The G.O.A.L. Approach

A Goal-Oriented Algebraic Language

Jérôme Dantan

1,2

, Yann Pollet

and Salima Taibi

CEDRIC, CNAM, Paris, France

LAMSAD-Agri'terr, Esitpa-APCA, Mont-Saint-Aignan, France

Keywords: Goal, Algebra, Semantic Web, Ontology, Semantic Web Service.

Abstract: In a global context of sharing information, such as Big Data and cloud computing paradigms, researchers

are developing many means to deal with new data models and algorithms. However, the development and

reuse of these ones is complex because of the heterogeneity of environments, data formats and contexts of

use all around of the world. That's why a way to share and reuse algorithms and treatments through a

common formalism is needed, for both machines and computers. The ultimate goal of our work is to provide

a collaborative platform for not only experts but also machines automatically develop, reuse and chain

treatments, computations and models. For this, we rely on a goal-oriented approach which is associated with

the Semantic Web, to establish a common formalism to design models for worldwide researchers. In this

article, we propose the formalization of our approach thanks to an algebra which is linked to the Semantic

Web standards. Finally, we provide a high-level language dedicated to both computers and experts,

illustrated with examples that are linked to the agriculture domain.

1 INTRODUCTION

Assessing the sustainability of human activities is

becoming a worldwide major concern. In this paper,

we consider examples that belong to the agriculture

domain such as agronomy, ecology, economy,

sustainability modeling, etc. Researchers in the

agriculture domain do not only seek to assess and

improve the performances of farms, but also their

sustainability, through many indicators. The

development and computation of such indicators

require being expert in several areas. Indeed, some

aspects to be treated may be e.g. soil biology, animal

welfare, waste management, economic durability,

sales, etc.

For this, researchers from all over the world

compute data from many sources, both local and

from the Internet, that cannot be used directly, e.g.

data from RDBMS (Relational DataBase

Management Systems) and complex models.

Consequently, many computer programs and

database queries have been developed for

interpreting data. However, they are specific to each

data source but experts need to reuse the code

already written e.g. to extract data from other

sources. That's why they need a common platform

not only for experts design support but also for

computers to assist and/or automate the development

and reuse of treatments, models, etc. Large amounts

of data and treatments that are stored on the Internet

and from disparate sources must be computed,

according to a common formalism that we propose

in this paper.

The remainder of this article is structured as

follows: in section 2 we provide a state of the art. In

section 3, we present the features of our approach.

Then, in sections 4 and 5, we respectively provide

definitions and algebra we have defined. In section

6, we introduce he high-level language we have

developed, illustrated with examples in agriculture.

Finally we conclude in section 7.

2 STATE OF THE ART

Our approach involves several areas in software

engineering. We need to index treatments and data

though a common formalism. This common

formalism must be understandable for both humans

and machines. We have identified two

complementary approaches that may meet our

expectations. First of all, the goal-oriented

approaches are usually used to formalize a problem

thanks to hierarchies of goals. Then the Semantic

173

Dantan J., Pollet Y. and Taibi S..

The G.O.A.L. Approach - A Goal-Oriented Algebraic Language.

DOI: 10.5220/0004447401730180

In Proceedings of the 8th International Conference on Evaluation of Novel Approaches to Software Engineering (ENASE-2013), pages 173-180

ISBN: 978-989-8565-62-4

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

Web approaches make models understandable by

both machines and humans, with the objective to

reuse software components.

2.1 The Goal-oriented Approaches

The existing goal-oriented approaches are intended

to formalize a problem by defining the steps and

components of software design in terms of goals. A

goal is by definition a target to reach, part of the

Requirement Engineering. The way the goal is

achieved is represented by a decomposition of the

goal into sub-goals, which “consists in identifying

goals and refining them into sub-goals until the latter

can be assigned as responsibilities of single agents

such as humans, devices and software” (Letier,

2001). According to (Lapouchnian, 2005), “The

main measure of the success of a software system is

the degree to which it meets its purpose. Therefore,

identifying this purpose must be one of the main

activities in the development of software systems”.

In KAOS (Knowledge Acquisition in autOmated

Specification or Keep All Objects Satisfied), users

are able to structure the goals: each goal, except the

roots, is justified by at least another goal that

explains why the goal was introduced in the model

and each goal, except the leaves, is decomposed in

sub-goals, describing how the decomposed goal can

be reached.

