BUILDING CLASS DIAGRAMS SYSTEMATICALLY

M.J. Escalona

Department of Computer Languages and Systems. University of Seville. Spain

J.L. Cavarero

Laboratoire CNRS / I3S, Sophia – Antipolis. University of Nice, France

Keywords: Static model, class diagram, web

methodology, systematic process

Abstract: The class diagram has become more important since the object-oriented paradigm has acquired more

acceptance. This importance has been translated also in the new field of web engineering. However, in a lot

of cases, it is not easy to get the best class diagram in a problem. For this reason, it is necessary to offer

systematic processes (as cheaper and easier as possible) to give a suitable reference to the development

team. This work presents two different processes developed in the University of Nice and in the University

of Seville and applies them to the same problem comparing the results and getting some important

conclusions.

1 INTRODUCTION

The object oriented paradigm has been the most

accepted in the last years to model software

information systems. The class diagram has been

assumed by, not only the research community, but

also by the company environment as the best model

to represent the static model of information systems.

The importance of the class diagram is also very

vious in web development (Desphande et al.,

2002). In the last years, most of the groups working

in web methodologies have accepted the class

diagram as the way to represent the conceptual data

model, that is, the static model in web environment

(Barry & Lang, 2001), (Koch, 2001) (Escalona &

Koch, 2004). Moreover, the developer who applies a

class diagram to represent his system can very often

discover that it is not easy to get a good class model.

(Cachero & Koch, 2002) (Insfran et al., 2002). For

this reason, the research community is looking for

systematic processes, ways and heuristics, which

help the developer to build the class diagram. But,

the reality is quite different, if we achieve a

comparative study on the actual methodologies

(Barry & Lang, 2001), (Koch, 2001) (Escalona &

Koch, 2004), we can deduce that they do not offer

suitable, easy and cheap processes to build the class

diagram from the requirement catalogue.

In this work, we present two systematic processes to

get the class diagram

from requirements. These

processes were applied to a real project develop in

Seville, which is presented in the third section, and

very interesting results were obtained.

2 SYSTEMATIC PROCESSES TO

BUILD CLASS DIAGRAMS

If we analyse the different proposals in software

engineering and also in web engineering, we can

discover that all methodologies give a great

importance to this model (Barry & Lang, 2001),

(Koch, 2001) (Escalona & Koch, 2004). For this

reason, some research groups have worked in

developing systematic processes, which help in a

very easy, cheap and fast way to get this model.

Actually, there are very few systematic processes At

his point, we are going to present two proposals

which offer algorithms to get a class diagram in a

systematic and, even, automatic way.

2.1 NDT- Navigational development techniques

NDT(Escalona, 2004) is a methodological proposal

to specify a

nd analyse web information systems.

NDT development process can be defined as a

378

J.Escalona M. and L. Cavarero J. (2005).

BUILDING CLASS DIAGRAMS SYSTEMATICALLY.

In Proceedings of the Seventh Inter national Conference on Enterprise Information Systems, pages 378-381

DOI: 10.5220/0002511103780381

 SciTePress

bottom-up process. The development process is

focused on a very detailed definition of requirements

guided by objectives, which covers three subphases:

requirements capture, requirements definition and

requirements validation. The NDT process only

covers the first phase in the life cycle. It is a bottom-

up process where models are independent. Also, it is

important to note that the processes to get analysis

models from requirements are systematic. They can

even be automatic if the development team uses

NDT-Tool (Escalona et al., 2003, 2004), which is a

case tool that supports the NDT development

process.

2.2.1 From storage requirements to the basic

conceptual model

In this work, only one of the systematic processes in

NDT is presented. This process allows you to get the

basic conceptual model starting with the storage

information requirements definition. At this point,

we are going to explain this process in a global way

(more information in (Escalona, 2004)).

As we said, in NDT, each requirement is dealt in

a different way depending on its type. One of the

most important is the group of storage information

requirements. These requirements describe the

information which the system deals with. To

describe this kind of requirements, NDT proposes a

special pattern. In table 1, an example of this pattern

is presented. The process is quite simple, each

storage requirement pattern expresses which

information should be stored in the system as an

abstract concept. For example, in table 1, we show

the information that the system has to store for each

monument. One of the most important fields in this

pattern is the specific data. The specific data field

describes which concrete information is stored and

its nature. For instance, for each monument, the

system stores the name. Its nature is String. The

nature can be of two kinds in NDT: the basic natures

or other storage requirements. Basic natures can be

String, Number, etc. When a specific data has a

nature of another storage requirement, it means that

the structure of this specific data is described in the

pattern of this storage requirement. For instance, in

table 1 the specific data “Conservation Data” has a

nature SR-02. It expresses that its structure is

described in the pattern SR-02.

