FROM ONTOLOGY CHARTS TO CLASS DIAGRAMS

semantic analysis aiding systems design

Rodrigo Bonacin

, M Cecilia C. Baranauskas

, Kecheng Liu

Institute of Computing, State University of Campinas, Caixa Postal 6176 13083 970 Campinas, SP - Brazil

Department of Computer Science, University of Reading, Reading, RG6 6AY, UK

Keywords: Organisational Semiotics, Semantic Analysis, UML, Unified Process

Abstract: Despite the wide adoption of the Object Oriented paradigm for software development and the usefulness of the

Unified Modelling Language, there still are aspects of business modelling not well captured and

represented. Previous literature in Organisational Semiotics has shown that its methods could facilitate a

converging process for reaching a semantic representation, which delivers an agreed business model. In this

paper we define a process for informing UML class diagrams with results of Semantic Analysis. We provide

a group of heuristic rules to aid the construction of a preliminary class diagram from an ontology chart.

1 INTRODUCTION

Information and communication technologies are the

new power for innovation in companies nowadays.

The approach to the design of these technologies is

crucial for the adequacy of a technological artefact

in organisational contexts. Although there are

successful applications of technological artefacts in

organisations, there are also many stories of fails

(Booch, 1998). In order to cope with this problem

researchers have been developing new

methodologies and pointing out that the process for

designing systems for the organisational context is

still too far from being a solved problem.

According to Xie et al. (2

003, pp. 89)

“Understanding the business itself is the foundation

for any successful software development. For the

information systems analysts and designers,

successful communication with the domain experts

so as to properly understand, interpret, and apply

their business knowledge into software design and

implementation has always been a challenging part

of the job.”

During the last years, new standards have

merged in the software industry. Particularly the

standards based on the Object Oriented (OO)

approach became the most diffused after the

popularisation of the OO programming languages.

The Unified Modelling Language (UML) (OMG,

2003) is nowadays widely used for OO modelling by

system analysts, designers and developers. The RUP

(Kruchten, 1999), a specific and detailed instance of

a more generic process, the Unified Process (UP),

introduced by Jacobson et al. (1999) became also

widely used by the software industry.

The RUP captures many practices in software

evelopment paradigm from the Business Modelling

to the Deployment of the system. Nevertheless,

literature in the Organisational Semiotics (OS) have

pointed out some weaknesses of applying the

traditional modelling approach based on an

objectivist view to business modelling (Liu, 2000;

Stamper, 2000; Xie et al., 2003).

Assuming that the semiotic approach ca

contribute with improvements in business

modelling, we can have both: the organisational

semiotics with a different and valuable view of the

organisation on one hand and a de facto industrial

standard based on the OO approach on the other

hand.

In line with Xie et al. (

2003) who argue that

Organizational Semiotics can improve the OO

modelling, in this paper we propose steps and

heuristic rules to construct a preliminary version of a

class diagram based on outcomes from the Semantic

Analysis Method (SAM). The paper is organised in

the following way: Section 2 presents the theoretical

background; Section 3 discusses and shows how

SAM could inform the construction of class

diagrams; and Section 4 concludes.

389

Bonacin R., Cecilia C. Baranauskas M. and Liu K. (2004).

FROM ONTOLOGY CHARTS TO CLASS DIAGRAMS - semantic analysis aiding systems design.

In Proceedings of the Sixth International Conference on Enterprise Information Systems, pages 389-395

DOI: 10.5220/0002631103890395

 SciTePress

2 THEORETICAL BACKGROUND

2.1 The Semantic Analysis Method

In opposition to the objectivism, which presupposes

that there exists a world independent of the observer,

an objective reality composed by a structure of pre-

existent entities, the SAM is based on the

subjectivist paradigm (Liu, 2000), which

understands reality as a social construct based on the

behaviour of agents participating on it.

The building blocks of the Semantic Analysis

involves the concepts of affordance, ontological

dependency and agent.

Affordance, the concept introduced by Gibson

(1979) can be used to express the invariant

repertories of behaviour of an organism made

available by some combined structure of the

organism and its environment. In Semantic Analysis,

affordances are social constructs in a certain social

context (Liu, 2000). The social world acts as the

environment that is constantly affecting the agents’

behaviour, and is affected by the agents’ actions.