More recently, (Guzelian et al., 2004) describe

the modelling of information systems added to the

methods of object design, with problems “expressed

as goals to be reached and the information system is

the result of a process of meeting these goals”.

Moreover, they present their approach by defining

meta models including decompositions of goals into

refined sub-goals, formalized thanks to UML. A

model, allowing prioritizing and organizing

components with a set of decision trees and a system

of components reuse are proposed. However, neither

algebra nor high level language for goals assessment

is presented.

Finally, the goal-oriented approaches are used in

many fields of software engineering: Requirement

Engineering, business modeling, specification of

reusable components, definition of user models and

development of interactive systems as information

systems on the Web, agent systems, information

retrieval systems, etc (Guzelian et al., 2004).

2.2 The Semantic Web

The Semantic Web (SW or Web of Data) is an

extension of the existing Web, providing access to

computers to structured collections of information

and sets of inference rules that they can use to

achieve automated reasoning (Berners-Lee et al,

2001). For this, the Semantic Web is based among

others on representations of human knowledge based

on domain ontologies and semantic web services,

which are autonomous computing entities that

compute semantically indexed data.

According to (Castellani et al, 2011), Ontologies

offer significant benefits to collaborative service-

oriented systems, such as interoperability and

reusability. They are usually used as an index to

retrieve specific data (Garcia and Celma, 2005), to

infer new knowledge (W3C, 2012) (Berners-Lee et

al., 2001), to semantically annotate multimedia data

(Castano et al., 2007), to find out Web Services

automatically (Martin et al., 2007), or to match

knowledge with other knowledge for a more general

purpose (Cruz and Nicolle, 2011). Furthermore, the

Web Ontology Language (OWL) provides model

flexibility, explicit representation of semantics, out-

of-the-box reasoning and freely available

background knowledge (Castellani et al., 2011).

To manage knowledge, the Semantic Web

Services (SWS) are based on semantic description

frameworks for Web Services. The composition of

SWS is the automated processing of Web services

autonomously simple to complex automated

processes. Then, a "generic inference mechanism

shall be developed for handling SWS" (Charif and

Sabouret, 2006). Various technologies exist, such as

OWL-S (W3C, 2004), WSMO (W3C, 2005),

METEOR-S (Sheth et al., 2008), IRS-III (Domingue

et al., 2008), etc. The main drawbacks of these

approaches are that the user must be a computer

specialist, whereas the services composition

solutions are intended to help ordinary users in the

web, and some manual steps must be performed by

the user (Charif and Sabouret, 2006).

2.3 Combining the Approaches

Our goal is not only to analyze the requirements of a

system. Our platform has also to find the ad hoc

compositions of treatments solving both users’ and

computers' requirements, to formalize models,

algorithms and data to assist or automate their reuse,

to assess them, and consequently to be both

worldwide researchers and machine understandable.

In (Dantan et al., 2012), two kinds of virtual

properties have been defined, as an extension of

OWL (W3C, 2012): these virtual properties,

attached to core OWL ontology classes, may be

hierarchically organized goals that have to be

ENASE2013-8thInternationalConferenceonEvaluationofNovelSoftwareApproachestoSoftwareEngineering

174

evaluated. We provide a goal-oriented model that is

linked to the Semantic Web properties. We combine

a core OWL ontology extended by virtual properties

with a goal ontology that contains hierarchically

organised goals and sub-goals. Indeed, we consider

that the assessment of a higher-level goal is achieved

when the assessment of its sub-goals, i.e. the goals at

a more specific level, are achieved. At the below, the

goals are raw data extraction. For this, we first

formalize a rigorous algebra based on well defined

operators and then we provide a common high-level

language for both users and computers to

automatically develop, reuse and chain the

treatments, computations and models.

3 OVERVIEW

We propose a goal-oriented model where each goal

is attached to both the domain ontology, as a “goal

property” and to a semantic goal-oriented service,

according to the following procedures:

(1) A core OWL ontology which is part of the

Semantic Web is used to model the considered

domain. We attach goal assessments as classes’

properties of the selected core ontology, which

consists of an extension of core ontology.

(2) Formalization of data models, algorithms and

non-functional properties in a common formalism

for expressing goals and their assessment, by

defining an algebra and a high level language that is

both machine and human understandable.

