A CONCERN-ORIENTED AND ONTOLOGY-BASED APPROACH

TO CONSTRUCTING FACETS OF INFORMATION SYSTEMS

Crenguţa Bogdan

Faculty of Mathematics and Computer Science,Ovidius University, 124 Mamaia Blvd., Constanta, Romania

Luca Dan Şerbǎnaţi

Faculty of Engineering in Foreign Languages, PolitehnicaUuniversity, 313 Spl. Independentei, Bucharest, Romania

Keywords: Information systems analysis, concern, ontology, view, facet.

Abstract: A concern-oriented analysis approach for developing information systems is presented. The method uses the

concerns of various stakeholders of an information system (IS) for partitioning the system conceptual

domain in stakeholder-oriented sub-domains. Mental representations descriptions of stakeholders’ beliefs

and knowledge related to each concern are identified and on their basis, a domain ontology can be created.

Furthermore, facets of the future IS are constructed by an abstraction mechanism applied on the domain

ontology.

1 INTRODUCTION

In this paper, we present some results we obtained in

our research on information systems analysis. We

consider an information system as an informational

model of one or more work systems belonging to an

enterprise, or in general to an organization. Such a

system gives informational support to the

represented systems. It provides functionalities of

capturing, transmitting, storing, retrieving,

manipulating, and supplying data, information, and

knowledge.

The conceptual mechanisms we generally use for

managing the IS complexity are decomposition and

composition, and refining and abstraction. We

considered them in order to propose an analysis

approach for constructing views on ISs. These

mechanisms are based on the fundamental principle

of separation of concerns that we present in what

that follows.

Separation of concerns is an old decomposing

and composing principle which partitions a system

into smaller more manageable and comprehensible

parts (Parnas, 1972). Each decomposing criterion is

derived from a concern or need belonging to a

particular area of interest.

The stakeholders of an IS are people who

influence the system development and/or use.

Stakeholders are employees, managers, customers,

suppliers, etc.

We define the (stakeholder’s) concern as a

problem-originated care of one or more stakeholders

involved in the construction or evolution in its

natural environment of an IS. The care of a

stakeholder derives from his/her interest or

responsibility in the IS’ real world, his/her thinking

to improve or modify something in this world for

better matching his/her expectations, or worrying

about something wrong or undesired could occurs.

In our approach, the specification of a concern

problem uses a pair composed from two

descriptions: a) the initial state description of the

current situation, as the stakeholder perceives it, and

b) the final state description of the situation that

matches expectations, interests, or desires of the

stakeholder. These two elements are respectively

considered as hypothesis and conclusion of the

problem specification. The problem’s initial state

contains all information and knowledge necessary to

obtain the final state of the problem and, thus, to

solve it. The high-level specification of a concern

that a stakeholder tries to solve is an association of

such a pair of states and the role the stakeholder

plays in the system.

220

Bogdan C. and Dan ¸Serb

ana¸ti L. (2007).

A CONCERN-ORIENTED AND ONTOLOGY-BASED APPROACH TO CONSTRUCTING FACETS OF INFORMATION SYSTEMS.

In Proceedings of the Second International Conference on Software and Data Technologies - Volume ISDM/WsEHST/DC, pages 220-223

DOI: 10.5220/0001326202200223

 SciTePress

2 MULTIFACETED

INFORMATION SYSTEMS

According to our approach, we can analyze an

information system as an aggregation of views that

are created by composing facets. These facets can be

explicitly constructed based on the stakeholders’

beliefs and knowledge.

We consider a belief as a state of mind about a

mental representation that symbolizes a mental

object that depends on a perception (Ferrario,

Oltramari, 2004). In the cognitive psychology, a

mental representation is defined as a psychological

mechanism that allows the reflection and the

knowledge of an entity, phenomenon, or of a state of

affairs in its absence. The condition is that, this was

previously perceived in the real world (Zlate, 2004).

There is a strong relation between knowledge

and beliefs: a credible belief accepted by all people

who are interested in, it’s a piece of knowledge.

Nevertheless, we do not consider all the beliefs

and knowledge of a stakeholder, but only those

which belong to the explanations of the cause of

problems that are related to their concerns. We

called semantic rationale such a motivation of a

concern.

We use the semantic rationales of the

stakeholders’ concerns to firstly identify their

vocabularies, and then describe their intended

meaning in order to obtain an ontology.

We proposed in (Bogdan, Şerbǎnaţi, 2006) and

(Bogdan

, Luzi, Ricci, Şerbǎnaţi, 2007) a concern-

oriented approach of IS analysis with the following

11 steps: 1) identification of stakeholders; 2)

identification of concerns; 3) concern classification;

4) identification of relations between concerns; 5)

priority of concerns solving; 6) identification of

semantic rationales; 7) identification of the concepts

used in the semantic rationales; 8) ontological

analysis of the intension of the concepts; 9) choosing

a foundational (top-level) ontology to be extended

by the new ontology; 10) classification of the

concepts conforming the foundational ontology; 11)

definition of the ontology using a formal logical

language.