An ontological dependency is formed when an

affordance is possible only if certain other

affordances are available. We say that the affordance

“A” is ontological dependent on the affordance “B”

to mean that “A” exists only when “B” does.

An agent is a special kind of affordance, which

can be defined as something that performs

responsible behaviour. An agent can be an

individual person, a cultural group, a language

community, a society, etc.

The SAM addresses issues that are not

represented in any of the UML diagrams and it

provides a different way of thinking about the

organisation if compared with the Object Oriented

paradigm.

2.2 The OO Support for Business

Modelling

In the set of UML models, one diagram can support

the construction of another providing different views

of the system. According to the official UML

specification (OMG, 2003, pp. 46): “The choice of

what models and diagrams one creates has a

profound influence upon how a problem is attacked

and how a corresponding solution is shaped. (…) No

single view is sufficient”.

In the Rational Unified Process (RUP) the

business modelling has two major parts (OMG,

2003, pp. 587; Ng, 2002): (1) The business use-case

model describes the business processes and their

interaction with external parts as a sequence of

actions; (2) The business object model describes

business processes from an internal perspective.

“Whereas a business use-case model tells what a

business process will do, a business object model

tells how it will be done. It serves as an abstraction

of how business workers and business entities need

to be related and to collaborate in order to perform

the business” (Heumann, 2001, pp. 3).

2.3 Previous work on Organisational

Semiotics allied to Object

Orientation

Xie et al. (2003) have argued that RUP could be

improved with the semantic and norm analysis.

Applying RUP in a case study, they identified some

problems in business modelling that Organisational

Semiotics could help to understand and improve.

These problems are related to: “Facilities to

rigorously analyse and define the meanings of the

business entities, use of notations that help to reach

and disseminate the common understanding of the

business entities, method to reveal the fundamental

and essential relationships among the business

entities, method for responsibility-oriented workflow

analysis, criteria for the level of detail of the activity

descriptions and criteria for the termination of the

iterations” (Xie et al., 2003, pp. 94)

To maximize the benefits of SAM to the UP, we

need a method to build OO diagrams from the

concepts worked by Semantic Analysis. Liu (2000)

presented some principles of transformation from

SAM to an OO design. In other work Liu and Xie

(2003) have proposed a mapping from some

structures of the ontology chart to a class diagram:

(1) Agents and Affordances can be mapped to

Classes and some can be mapped to methods of

classes; (2) Regarding Ontological dependencies,

they can be mapped to: nested classes; the dependent

class can be included by value in the antecedent

class; the dependent can be contained in the

antecedent class as one of its methods; model the

antecedent as parameters of a method; (3) Role

names are mapped to inherited classes; (4)

Determiners are mapped to either classes or

attributes of classes.

3 BUILDING UML CLASS

DIAGRAMS INFORMED BY

SAM

We highlight the differences between the ontology

chart and the UML based business models in two

ICEIS 2004 - INFORMATION SYSTEMS ANALYSIS AND SPECIFICATION

390

aspects: (1) while the RUP business models describe

the business process, the semantic analysis is not

focusing a process view, but a description of the

organisation’s signs and the relations among them,

and (2) we have different concepts in the OO

approach and SAM and consequently the models

show different representations of reality;

In this work we review fundamental concepts

and define a process for informing the UML class

diagram with results of SAM. We provide a

sequence of steps and a group of heuristic rules to

construct a preliminary class diagram from an

ontology chart.

The proposed approach was generated and

refined during the development of Pokayoke: a

CSCW system for supporting problem solving in a

manufacturing organisation (Bonacin and

Baranauskas, 2003). This system was designed using

the Semiotic Participatory Method (SPaM),

integrating Participatory Design and Organisational

Semiotics techniques (Bonacin and Baranauskas,

2003).

3.1 Heuristics to construct Class

diagrams from Ontology Charts

We are proposing four steps to the construction of a

first version of a class diagram: (1) from the

affordances names, create a table of potential classes

and operations; (2) model the relations between

classes and operators from the relations in the

ontology chart; (3) model the attributes from the

determiners, and (4) give names to associations and

reorganise the class diagram if necessary. Figure 1

is an example of an Ontology Chart used to illustrate

the proposed steps and rules.