(3) Linking with the Semantic Web. Every data

models and algorithms are semantically indexed and

each goal property is defined thanks to an URI

(Uniform Resource Identifier). Thus, each abstract

service is considered as both a URI and a property of

classes from the core ontology. Their concrete

services are considered as their instances (objects).

We therefore structure models, algorithms and data

by using the ontology of goals that are formulated

either by the machine or the user who is not a

computer scientist. The various solutions result from

the various feasible compositions of goals into sub-

goals.

4 DEFINITIONS

In this section, we define the notions of goal, goal

property and goal-oriented service and finally we

illustrate our meta-model.

4.1 Goal

We define a goal as a semantically indexed quantity

to evaluate, that is attached to one or more elements

of real life, e.g. the soil quality of a parcel, the

sustainability of a farm, the sales of a farm, etc.

There are goals defined by composition, i.e. goals

that can be composed into sub-goals, which must be

reached for the higher level goal to be satisfied. In

other words, a process defined by composition of

sub-goals ensures a higher level goal.

4.2 Goal Property

Each goal is linked to a goal property that belongs to

an ontology. A goal property is the assessable

property that is attached to an OWL ontology class.

In other words, we define a goal property as the

projection of an ontology class to an assessable goal

property, such as a datatype property, or a virtual

property (Dantan et al., 2012) that may be either

quantitative or qualitative data.

For example, in the agriculture domain, the

evaluation of the economic durability of a farm is a

goal. The “economic durability” goal property may

be assessable via the projection of the “farm” class

to the “economic durability” goal. The “economic

durability” goal property may be assessed thanks to

(1) simple datatype properties such as the sales or

the surface of a farm, and (2) other goal properties

that are more complex, such as the soil quality

which may be assessed by several algorithms. We

therefore define two main types of goal properties:

(1) Complex: complex processes using quantitative

or qualitative data (e.g. sustainability computations,

complex statistical treatments, etc).

(2) Data: processing of raw data (e.g. subscription to

events such as live results of soil analyses, data

extraction, and aggregate operators).

Finally, a composition of goal properties is a set of

goal properties that are linked thanks to operators

such as “and”, “or”, and aggregate operators, which

leads to a higher level goal.

4.3 Goal-oriented Service

A goal-oriented service is a computer entity which is

the mean both to reach a goal and to assess a goal

property, with inputs and outputs. The function

associated to a goal-oriented service is the model,

treatment, algorithm or simply a datatype property

that evaluates the associated goal property.

The inputs of a goal-oriented service exactly

match the outputs of the sub-goal services that

TheG.O.A.L.Approach-AGoal-OrientedAlgebraicLanguage

175

compose them. Indeed, the principle of approached

matching is not able to provide a service which

actually meets to the need formulated (Pollet, 2010)

(Dantan et al., 2012). We therefore define

compositions of goal properties by enabling either

non computer scientist users or computers to manage

compositions of goal properties in a declarative way.

To do this, the experts of the considered domain first

work in an generic or abstract way, and the technical

details and data restrictions of concrete services are

not known. The discovery and composition of

services are automatically performed. Each generic

goal-oriented service is identified thanks to a URI.

The classes of the ontology are URIs and their

instances are the concrete implementations for a

particular set of data.

4.4 The Meta-model

The meta-model, which illustrated in figure 1,

contains three layers: (1) the ontology layer that is

the core ontology, extended with the goal properties;

(2) the goal layer that is the hierarchy of goals; their

associated goal properties extend the core domain

ontology; (3) the goal-oriented services layer that

contains the goal-oriented services.

Figure 1: The meta-model.

5 ALGEBRA

We define an algebra that includes operators, which

operate on both ontology classes and properties. We

have built algebraic expressions to express a goal in

a formal way.

In the definitions that will follow; let

, …,

be ontology classes and let

, …,

be one of their respectively attached

goal properties.

In the examples that will follow:

 Let “farm”, “parcel” be ontology classes.

 Let “sustainability”, “sales” “productivity”,

“economicDurability”, “area”, “culture”,

“ISO14000-sales”, “ISO14000-productivity” be

goal properties attached to both the “farm” class

and classes that are contained in the farm class.

 “sales” and “productivity” goal properties

contribute to the “economicDurability” higher

level goal property.

