THE HAV DATA INTEGRATION APPROACH

The Mapping in HAV

Fatima Boulçane

LIRE Laboratory

Mentouri University of Constantine

Keywords: Mediator, data integration, GAV, LAV.

Abstract: This paper provides an overview on a hybrid approach of heterogeneous data integration which we term

Hybrid As View (HAV) and it focuses on the mapping between the global schema and source schemas

through the partial schemas. The contribution of this approach is on two complementary axes: (i) to propose

a multi-mediators architecture essentially made up of two types of components: specialized mediators and a

global mediator. Each of the specialized mediators provides an integrated view of sources with the same

model. The global mediator integrates the partial schemas provided by the set of the specialized mediators

to provide an access on a uniform view represented by a global schema; (ii) to model the relation between

the global schema and the sources through the partial schemas by combining the best of the two approaches

Global As View (GAV) and Local As View (LAV).

1 INTRODUCTION

The data integration systems aim to offer a uniform

view on a set of heterogeneous sources for end-users

or applications. Data integration refers to the

problem of combining data residing at autonomous

and heterogeneous sources, and providing users with

a unified global schema (Xu et al., 2004). The

integration systems are often based on the three

levels mediation architecture proposed in (Koffina,

2005) (Figure 1). A mediator layer represents the

integration part. It interacts between the data sources

and the applications or the users. To reach the data

sources, the mediators call upon the wrappers, which

convert the data of the sources in a data model used

by the mediators.

The problems of data integration systems are

closely related to the heterogeneity of the data, with

their semantic difference, the differences in terms of

accessibility of the sources, functions offered and

availabilities of cooperation (Ullman, 1997).

We can classify the data integration systems

according to the relation between the schemas of the

local sources and the mediator unified global schema

(Lenzerini, 2002, Cali, 2002). The definition of the

global schema can be done according to the two

basic approaches: the GAV approach and the LAV

approach. Furthermore, hybrid approaches based on

both GAV and LAV have been recently proposed

(Koffina, 2005).

The GAV approach consists in defining the

global schema as a view on local schemas. The

principal advantage of this approach lies in the fact

that the rewriting of the requests is simple. On the

other hand, it is difficult to add new sources to the

system (Xu et al., 2004).

In the LAV approach, the local sources are

defined like views on the global schema. With this

approach, the problem of rewriting requests is more

Wrapper

Mediator

Application

Figure 1: Architecture of mediation.

490

Boulçane F. (2007).

THE HAV DATA INTEGRATION APPROACH - The Mapping in HAV.

In Proceedings of the Ninth International Conference on Enterprise Information Systems - DISI, pages 490-493

DOI: 10.5220/0002398004900493

 SciTePress

complex because a request on the global schema

must be reformulated according to the schemas of

the sources (Li Xu, 2004).

Both GAV and LAV have some drawbacks.

Thus, combining Global and Local approaches have

been proposed. The first one is called GLAV

(Friedman, 1999). In this approach, we are able to

express a local source in terms of the global schema

(LAV), a global source in terms of the local sources.

Query rewriting in this approach is shown to be no

harder than it is for the LAV approach (Koffina,

2005).

BAV is another data integration approach. It is a

rich integration framework, which is based on the

use of reversible sequences of primitive schema

transformations, called transformation pathways

(Boyd, 2004).

BGLaV is an alternative point of view that is

neither GAV nor LAV. The approach uses source-

to-target mappings based on a predefined conceptual

target schema, which is specified ontologically and

independently of any of the sources. The proposed

data integration system is easier to maintain than

both GAV and LAV, and query reformulation

reduces to rule unfolding (Li Xu, 2004).

The specificity of our architecture, is that in

addition to the characteristics of the current systems,

seeks to promote an architecture which improves the

integration of heterogeneous data by supporting the

combination of two approaches GAV and LAV:

HAV and to make us benefit from their advantages.

The various information integration systems

containing mediator are characterized by, on the one

hand, the languages used to model the global

schema, the schemas of the data sources to be

integrated and the requests of the users, and on the

other hand, the mapping between the global schema

and the schemas of the data sources to be integrated

(Lenzerini 2002). This paper focuses on the mapping

in HAV.

We organize the contributions in this paper as

follows. Section 2 presents an overview of HAV

approach. Section 3 will be devoted to the

presentation of the HAV formal definition,

especially the mapping. Finally, we summarize and

we release some prospects in section 4.

2 OVERVIEW OF THE HAV

APPROACH

In this section we present an overview of a multi-

mediators architecture described in a previous work

(Boulçane, 2006) essentially made up of two types

of components: specialized mediators and a global

mediator (Figure2). The architecture which rises

from the HAV approach is an architecture where

specialized mediators place themselves in the heart

of the mediation architecture. They are considered as

virtual sources to be requested by the global

mediator via the specialized wrappers.