Starting the OO modelling after working in

Semantic Analysis, the first question to be answered

is “Where are the objects in the semantic diagrams?”

The presence of affordances in the ontology chart

suggests classes to be modelled in the class diagram;

e.g.: a department in the SAM perspective is an

affordance of the society and in the OO perspective

it is an object with internal attributes and operations.

If the affordance department is represented in the

ontology chart this suggests, from the OO

perspective, that there is such object in the context

and we can refer to this class of objects using the

name “department”. As a first heuristic we suggest

that affordance names that are “nouns” should be

translated to classes. Figure 1 shows some signs that

refer to objects from the OO perspective: society,

organisation, department, person, employee,

employer and task.

After listing some potential objects to be part of

an initial class diagram, “what do the other

affordances suggest?” For example, works is an

affordance and it could be seen as an operation of

some object, from the OO perspective. The presence

of works in the ontology chart suggests that there is

some class with the operation works. As a second

heuristic we suggest that the affordance names that

are “verbs” should be translated to operations.

A table with the potential classes and operations,

as shows Table 1, is the outcome of step 1.

Table 1: Potential Classes and Operations

Potential Classes Potential Operations

Society Works

Organisation Assigned to

Department Employs

Person Works on

Employee Responsible for

Employer

Project

Task

Step 2 involves to discover the relations among

the classes and operations to construct a first version

of a class diagram. The ontology chart can suggest

some relations among the concepts of the table; e.g.:

there is an ontology dependency between the

affordances society and person in Figure 1, and they

are classes in Table 1. The ontology dependency

suggests that there is an association between them.

“Why does the ontological dependency suggest

this kind of relation?” From the OO perspective we

could say that the object derived from the dependent

affordance (e.g. person) is created and destroyed

during the existence of the object derived from the

antecedent affordance (e.g. society). This

interpretation suggests that the object person is

possible only if the object society is also possible.

This kind of “existential” dependence cannot be

directly represented in the class diagram, since this

Figure 1: Ontology chart for project management

(

from Liu, 2000,

. 79

)

FROM ONTOLOGY CHARTS TO CLASS DIAGRAMS: SEMANTIC ANALYSIS AIDING SYSTEMS DESIGN

391

diagram do not specify when the objects are

constructed and destroyed. The use of UML

behaviour diagrams could represent lifecycle

dependency (this dependency can be stored in a list

of lifecycle dependencies and used later). As a third

heuristic we propose to model an association

between classes, whenever one object cannot exist if

another does not exist.

The case described in the last paragraph is only

one of possible cases of relations among the

elements of the group of potential classes and

operations. We have identified 10 other different

cases. Tables 2 to 5 describe the heuristic rules to be

applied for each case and the rationale behind the

rule. These cases are distributed in four groups:

Group A (Table 2). Rules to be applied to

relations between affordances whose names are

“nouns” in the ontology charts. For example, in

Figure 1, we have: society-person, society-

organization, person-employee, organization-

employer, organization-department, organization-

project and project-task;

Table 2: Rules of Group A

Rule Rule Description, Example and Rationale

a.i

If the relation is a “whole-part” in the ontology chart then the

class diagram will have a composition. (e. g. the affordance

department is part of organisation in the ontology chart and in the

class diagram organisation will be composed of department).

A whole-part relationship means that an affordance is not only

part of its antecedent but also that is ontologically dependent on

it. In OO an aggregation represents that one object is part of

another. We propose the use of “composition”, a special kind of

aggregation, which specifies the composite object is responsible

for the creation and destruction of the parts. Therefore the

lifecycle of the “part” is enclosed in the lifecycle of the composite

object.

a.ii

If the relation is “ontological dependency” there is an

association between the corresponding classes. The existential

dependency must be represented in the behaviour diagrams. (e.g.

the affordance problem is ontologically dependent on

organisation; the class diagram will have an association between

problem and organisation; the behaviour diagram of the class

problem is instantiated and destroyed only if it exists an object of

the class organisation).