 Let “contains” be the contains object property

attached to the farm class.

5.1 Projection of a Class to a Property

We define a projection as a restriction of an

ontology class to one of its properties. The inputs of

the projection operator are either an ontology class

or a property of this class, which can be: (1) either a

goal property. In this case, the result of the

projection is an assessable goal property or (2) an

object property. In this case, the result of the

projection is a set of object properties.

5.1.1 Projection to a Goal Property

The projection of a class to a goal property is

denoted:

, where C is an ontology class and G

is the goal property to which the class is restricted

e.g. the projection of the farm class to its

sustainability goal property. It is denoted:

ustainability

farm

. The result is the assessable

“sustainability” goal property, attached to the farm

class.

5.1.2 Projection to an Object Property

The projection of an ontology class to an object

property is denoted:

π'

, where C is an ontology

class and O is the object property to which the class

is restricted. The projection of the farm class to its

“contains” object property is denoted:

π'

contains

farm

. The result is the set of classes that

the farm class contains, e.g. parcels, barns, etc.

5.2 Conjunctions and Disjunctions

The goal-oriented model we propose may express

various solutions to assess a given high-level goal.

These goal properties are basically described thanks

to “and” and “or” logic operators, by building

decisions trees, thus enabling the assessment of a

high-level goal property and defining a complete

process of goals/sub-goals expression. By extension,

in our model, each conjunction/disjunction of goal

property meets one-to-one to their

conjunction/disjunction of goals.

(1) The “AND” conjunction. A goal property is

ENASE2013-8thInternationalConferenceonEvaluationofNovelSoftwareApproachestoSoftwareEngineering

176

achieved when its preconditions are achieved. In fact

we define the achievement of a high level goal

property as a “and” conjunction of lower level goal

properties, denoted:

)

∧

)

. For instance,

the conjunction of the “sales” and “productivity”

farm goal properties is denoted:

ales

farm

∧

productivity

farm

. We will see that such

conjunction of goals may reach a higher level goal.

(2) The “or” disjunction: when several scenarios of

conjunctions or compositions exist, we express the

alternatives through the “or” conjunction. It means

that each alternative conjunction or composition

below is interchangeable to reach the same goal.

Two alternative compositions are denoted:

)

∨

)

. So

)

and

)

are

alternative compositions.

5.3 Chronology

It is possible to define the chronology of sub-goals

and goals, to link goals to sub-goals explicitly thanks

to a temporal conjunction. This may enable the user,

while expressing a request, to suggest its preferences

to the composition engine and to other users. The

composition engine of services can achieve the

higher level goal property only if its precondition

goal-properties (sub-goals), have been achieved. So,

a composition of sub-goals

(π

)

... π

)

leading to a higher level

goal

is denoted as follows:

(π

)

... π

)

) →π

(1)

For example, the conjunction of the farm goal

properties mentioned just before, leading to the farm

economic durability are denoted as follows:

sales

farm

∧

productivity

farm

→π

economicDurabilit

farm

(2)

5.4 Aggregates

The standard relational algebra aggregates are: sum,

average, minimum, maximum and count. Each

aggregate has a list of assessable goals for input and

has an assessed goal for output. An aggregate

function is denoted:

AGG(π

C )

, where AGG is

one of the following aggregate operators: SUM,

AVG, MIN, MAX and COUNT.

5.5 General Example

The following example is a combination of the cases

already mentioned. Suppose that the sum of the

productivities of the classes that compose a farm is

“equivalent” to the whole productivity of a farm.

The conjunction of the farm goal properties

mentioned just before, with the sum of productivities

alternative conjunction about the classes that are

contained into the farm, leading to the farm

economic durability are denoted as follows:

sal es

arm

∧

(π

productivity

arm

∨

SUM(π

productivi ty

(π

contains

arm)))

→π

economicDurabilit

farm

(3)

Note that the use of the SUM operator involves that

we take into consideration the whole classes that

may be included in the farm.