The specialized mediators provide each one an

integrated view of sources with the same model

called partial schemas, which are integrated by the

global mediator into a global schema. It is essential

to determine the relation between these schemas and

the sources. This essentially consists in defining the

correspondence between the global schema and the

sources via the partial schemas. To build such a

system, we propose to define the partial schemas

with the LAV approach and the global schema with

the GAV approach.

Figure 2: The multi-mediators Architecture.

3 FORMAL DEFINITION OF HAV

We use the formalism suggested by (Lenzerini,

2002) for the data integration systems based on a

global schema and which we adapt to the integration

system based on the HAV approach.

Definition 1 (Lenzerini, 2002): A data integration

system is a triplet (G, S, Mg,

) where:

 G is the global schema expressed in a language

, over the alphabet A

. The language L

determines the expressiveness allowed for

Global mediato

Specialized

Adapter1

Global Sche

(

GAV

)

Specialized

Adapter2

Specialized

Adapter3

Relational Partial

Schema (LAV)

Object Partial

Schema (LAV)

XML Partial

Schema (LAV)

Specialized

Mediator 1

relational

Source

Ada

Specialized

Mediator 2

relational

Source

Ada

Object

Source

Ada

Object

Source

Ada

XML

Source

Ada

XML

Source

Ada

relational

Schema

relational

Schema

Object

Schema

XML

Schema

Xml

Schema

Specialized

Mediator 3

Object

Schema

THE HAV DATA INTEGRATION APPROACH: The Mapping in HAV

491

specifying the global schema, i.e., the set of

constraints that can be defined over it.

 S is the set of the local schemas. It is modelled

in the source language L

over the alphabet

. The language determines the set of

constraints that can be defined over it.

 Mg,

is the mapping between G and S.

For the needs of the HAV data integration

system, we recursively use the definition 1.

Definition 2: A data integration system in HAV is a

triplet (G, I

, Mg,

) where:

 G is the global schema expressed in a

language L

, over alphabet A

. The language

determines the set of constraints that can

be defined over it.

 I

, is a set of data integration systems like a

triplet (S, S

,,M

SS,S

) where:

▪ S is the schema of a specialized mediator,

expressed in a language L

on an alphabet

. The language determines the set of

constraints that can be defined over it.

▪ S

is the schema of the source with the

same model as S on an alphabet A

▪ M

SS,S

is the mapping between S and S

constituted by a set of assertions of the

form:

 Mg,

is the mapping between G and S,

constituted by a set of assertions of the form:

In other words, the global schema G provides an

integrated view of the partial schemas S, where each

one is the result schema of a specialized mediator.

Definition 3: Given the definition 2, mappings M

SS,S

and Mg,

in HAV approach are in the form:

SS,S

: S

(X) S

), S

), …, S

, Z

)

Where X = U

are relations of partial schemas

are local relations.

Mg,

: Gi(X) S

), S

), …, S

)

Where X = U

Gi are global relations

are relations of partial schemas

Example:

We call upon two relational sources S1 and S2,

and two semi-structured sources S3 and S4.The

relational sources S1 and S2 are integrated like local

views on partial relational schema PS1.The sources

S3 and S4 are integrated to give partial schema PS2

in XML. This corresponds to the integration with the

LAV approach. The global schema is relational and

is defined like a global view on partial schemas PS1

and PS2. This corresponds to the integration with the

GAV approach.

The sources to be integrated contain information

on films. The partial schema PS1 contains films

since 1960 and their criticisms since 1990. PS2

contains films. The global schema consists of two

tables. One contains information on films and the

other contains articles relating to films.

The mappings M

SS,S

We assume that the specialized shema1 (PS1)

consists of two relations:

Film (Fid, title, productor, year)

Critiques (Fid, critique)

We integrate the two sources S1 and S2 on the

specialized schema1. The description of these

sources is:

For S1:

Film (Fid, title, year, productor)

Film (Fid, title, productor, year)

Film.year>1960

For S2 :

Critique (Fid, title, critique)

Film (Fid, title, productor, year)

Critiques (Fid, critique)

Film.year>1990,

Film.Fid =Critiques.Fid

From the description above we can conclude that

the relation Film of S1contains the identifier of the

film, the title of the film, the productor of the film

and the year, only for films since 1960. While the

relation Critique of S2 contains the identifier of the

film, the title of the film and the critique of the film

since 1990.

The mappings : Mg,

We assume that there are two specialized schemas

PS1 and PS2. We suppose that after the translation

of PS2 into the relational model the relational

schema of PS2 consists of one relation:

Film (Fid, title, director, kind).

