J., Silva, S, 2002a. An XML

Knowledge Base System for Scheduling Problems. In

IICS’02 - Proceedings of the Innovative Internet

Computing System, Kuhlungsborn. Germany. Springer-

Verlag in the Lecture Notes in Computer Science

series.

Varela, M. L. R.; Aparício, J. N.; Silva, S. C., 2002b.

Developing a Web Scheduling System Based on XML

Modeling, In: Knowledge and Technology Integration

in Product and Services – Balancing Knowledge and

Technology in Product and Service Life Cycle,

BASYS’02, Cancun, Mexico. Kluwer Academic

Publishers, pp. 61-70.

Varela, L., Aparício, J., Silva, s., 2002c. Scheduling

Problems Modeling with XML. In The Proceedings of

the 4

International Meeting for Research in Logistics,

Lisbon. Portugal.

Varela, M. L. R.; Aparício, J. N.; Silva, S. C., 2003a. A

Scheduling Web Service based on XML-RPC, In:

Proceedings of the ICAPS-03 Workshop on Planning

for Web Services, Trento, Italy, pg. 43-50.

Varela, M. L. R.; Aparício, J. N.; Silva, S. C., 2003b A

Scheduling Web Service, In: Proceedings of the 1st

Multidisciplinary International Conference on

Scheduling: Theory and Applications, MISTA’03,

Nottingham, UK. ASAP, The University of

Nottingham, pg. 540-551.

ICETE 2004 - GLOBAL COMMUNICATION INFORMATION SYSTEMS AND SERVICES

176