An “ontological dependency” means that the dependent

affordance is only possible if we have the antecedent. From the

OO perspective we could say that the object derived from the

dependent affordance should be created and destroyed during the

existence of the object derived from the antecedent. This kind of

dependence cannot be represented in the class diagram, because it

does not specify when the objects are constructed or destroyed.

This is the reason why we propose the use of behaviour diagrams

to represent it. We also propose to model an association between

the classes.

a.iii

If the relation is a “role-name”, the class corresponding to the

role name will be a subclass of the antecedent and will have an

association with the dependent (if the dependent can not be

operation of the role name class). (e.g. the role-name employee is

dependent on person and employs; then, in the class diagram the

employee will be a sub-class of person and will have a relation

with the class that contains the operation employs)

A “role-name”, means that an agent has a specific role. From

the OO perspective we could have the object that was derived

from the “role-name” as a specialisation of the antecedent In

some cases another alternative is the use of the OO concept of

“role” .

a.iv

If the relation is a “specialisation”, it can be translated to a

hierarchic relation in the class diagram. (e.g. natural person and

corporate body are specific terms to legal person; in the class

diagram the natural person and corporate body will be sub-

classes of person).

A “specialisation” in the SAM perspective means that we have

one generic affordance and the specialised ones. In the OO

perspective we have an object that is the generic one and another

that is the specialised one, this concept is modelled as a hierarchic

relation in the class diagram.

Group B (Table 3). Rules to be applied to

relations between affordances where the antecedent

name is a “noun” and the dependent name is a

“verb”. For example, in Figure 1, we have:

employee-works, department-works, employee-

works on, task-work on, department-responsible for,

project-responsible for, project-assigned to and

employee-assigned to;

Table 3: Rules of Group B

Rule Rule Description, Example and Rationale

b.i

If the relation is an “ontological dependency” we have two

options: (1) the operation corresponding to the dependent

affordance will be part of the class definition corresponding to

one of its antecedents, or (2) if it is already part of some class, the

class diagram will have an association between this class and the

class corresponding to the antecedent affordance. The choice of

which class will contain the operation is made according to the

OO principles (e.g. walk can be an operation of people). The

existential dependency will be represented in the behaviour

diagrams. (e.g. the affordance walk is dependent of person and

surface; in the class diagram the walk will be operation of person

and an association between person and surface will be modelled

in the behaviour diagram; no execution of the method walk is

possible without an object of the class surface).

If the operation (from the dependent affordance) is part of the

definition of the class (from the antecedent affordance) the

operation is only possible if the class is possible, since there is not

operation without a class in the OO approach. If the operation is

already part of the definition of another class we propose an

association between the classes, because the operation (from the

dependent affordance) is only possible if a certain class (from the

antecedent affordance) is possible. The class diagram does not

specify when the objects are constructed and destroyed; for that

we propose the use of behaviour diagrams to represent it.

b.ii

If the relation is a “specialisation”, the class diagram will have

the specific (in the ontology chart) as operation of the generic.

(e.g. a generic affordance attitude have the specific affordances

desire, want and like; in the class diagram the class attitude will

have the operations desire, want and like)

A “specialisation” means that we have one generic affordance

and other specialised. We have not this kind of hierarchal relation

between a class and an operation in the class diagram. We

propose that the operation could be modelled as an operation of

the generic one because it suggests that they could have common

behaviour

ICEIS 2004 - INFORMATION SYSTEMS ANALYSIS AND SPECIFICATION

392

Group C (Table 4). Rules to be applied to

relations between affordances with a “verb” as

antecedent and the dependent is a “noun”. An

example of this group could be (this case did not

show up in Figure 1): the affordance error is

ontologically dependent on the affordance notify;

Table 4: Rule of Group C

Rule Rule Description, Example and Rationale

c.i

If the relation is an “ontological dependency” the class diagram

will have an association between the class (from the dependent)

and the class where the operation was specified. The existential

dependency will be represented in behaviour diagrams. (e.g. the

affordance error is ontologically dependent on notify, then there

will be an association between the class that notify is part (e.g.

mail server) and the error class)

The fact of an object exist only if an certain operation exist can

be interpreted as the object has to be created and destructed

during the operation execution (possible by chain of methods

invocations). This fact can be represented by the behaviour

diagrams of UML and invocations between classes suggest

associations between classes in the class diagram

Group D (Table 5). Rules to be applied to

relations between affordances whose names are

“verbs” in the ontology charts. An example of this

group could be (this case did not show up in Figure

1): the affordance stumble is ontologically

dependent of the affordance walk.