We integrate PS1 and PS2 on the global schema.

The description of the global schema is:

Films (Fid, title, realisator, year, kind)

PS1.Film(Fid, title, realisator, year, NULL)

PS2.Film(Fid, title, director, NULL, kind)

Articles (title, critique)

PS1.Film(title, Null)

PS1.Critiques(title, critique)

ICEIS 2007 - International Conference on Enterprise Information Systems

492

Example of requests on the global schema:

Which are criticisms of films having for title:

‘Freedom’?

We send the following request to the global schema:

Select Films.title, Articles.critique

From Films, Article

Where Films.title = ` Freedom' And

Films.title = Articles.title

The translation of the request over PS1 and PS2

(GAV) is:

Select PS1.Film.title, PS1.Critiques.critique,

PS2.Film.title

From PS1.Film, PS1.Critiques, PS2.Film

Where (PS1.title=’Freedom’ And

PS1.Film.Fid=PS1.Critiques.Fid Or

PS2.title=’Freedom’ And

PS2.Film.Fid=PS1.Critiques.Fid)

The translation of the request over the sources S1,

S2, S3, and S4 (LAV) is:

Select S1.title, S2.Critique.critique

From S1.Film, S2.Critique, S2.Film

Where S1.title=’Freedom’ And

S1.Film.Fid=S2.Critique.Fid Or

S2.title=’Freedom’ And

S1.Film.Fid=S2.Critiques.Fid

Union (the relational request on S3 and S4 is

as follows)

Select S3.Film.title, S4.Film.critique

From S3.Film, S4.Film

Where S3.title=’Freedom’ And

S3.Film.Fid=S4.Film.Fid

4 CONCLUSION

This paper presents HAV, a hybrid approach to data

integration which combines GAV and LAV to make

us benefit from their advantages, and defines the

HAV mapping.

The contribution of such an approach is that

HAV is more effective and practicable because the

set of the partial schemas concerned by the HAV

integration approach is small and stable. Thus, in

HAV It will be less complex for the specialized

mediators to carry out the sub-requests (because

each one has a reduced number of sources to

integrate, and these sources are in the same model),

than if the global schema itself were built with LAV

approach, so the complexity to reformulate queries

is reduced.

Noting that it does not exist yet a truly definite

Benchmark which makes it possible to evaluate the

performances of a mediator. We intend to adapt one

among those which exist with our context. In

particular, the result carried out by (Dang Ngoc,

2003) seems to be promising: that it is possible to

use a tree structure of mediators without harming the

evaluation.

REFERENCES

Boulçane, F., 2006. An approach of mediation. In

proceedings of the IEEE ICTTA’06, International

Conference on Information & Communication

Technologie. Volume 2, Page(s): 3546-3551.

Boyd, M., Lazanitis, C., Kittivoravitkula, S., Brien, P. M.,

and Rizopoulos, N., 2004. AutoMed: A BAV Data

Integration System for Heterogeneous Data Sources.

In CAiSE’04, 16th International Conference on

Advanced Information Systems Engineering. Riga,

Latvia, June 7-11, Proceedings Springer-Verlag.

Cali, A., Calvanese, D., Gracomo, G., Lenzerini, M.,

2002. On the expressive power of data integration

systems. Lecture notes in computer science, volume

2503.

Dang Ngoc, T.T., 2003. Fédération de données semi-

structurées avec XML. Thèse de Doctorat.

Université de Versailles-Saint-Quentin.

Friedman, M., Levy, A., Millstein , T., 1999.

Navigational plans for data integration. In Proceedings

of the sixteenth national conference on Artificial

intelligence and the eleventh Innovative applications

of artificial intelligence conference innovative

applications. Pages: 67 – 73, Orlando, Florida, United

States.

Koffina, I., Serfiotis, G., Christophides, V., 2005.

Foundations for Information Integration: A State of

the Art. In FORTH-ISL, Sixth Framework

Programme.

Lenzerini, M., 2002. Data Integration: A Theoretical

perspective. In Proceeding of the twenty-first ACM

SIGMOD-SIGACT-SIGART, Symposium on Principles

of Database Systems.

Ullman, J.D., 1997. Information Integration using logical

views. In Proc.of the 6th Int. Conf. on Database

Theory (ICDT’97). Volume 1186 of Lecture Notes in

Computer Science, pages 19-40, Springer.

Xu, L., Embley, D.W., 2004. Combining the Best of

Global-as-View and Local-as-View for Data

Integration. In Information Systems Technologies and

Its Applications. ISTA. 123–136.

THE HAV DATA INTEGRATION APPROACH: The Mapping in HAV

493