The approach also includes guidelines for the

creation of the informational views on the IS under

study from the obtained domain ontology. For this,

other four steps should be added to the basic

method: 12) construction of the UML ontological

model of each piece of knowledge or belief; 13)

construction of facets for each concern rationale; 14)

the analysis of the independence degree of the

facets; and 15) construction of the informational

view by grouping facets of some related concerns.

This paper is focused on the steps 12)-15) of the

method.

2.1 Analysis of Concerns

In order to identify the concerns and their relations

in the development process of an IS, our approach

firstly recommends to analyze the stakeholders’

preoccupations, interests and beliefs, and identify

how they generate concerns, in other words how the

stakeholders reason.

The analysis of the concerns according to the

above four perspectives begins with the construction

of their high-level specifications composed from a

problem specification and the roles of stakeholders

that manifest that concern.

For each concern, the beliefs and pieces of

knowledge that constitute their semantic rationales

have to be identified. For this we can use the

problem specification, the stakeholder’s work

practice, and his/her explicit and tacit knowledge.

2.2 Building an Ontology

The mental representations of the stakeholders’

knowledge and beliefs are formed by concepts that

refer instances belonging to three categories:

physical entities and their relations in the real world,

ad hoc conceptualizations resulted from the

stakeholder’s experience, and abstract (non-physical

or social) entities that were produced by the human

mind and are shared by various communities.

The identification of concepts from every

semantic rationale represents the activity in which a

vocabulary is created. The vocabulary is a set of

concepts that we use them in order to refer concrete

and abstract entities as well as relationships between

them from the domains associated to the problems

related to the identified concerns. From each

concern rationale the participating concepts are

gathered in the vocabulary. In our approach the

vocabulary is used for solving the problem

associated to the concern. This activity is repeated

until the whole conceptual domain of the problems

associated to the concerns shared between

stakeholders is obtained.

Furthermore, the concepts are ontologically

analyzed according to the OntoClean methodology

in order to obtain a backbone taxonomy based on a

combination of some properties, like rigidity,

identity, and dependence (Guarino, Welty, 2004).

Then the foundational ontology is chosen.

Subsequently a new taxonomy is created by

subsuming the existing categories in the

foundational ontology taxonomy. In addition, on the

basis of the conceptualization of foundational

A CONCERN-ORIENTED AND ONTOLOGY-BASED APPROACH TO CONSTRUCTING FACETS OF

INFORMATION SYSTEMS

221

ontology, the domain ontology is created by

formally describing the intension of each concept

and their intentional relations.

During our research, we used the top-level

ontology DOLCE (Masolo, Borgo, Gangemi,

Guarino, Oltramari, 2003) and one of its modules

D&S (Gangemi, Mika, 2003). Other top-level

ontologies might be used.

2.3 Views and Facets

In our approach a view is a model of an IS related to

a particular, homogeneous from a logical point of

view, set of concerns. We consider that the concerns

emerge from a particular perspective of the IS

developing process: social, functional,

informational, or technological perspective.

Therefore, depending on the perspective applied, we

obtain social, functional, informational, or

technological views. The views are models of a

future or existing IS resulting from a projection of

the system in a large area of concerns belonging to

more stakeholder roles. In this paper, we consider

only the informational views.

An informational view is a structural model of

the system to be modelled, basically a UML class

diagram (UML, 2003). It contains categories in the

system’s conceptual domain, their relations, as well

as constraints regarding the model interpretation.

We also consider an informational view as a

cluster of facets. Each facet is a simplified model of

the informational view and conceptually represents a

concern-driven abstraction of the informational view

according to a stakeholder’s paradigm. This

paradigm is shaped in time by stakeholders playing

the same role or having the same responsibilities.

Confronted with similar situations the stakeholders

manifest similar concerns and build similar solutions

for solving these concerns. We can say that a facet

describes the semantic rationale of a concern.

Technically, a facet is constructed according to a

template that contains the following fields: the codes

of the facet and concern, the dependency graph of

beliefs and knowledge for a semantic rationale of the

concern and the facet semantics represented as a

UML class diagram that contains the participating

concepts and their ontological relations extracted

from the domain ontology.

The template associates the semantic rationale of

the concern to the concern semantics as it is derived

the knowledge and beliefs of a concern’s semantic

rationale. In the next subsection, the UML

ontological model is defined.

2.3.1 UML Ontological Models

An UML ontological model is a class diagram that

semi-formally describes the semantic of a piece of

knowledge or belief of a concern’s rationale. Such

model is constructed from the domain ontology of

an IS trying to preserve its semantic. It uses concepts

of the UML metamodel like class, data type,

association, and dependence (UML, 2003). In our

research, we found that the correspondence between

these concepts and the categories and conceptual

relations of the domain ontology (constructed using

the top-level ontology DOLCE+D&S) is expressed

in the following rules:

1. All categories of the domain ontology, excepting

the abstracts and formal roles, are mapped to

classes.

2. The material roles are mapped to association

classes and the formal ones are mapped to

association roles.