5.6 Restriction

The restriction operator selects a composition of

goal properties for given sets of goal properties. So,

the restriction is performed through a restriction

expression expressed thanks to a particular

conjunction of assessed goal properties. The

restriction of a composition of goal properties is

denoted:

(

∧

)

, where

(

∧

)

is a composition

of goals and R is the restriction expressed thanks to

a conjunction of goals, as defined above. The

restriction operator requires the evaluation of the

composition of goal properties, e.g. the restriction of

the economic durability composition, restricted to

the farms that belong to the Normandy area and that

cultivate potatoes is denoted as follows:

( π

( area= normandy)

farm

∧

(cu ltu r e = p o ta to es )

farm)

(4)

(π

ales

farm

∧

roductivit

farm →π

economicDurabilit

farm)

5.7 Operator and Property Overload

Goal properties and aggregate operators can be

overloaded either by taking into account either

theoretical implementations or new input/output

data. They are not theoretically implemented yet but

may overload the existing goal properties, e.g. by

inheritance, or by aggregate operators overloading,

taking account of specific computing methods or

new data types.

(1) The overload of a goal property is specified by

adding a "+" as exponent:

)

(2) The overload of an aggregate operator is

specified by adding a "+" as exponent:

AGG

(π

C )

Examples: (1) the “agronomicSustainability” parcel

goal property may be overloaded by the

“agronomicSustainability” goal property of a farm.

(2) The “MAX” aggregate operator that extracts the

TheG.O.A.L.Approach-AGoal-OrientedAlgebraicLanguage

177

maximum from a set of quantitative values may be

overloaded to extract the maximum from a set of

qualitative values, by implementing a new way to

compute them.

5.8 Inheritance

Inheritance is a specialization of a goal that is

different from a “restriction” which restricts a goal

property to a particular data set, as a geographical

area e.g. inheritance rather enables to declare a new

goal from another specializing it "functionally", e.g.

based on a particular standard. Sub-goals that

compose a higher level goal may correspond to

several computations in addition to potential

restrictions on a data set. The inheritance of a goal

property from another goal property can be

explicitly formulated by new operations on goal

properties. It is denoted:

C ←π

(G')

, where G'

is the goal property inherits from G.

For example, the "ISO14000-sustainability" farm

goal property inherits from the "sustainability" farm

goal property because it specializes it by achieving

their computations according to ISO14000 standards

which defines standards for environmental

management. The inheritance of farm ISO14000-

sustainability from farm sustainability is denoted as

follows:

ustainability

farm ←π

ISO14000-sustainabilit

farm

(5)

5.9 Goal-equivalence

Two conjunctions of goal properties are called

equivalent if they both contribute to the same goal

property whatever restrictions that can be applied to

them. The goal-equivalence of two goal properties is

denoted:

)

C = π

)

, where

and

are

two goal properties from two goal-equivalent

compositions. The goal-equivalence relation is an

equivalence relation (reflexive, symmetric,

transitive). The following conjunctions are goal-

equivalent:

((π

( ISO14000− sal e s)

)

∧

(π

( ISO14000− productivity )

arm))

(π

(

SO14000−

ales

)

farm

∧

SUM(π

(

SO14000−

roduct ivit

)

(π'

contai ns

farm)))

(6)

Explanation: one of the criteria of ISO 14000 is just

to retrieve all the data related to their goals that are

available and consequently the sub-goals are

theoretically always satisfied. So, they both satisfy

the higher-level goal property:

( ISO14000− economicDurability )

farm

, regardless of the

restrictions that can be applied to them.

5.10 Goal-sub-Equivalence

Two conjunctions of goal properties are called goal-

sub-equivalent if they both contribute to the same

goal property, restricted to a particular conjunction

of assessed goal properties. The goal-sub-

equivalence of two goal properties is denoted:

 

π C π C

, where C is an ontology

class,

and

are two goal properties from two

equivalent compositions. The goal-sub-equivalence

relation is an equivalence relation (reflexive,

symmetric, transitive). The following conjunctions

are goal-sub-equivalent:

 



 











  



salesfarm productivity

area= normandy culture= potatoes

country=normandy culture= potatoes

containssales productivity

σππfarm

farm π farm

π farm SUM ππ'farm



(7)

Explanation: in the area of Normandy all potatoes

fields have all the data related to their goals that is

available and consequently the sub-goals are

theoretically always satisfied. So, they both satisfy

the higher-level goal property, for farms that

cultivate potatoes in Normandy:

 



ISO14000 economicDurability

area=normandy culture=potatoes

σπfarm

farm π farm

















(8)

5.11 Priorities of Operators

Parentheses may prioritize the operators. However,

this is the list of operators in decreasing order of

priority: Projection, And, Or, Then, Restriction. The

priorities for aggregate, overload, inheritance, goal-

equivalence and goal-sub-equivalence are not

applicable because they must be explicitly defined

with parenthesis.