Basically, in the systematic process, each storage

pattern produces a class. The specific data with basic

nature are translated into an attribute in this class.

The specific data with nature of another storage

requirement are translated in associations.

Table 1: A storage requirement definition.

SR-01 Information of the monument

Associated

objectives

OBJ-01: To manage the information of monuments

Description The system must store information about monuments.

Specifically:

Name and description Nature

Name: it’s the name of the monument. String

Address: it’s the address of the monument. String

… …

Typology: it’s the set of the monument

typologies.

String

Cardinaly:0..n

Image: it’s a set of images where the

monument appears.

Image

Cardinality: 0..n

Specific data

Conservation data: it stores the piece

conservation studies.

SR-02

Cardinality: 0..n

2.2 The Composition method

The Composition method (Cavarero & Lecat, 2000)

is a systematic process that starts with attributes and

methods. It proposes 4 steps to get the distribution of

these attributes and methods in classes, in the most

efficient way.

The process starts by identifying the methods and

attributes and by designing a matrix. The methods

are in columns and the attributes are put in rows.

Then, it is necessary to study how each method deals

with each attribute. If a method modifies one

attribute, a symbol O is included. If a method

consults but not modifies one attribute, we express

the relation with X. In our example, attributes and

methods are identified with NDT, so it is easier to

design the matrix which is presented in table 2.

Table 2: A part of the matrix.

Attribute m1 m2 .... m15 m16

Town X X X X

City X X X X

Authors

Characterizati

X X X X

Code X X X X

......

Type O O

After designing the matrix, the next step is to study

the methods which modify any attribute. Each of

these methods is put in one class with the attributes

that they modify. In the third step, if two classes

have the same attributes or if the set of attributes in

one class is contained in another one, the two classes

are joined. After that, classes with common

attributes, but not contained, are connected. For

instance, if class A has the attributes A1, A2 and

class B has the attributes A1, A3, then attribute A1

is deleted in class B and in class A, a new public

BUILDING CLASS DIAGRAMS SYSTEMATICALLY

379

Figure 1: The result of the application of NDT and the composition method.

method, getA1(), is added (to make it possible for

the method in B to read A1) and a relation between

A and B is created. In the last step, we add in the

diagram the methods that only consult attributes.

These methods must be added by trying to get the

least new relations as possible.

3 AN EXAMPLE: THE SYSTEM

TO MANAGE MONUMENTS

Since the last years, a lot of people and tourists are

interested in getting information about monuments

and cultural heritage. In Andalusia, in the south of

Spain, there is a public organization which manages

the information about monuments: the Andalusian

Institute of Historic Patrimony (IAPH, 2004). Some

years ago, this organization started to develop the

Information System of Andalusian Historical

Patrimony (SIPHA, 2004). This system stores

information about monuments in very different

media: image, text, documents, etc. It is a very big

system because in Andalusia there are a lot of

monuments. The system stores about 130 attributes

for each monument where we can find data like the

name of each monument, the name of its authors,

images of the monuments, etc. This amount of data

makes it necessary to design a class diagram very

efficiently to get the most suitable results.

4 BUILDING THE CLASS DIAGRAM

Starting with the example of monuments, which was

completely specified by NDT, we reduced the

number of attributes and methods, working only

with the most significant We applied the NDT

method and the Composition method on a part of

this system, which contains originally 40 attributes

and 16 methods. The name of these attributes and

methods are quite intuitive, so we are not going to

explain all of them because the most interesting

issue is to show the results.

Figure 2: The final model.

ICEIS 2005 - INFORMATION SYSTEMS ANALYSIS AND SPECIFICATION

380

In NDT the work started with the storage

information requirements patterns. The problem of

monuments was completely developed with NDT in

(Escalona et al., 2004)). In table 1, we can observe

one of them. From them the conceptual model A in

figure 1 was gotten. To apply the composition

method, the first task was to build the matrix. In

table 2 we present a part of the matrix for this

system. Applying the algorithm, we got the

conceptual model B in the figure 1. After the

application of the algorithms and, if we continue the

process in NDT and in the Composition method, we

can find that the final model is the same for both

methods. The classes Piece, PieceDescription,

PieceSituation and also PieceAnalysis, are joined in

the model of the Composition method and the

cardinalities of the other associations are the same as

in NDT. In NDT, the final distribution of the

methods is the same as in the Composition method.