Table 5: Rules of Group D

Rule Rule Description, Example and Rationale

d.i

If the relation is a “whole-part”; we have two options: (1) if the

operations are part of the same class, the operation from the

antecedent will invoke the other operation, and (2) if they are part

of different classes, there will be a link between the two classes,

and the antecedent will invoke the other operation. (e.g. a part of

the affordance register (as verb) is fill out; the class diagram

could have register and fill out as part of the same class (e.g.

person), and the behaviour diagram reflects that the execution of

fill out is only possible during the execution of register)

A “whole-part” relation means that an affordance is part of

another in the SAM. From the OO perspective it could mean that

an operation is part of another operation. We propose that an

operation invoke (may be not directly) the other operation

because the invoked operation is part of the whole operation. This

aspect will be represented in the behaviour diagrams of UML. If

they are specified in different classes, we propose an association

between the classes

d.ii

If the relation is an “ontological dependency” we propose apply

the same of rule “d.i”. (e.g. the affordance stumble is

ontologically dependent on walk; in the class diagram will have

walk and stumble as part of the same class, and the behaviour

diagrams will reflect that the execution of stumble is only possible

during the execution of walk)

From the OO perspective an operation exists only if another

operation also exists. A method should execute only if another

method is also executing; for this reason we propose that the

operation from the antecedent should invoke (may be not directly)

the other operation. This aspect will be represented in the

behaviour diagrams of UML. If they are specified in different

classes we propose an association between the classes

d.iii

If the relation is a “specialisation”, we have two options: (1) if

generic and the specific affordances was translated to operations

of the same class, the specific could use the generic one in its

execution, (2) if the generic and the specific are translated to

operations of different classes, it can suggest a hierarchic relation.

(e.g. the affordance move and the specialisations walk and run; if

they are translated to the same class (e.g. person) run could

invoke the move)

In the OO perspective we could have an operation that is

generic and another operation that is the specialised one. In the

class diagram there is not hierarchic relation between operations.

The specific operation can use the generic operation to execute

some of the generic behaviour. If they are in different classes, it

could suggest that the classes have at least a common behaviour

for these operations

3.2 The Approach Illustrated

For the example of Figure 1, we could apply rule

(a.i) to represent the relations between organization-

department and project-task, rule (a.ii) to the

relations between society-person, society-

organization and organization-project; rule (a.iii) to

the relations person-employee and organization-

employer, and the rule (b.i) to the relations

employee-works, department-works, employee-

works on, task-work on, responsible-responsible for,

project-responsible for, project-assigned to and

employee-assigned to.

A class diagram generated from the applications

of the rules is the outcome of the step 2. The step 3

involves the translation of determiners to attributes

or classes. In Figure 1, the determiners function and

hourly rate were translated to attributes of employee

(see Figure 2). Determiners of affordances which

are verbs can be translated to variables of the

operations. Some determiners suggest the necessity

of new classes.

FROM ONTOLOGY CHARTS TO CLASS DIAGRAMS: SEMANTIC ANALYSIS AIDING SYSTEMS DESIGN

393

In the step 4 we propose to refine the class

diagram, by giving names to the associations and by

changing the position of the classes in the diagram

or by representing the directions of the associations.

For example an association between the

organisation and society classes could receive the

name “exits at” and the order could be changed to

facilitate reading from the OO perspective. The

choice of the association names depends on context

interpretation according to the OO perspective; for

example it is possible to give the name “have” to the

association between society and organization. Some

associations can have the same name of the

operations (e.g. works, works on and assigned to in

Figure 2), but this redundancy can be eliminated

later.