6 THE G.O.A.L

To compose the goals, their attached goal properties

and goal-oriented services, we have specified a both

user and machine understandable high-level

language. We have called this language Goal

Oriented Algebraic Language (G.O.A.L); it relies on

the algebraic operators we have just defined.

6.1 Syntax

Table 1 summarizes the syntax of both our algebra

and our high-level language operators.

ENASE2013-8thInternationalConferenceonEvaluationofNovelSoftwareApproachestoSoftwareEngineering

178

Table 1: Algebraic and GOAL notation.

Operator name Algebraic operator GOAL notation

Projection

. (dot)

Aggregates

SUM AVG MIN

MAX COUNT

SUM AVG MIN, MAX

COUNT

And

∧

AND

∨

Then

→

THEN

Restriction

WHERE

Overload

exponent

SUPER

Inheritance

←

EXTENDS

Goal-equivalence

EQUIVALENT_TO

Goal-sub-

Equivalence



SUBEQUIVALENT_TO

6.2 Examples

Here are some examples linked to agriculture. In the

examples that will follow:

 Let “farm”, “territory” be ontology classes.

 Let “sustainability”, “sales”,

“productivity”,“economicDurability”,

“metalRate” “pesticideRate”, “country”,

“culture”, “ISO14000-sales”, “ISO14000-

productivity”, “ISO14000-sustainability” be goal

properties attached to the “farm” class.

 Let “confidentiality” and “reliability” be non-

functionnal properties.

 “sales” and “productivity” goal properties

contribute to the “economicDurability” higher

level goal property.

 Let “contains” be the contains object property

attached to the farm class.

6.2.1 Projection, Aggregate, Conjunction

and Disjunction Operators

The generic sustainability property of the farm class

can be expressed as follows:

(farm.sales AND (farm.productivity OR

farm.contains.?).productivity)

THEN farm.economicDurability

6.2.2 Complex Goal-oriented Services

Some goal properties are expressible through more

complex statistical operators, e.g. the landscape of

an area may be inferred from a decision tree whose

leaves are sub-goals, such as metal rates and

pesticide rates:

DECISION_TREE(AVG(territory.metalRate) AND

AVG(territory.pesticideRate))

THEN territory.landscape

6.2.3 Restriction

The concrete services are expressed by adding a

“WHERE” restriction clause with functionnal

properties, following the generic term of the goal

property. Next, a non-functionnal clause such as

QoS can be specified by adding a “WITH”

restriction clause, e.g.:

((farm.sales AND (farm.productivity)

THEN farm.economicDurability)

WHERE (farm.country=France AND

farm.culture=potatoes)

WITH (farm.sales.confidentiality=high AND

farm.productivity.reliability=medium)

6.2.4 Goal-equivalence

Two conjunctions of goal properties are called

equivalent if they both contribute to the same goal

property whatever restrictions that can be applied to

them, e.g.:

(farm.ISO14000-sales AND farm.ISO14000-

productivity)

EQUIVALENT_TO

(farm.ISO14000-sales AND SUM

(farm.contains.?)).ISO14000-productivity

The compositions below are equivalent because one

of the criteria of ISO 14000 is just to own all the

data related to their sub-goals, that are theoretically

always satisfied.

6.2.5 Inheritance

The "ISO14000-sustainability" farm goal property

inherits from the "sustainability" farm goal property:

farm.ISO14000-sustainability

EXTENDS farm.sustainability

6.2.6 Goal-sub-Equivalence

Two conjunctions of goal properties that are goal-

sub-equivalent are denoted as follows:

((farm.sales AND farm.productivity)

THEN farm.durability)

SUBEQUIVALENT_TO

((farm.sales AND SUM

(farm.contains.?).productivity) THEN

farm.durability)

WHERE (farm.area=normandy AND

farm.culture=potatoes)

The compositions below are equivalent because, in

the area of Normandy, all data fields of potatoes

required for the satisfaction of all the sub-goals of

TheG.O.A.L.Approach-AGoal-OrientedAlgebraicLanguage

179

these compositions are available.