3. Categories subsumed by abstract category are

mapped to the data type UML concept. A data

type is a type whose values have no identity

(UML, 2003).

4. All ontological relations, excepting parthood,

constitution, and subsumption are mapped in

associations in UML ontological models. An

association is a relation that describes semantic

connections between individuals that are

instances of the given classes (UML, 2003).

5. Temporal and temporary parthood, also

constitution relations are mapped in UML

aggregation relations (UML, 2003).

6. The subsumption relation is mapped to the UML

generalization/specialization relation. As it is

known, the subsumption relation holds between

two categories A and B (and we say, “A

subsumes B”) of an ontology if and only if, for all

possible states of affairs, all the instances of B are

also instances of A. Using UML language we

express the same semantic saying that B is a

subclass of A.

In order to construct an UML ontological model

of a piece of knowledge or belief, our approach

proposes the applying of the abstraction mechanism

and the following rules on its mental representation

description in the natural language:

1. If a concept from the mental representation

corresponds to a category from the domain

ontology, we map this category and the category

or categories from the foundational ontology that

subsume it into classes or data types.

2. If a concept corresponds to a quality from

foundational or domain ontology, the model will

contain the corresponding class and, in addition,

ICSOFT 2007 - International Conference on Software and Data Technologies

222

the class or classes that map the category or

categories in which the quality inheres in. We

inferred this rule from the fact of, according to

DOLCE, each quality is specifically, and

constantly dependent on the entity it inheres in

(Masolo, Borgo, Gangemi, Guarino, Oltramari,

2003).

3. In the case of a relation between two concepts, we

check if it is an ontological relation. If so, we

transform the relation into an UML one,

according to the rules 3-6 above enumerated.

4. If the relation between two concepts, excepting

the causality one, is not an ontological one, the

domain ontology is traversed on the basis of the

subsumption relation of the corresponding

categories and, on the basis of the reasoning

supplied by the ontology, we search the

ontological relation that has the same meaning

with the initial relation.

5. The causal relation between two concepts is

described in the model by the dependence

relation.

After we constructed the UML ontological

models of beliefs and knowledge of a concern

rationale, we can construct its facet. In order to

construct the facet structure, we apply again the

abstraction mechanism on the UML ontological

models and we only take the classes that will be part

of the future informational system and their

ontological relations.

3 CONCLUSIONS

In this paper, we have presented an approach based

on a concern-oriented analysis aimed to construct an

information system as a composition of multi-

facetted views. A view is an aggregation of the

facets of related concerns of the stakeholders.

The facets are constructed on the basis of

domain ontology by composing UML ontological

models of the beliefs and pieces of knowledge’s

mental representation descriptions of a stakeholder’s

concern.

The approach has been applied on a sub-process

of the clinical trial, namely the identification of the

subject selection criteria, but from the limited space

reason, we don’t present the example as a case study

in this paper.

REFERENCES

Bogdan, C., Şerbǎnaţi, L. D., 2006. Toward a Concern-

Oriented Analysis Method for Enterprise Information

Systems. In Proceedings of the IEEE International

Multi-Conference on Computing in the Global

Information Technology (ICCGI 2006), Bucharest,

Romania, IEEE Computer Society.

Bogdan, C., Luzi, D., Ricci, F.L., Şerbǎnaţi, L.D., 2007.

Towards an Ontology using a Concern-Oriented

Approach for Information Systems Analysis,

Enterprise Interoperability II, New Challenges and

Approaches, Gonçalves, R.J.; Müller, J.P.; Mertins,

K.; Zelm, M. Springer-Verlag.

Ferrario, R., Oltramari, A., 2004. Towards a

Computational Ontology of Mind. In Proceedings of

Formal Ontology in Information Systems (FOIS 2004)

IOS Press.

Gangemi, A., Mika, P., 2003. Understanding the Semantic

Web through Descriptions and Situations. In

Proceedings of the International Conference

ODBASE03, Italy, Springer.

Guarino, N., 1998. Formal Ontology and Information

System. In Proceedings of FOIS’98, Trento, Italy, IOS

Press.

Guarino, N., Welty, C., A., 2004. An Overview of

OntoClean. In S. Staab and R. Studer, editors,

Handbook on Ontologies in Inf. Sys., Springer.

Masolo, C., Borgo, S., Gangemi, A., Guarino, N.,

Oltramari, A., 2003. WonderWeb Deliverable D18.

Ontology Library. IST Project 2001-33052

WonderWeb: Ontology Infrastructure for the Semantic

Web

Parnas, D., L., 1972. On the Criteria to Be Used in

Decomposing Systems into Modules. In

Communications of the ACM, vol. 15, nr. 12.

UML, 2003. OMG, Unified Modelling Language

Superstructure, version 2.0, ptc/03-0802, 2003

Zlate, M., 2004. Psihologia Mecanismelor Cognitive,

Polirom.

A CONCERN-ORIENTED AND ONTOLOGY-BASED APPROACH TO CONSTRUCTING FACETS OF

INFORMATION SYSTEMS

223