The proposed steps and rules have produced

consistent diagrams when applied to the Pokayoke

system development. It was a quick way to construct

OO models from the ontology charts. But, in some

cases the application of the rules includes the choice

of the adequate design options from the OO

perspective, and also a deeper analysis of the

produced diagram. For example: a role-name in the

ontology chart suggests a hierarchical relation or a

“role” in the class diagram. In Figure 1 the role-

name “employee” produced a subclass of person

and the role-name “employer” a subclass of

organisation (see Figure 2). We could represent the

employer as a role of the organisation in its

association with the employee. Sometimes these

relations became clear along the modelling work;

they are exceptions that we can dealt with and they

were rare in the Pokayoke context.

4 CONCLUSION

person

employer

employs()

depart ment

budget

responsible for()

task

responsable for

project

budgetemployee

function

hourly rate

works()

wor ks_on()

ass igned to()

have a

wo rk s

wo rk s o n

assi gened to

organisation

defi ne

society

live at

exists at

It has been generally agreed that business modelling

is the foundation for the design of successful

software applications. Previous literature in OS has

shown that there could be a valuable contribution of

OS methods to enhance business modelling.

This paper presented a sequence of steps and a

group of heuristic rules developed to construct class

diagrams based on ontology charts. By applying

them it is possible to produce a first draft of class

diagrams to be useful to an OO programming

approach. Further work should be done in order to

articulate the proposed approach with a broader view

of the UP. Also further research involves to develop

tools to support the construction of UML diagrams

informed by the SAM outcomes.

ACKNOWLEDGMENTS

ure 2: Class Dia

ram

roduce

This work was partially supported by grants from

Brazilian Research Council (CAPES BEX2214/02-

4, CNPq 301656/84-3 and FAPESP 2000/05460-0).

The authors also thank Delphi Automotive Systems

in Jaguariúna, Brazil, and the members of the AIS

Lab (University of Reading) for collaboration and

partnership.

REFERENCES

Bonacin, R. and Baranauskas, M. C. C., 2003, Designing

Towards Supporting and Improving co-operative

Organisational Work Practices. Proceedings of 5th

International Conference on Enterprise Information

Systems, v.3, 233-238.

Booch, G., 1998, Software Development Best Practices, in

The Rational Unified Process: an introduction, P.

Kruchten, Addison-Wesley, 3-16.

Gibson,J. J., 1968. The Ecological Approach to Visual

Perception. Houghton Miffin Company, Boston,

Massachusetts.

Heumann, J. , 2001, Introduction to Business Modeling

Using the Unified Modeling Language (UML), The

Rational Edge, Rational Software,

www.therationaledge.com

Jacobson, I., Booch, G. B. and Rumbaugh, J., 1999, The

Unified Software Development Process. Addison-

Wesley.

Ng, PW, 2002, Business Process Modeling and Simulation

with UML, The Rational Edge, Rational Software,

www.therationaledge.com

Kruchten, P., 1999, The Rational Unified Process.

Addison-Wesley.

ICEIS 2004 - INFORMATION SYSTEMS ANALYSIS AND SPECIFICATION

394

Liu, K., 2000. Semiotics in information systems

engineering, Cambridge University Press.

Liu, K. and Xie, Z., 2003, Applying Semantic Analysis: a

user’s view, Research report EPSRC-SEDITA project.

OMG, 2003, OMG Unified Modeling Language

Specification, Version 1.5. www.omg.org/uml

Stamper, R. K., 2000, Information Systems as a Social

Science: An Alternative to the FRISCO Formalism. In

Information System Concepts: an Integrated

Discipline Emerging, E. D. Falkenberg, K. Lyytien

and A. A. Verrijn-Stuart, eds, Kluwer Academic

Publishers, USA, 1-51.

Xie, Z., Liu, K., and Emmitt, D., 2003, Improving

Business Modelling with Organisational Semiotics, In

H.W.M. Gazendam, R. J. Jorna, R. S. Cijsonw (eds.)

Dynamics and change in Organizations - Studies in

Organizational Semiotics, Kluwer Academic

Publishers, 85-102.

FROM ONTOLOGY CHARTS TO CLASS DIAGRAMS: SEMANTIC ANALYSIS AIDING SYSTEMS DESIGN

395