7 CONCLUSIONS

In a nutshell, we have presented the association

between a goal-oriented approach and the Semantic

Web. To formalize our approach, we have proposed

an algebra which is linked to OWL ontologies

classes and properties. Then, we have provided a

high level language called Goal-Oriented Algebraic

Language (GOAL), dedicated to both computers and

experts and based on our algebra. Finally, we have

illustrated our approach thanks to examples, linked

to agriculture. Future challenges are to provide an

ontology and actual use cases about the goal-

oriented Web services.

REFERENCES

Letier, E., 2001. Reasoning about Agents in Goal-

Oriented Requirements Engineering, Phd Thesis.

Université Catholique de Louvain, Dépt. Ingénierie

Informatique, Louvain-la-Neuve, Belgium.

Lapouchnian, A., 2005. Goal-Oriented Requirements

Engineering: An Overview of the Current Research.

Depth Report, University of Toronto.

Guzelian G., Cauvet C., Ramadour P., 2004. Conception

et réutilisation de composants : une approche par les

buts. In INFORSID 2004, 179-174.

Berners-Lee, T., Hendler, J., and Lassila, O., 2001. The

Semantic Web. Scientic American, 284(5):35-43.

Castellani, S. et al, 2011. A knownledge-based system to

support legal case construction. In Knowledge

Engineering and Ontology Development conference.

Paris, France: 15-27, may 2001.

Garcia, R., Celma, O., 2005. Semantic Integration and

Retrieval of Multimedia Metadata. In 4rd

International Semantic Web Conference, Galway,

Ireland.

World Wide Web Consortium (W3C), 2012. OWL 2 Web

Ontology Language Document Overview (Second

Edition) http://www.w3.org/TR/owl2-overview/.

Castano, S. et al, 2007. Ontology Dynamics with

Multimedia Information: The BOEMIE Evolution

Methodology. In International Workshop on Ontology

Dynamics (IWOD) ESWC 2007 Workshop. Innsbruck,

Austria.

Martin, D et al, 2007. Towards Semantic Annotations of

Web Services: OWL-S from the SAWSDL

Perspective. Proceedings of OWL-S Experiences and

Future Developments Workshop at ESWC 2007.

Innsbruck, Austria.

Cruz, C. and Nicolle, C., 2011. A graph-based tool for the

translation of xml data to OWL-DL ontologies. In

Knowledge Engineering and Ontology Development

(KEOD) IC3K 2011 conference. Paris, France, pp.

361-364.

Charif, Y., Sabouret, N., 2006. An Overview of Semantic

Web Services Composition Approaches. Electronic

Notes in Theoretical Computer Science. Volume 146,

Issue 1, 24 January 2006, 33-41.

World Wide Web Consortium (W3C), 2004. OWL-S:

Semantic Markup for Web Services

http://www.w3.org/Submission/OWL-S/.

World Wide Web Consortium (W3C), 2005. Web Service

Modeling Ontology (WSMO) http://www.w3.org/

Submission/WSMO/.

Sheth, A. P., Gomadam, K., Ranabahu, A., 2008.

Semantics enhanced services: METEOR-S, SAWSDL

and SA-REST. In IEEE Data Engineering Bulletins,

31(3):8-12, 2008.

Domingue, J. et al., 2008. IRS-III: A broker-based

approach to semantic web services. Journal of Web

Semantics, 6(2) :109-132, 2008.

Dantan, J., Pollet, Y., Taibi, S., 2012. Semantic Indexation

of Web Services for Collaborative Expert Activities,

IADIS International Conference on Information

Systems 2012

, pp.57-64.

Pollet, Y., 2010. Une approche algébrique pour la

réutilisation et l’orchestration de services dans les

systèmes d’information. Revue Ingénierie des

Systèmes d'Information. Journal Revue Ingénierie des

Systèmes d'Information, vol. 15(5), pp. 63-88, 2010.

Levy, N., Losavio, F., Matteo, A., Ramdane-Cherif, A.,

Hadj Salem, H., 2011. Quality Standards Ontology for

Web Service Discovery, In XXXVII Conferencia

Latinoamericana de Inforática (CLEI 2011), October

2011, pp.625-624, Quito, Ecuador.

ENASE2013-8thInternationalConferenceonEvaluationofNovelSoftwareApproachestoSoftwareEngineering

180