In this way, the final model in both techniques is

shown in figure 2.

5 CONCLUSIONS

The necessity of offering systematic process and

tools to get class diagram, in an easy and cheap way,

has been detected by several research groups. This

work has presented two proposals that offer a guide

reference to build class diagrams.

The first one, NDT, is focused in patterns,

specific data and relations. The second one is

focused in methods and attributes. After applying

both proposals, it is very interesting to observe that

the results are quite similar, although they are

focused in different aspects. Moreover, the final

result in each of them is the same and also similar to

the real solution found in the IAPH.

These similar results allow us to guarantee that

both proposals are based on correct processes, they

only offer different ways to build class diagrams.

The application of both models is not difficult. Each

of them offers tools that allow us to apply them

automatically. For these reasons, in very complex

systems, with a lot of attributes, methods, uses cases

or storage requirements, it could be interesting to

apply both of them and compare the results.

REFERENCES

Barry, C. & Lang, M. (2001) A Survey of Multimedia and

Web Development Techniques and Methodology

Usage. IEEE Multimedia. April-June 2001, 52-56.

Booch, G., Rumbaugh, J., Jacobson, I. Unified Modeling

Language User Guide. Ed. Addison-Wesley, 1999.

Cachero, C. , Koch, N. Conceptual Navigation Analysis : a

Device and Plataform Independent Navigation

Specification. 2

IWOST. June, 2002.

Cavarero, J.L, Lecat, R. La conception orientée object,

évidence ou fatalité. Technosup. Les filiéres

technologiques des enseignements supérieurs. Ed.

Ellipses. France 2000.

Deshpande, Y., Marugesan, S., Ginige,A., Hanse,S.,

Schawabe,D., Gaedke, M, B. White. Web

Engineering. Journal of Web Engineering. V. 1 N 1.

2002. Rinton Press

Durán A., Bernárdez, B., Ruiz, A., Toro M. (1999). A

Requirements Elicitation Approach Based in

Templates and Patterns. Workshop de Engenharia de

Reqisitos. Buenos Aires, Argentina.

Escalona, M.J. Modelos y técnicas para la especificación y

análisis de la navegación en sistemas software. Ph-

European Thesis. Dpto. Lenguajes y Sistemas

Informáticos. 10/2004. Sevilla, Spain.

Escalona, M.J., Koch, N Requeriments Engineering for

Web Applications: A Comparative Study. Journal of

Web Engineering”. pp.193-202. Volumen: 2, Nº·3

2004. Rinton Press. New Jersey, Febrero 2004.

Escalona M.J, Mejías M, Torres J, Reina A.M. NDT-

Tool: A tool case to deal with requirements in web

information systems. Proceedings of IV International

Conferences on Web Engineering. ICWE 2003. LNCS

2722. 2003.

M Escalona M.J, Mejías M, Torres J. Developing systems

with NDT & NDT-Tool. 13

International Conference

on Information Systems Development: Methods and

Tools, Theory and Practice. pp.149-159. Vilna,

Lituania. Septiembre 2004.

Insfrán, E., Pastor, O., Wieringa, R. Requirements

Engineering-Based Conceptual Modeling.

Requirements Engineering Journal, Vol 7 (1). 2002.

IAPH, Instituto Andaluz de Patrimonio Histórico.

Dirección General de Bienes Culturales. Consejería de

Cultura. Junta de Andalucía.

http://www.juntadeandalucia.es/cultura/iaph

Koch, N. Software Engineering for Adaptative

Hypermedia Applications. Ph. Thesis, FAST Reihe

Softwaretechnik Vol(12), Uni-Druck Publishing

Company, Munich. Germany. 2001.

Rossi, G. (1996) An Object Oriented Method for

Designing Hipermedia Applications. PHD Thesis,

Dpto. de Informática, PUC-Rio, Brazil, 1996.

Rumbaugh, J. OMT: Modelation and conception object

oriented. Masson, 1994.

SIPHA. El Sistema de Información del Patrimonio

Histórico de Andalucía.

http://www.juntadeandalucia.es/cultura/iaph/document

acion/documentacion.html

BUILDING CLASS DIAGRAMS SYSTEMATICALLY

381