DESIGNING XML PIVOT MODELS FOR MASTER DATA

INTEGRATION VIA UML PROFILE

Ludovic Menet and Myriam Lamolle

Laboratoire d’INformatique et Communication (LINC)

IUT de Montreuil, Université Paris 8, 140 rue de la Nouvelle France, F-93100 Montreuil, France

Keywords: Interoperability, Master Data Management, Metamodel, XML Metaschema, UML profile, XML Schema.

Abstract: The federation of data sources and the definition of pivot models are strongly interrelated topics. This paper

explores the difficulties of a mediation solution based on XML architecture and the concept of Master Data

Management. In this solution, pivot models use the standard XML Schema allowing the definition of

complex data structures. To date, the graphical modeling of XML Schema models is not standardized. The

introduction of a models definition formalism is a mean to make modeling more accessible. UML is a

modeling object language which is more and more used and recognized as a standard in the software

engineering field, which makes it an ideal candidate for the modeling of XML Schema models. In this

purpose we introduce features of the UML formalism, through a profile, to facilitate the collaborative

definition and the exchange of these models, and to introduce the capacity to express semantic constraints in

XML models. These constraints will be used to perform data factorisation and to optimise data operations.

1 INTRODUCTION

The evolution of computer networks and data

management systems has led to the advent of large-

scale information systems inside companies. These

systems increasingly using the Web for sharing

information are characterised by geographically

distributed and heterogeneous data sources.

Consequently, the information management

becomes complex, inefficient, insecure and

expensive. Most research into heterogeneous data

integration has opted for solutions based on

mediation architectures such as (Abiteboul, 2002),

(Garcia-Molina, 1995), (Lamolle, 2005). More

recently, the Master Data Management (Orchestra

Networks, 2000) has been defined as an activity

focused on the unification, the management and the

integration of reference data across an Information

System. These mediation architectures are based on

the definition of a data pivot model federating

several heterogeneous data sources. The definition

of this model is carried out by a semi-automatic

process requiring human intervention, or is done by

an expert in the field. However, all existing solutions

use various tools, technologies and formalisms to

define models. In this paper, we describe a means to

facilitate the definition of models for data integration

using standards components and formalisms. We

focus on solutions based on the standard XML

recommended by the W3C (W3C, 2004). In this

architecture, a pivot model is an XML Schema

document allowing complex, structured, typed and

rich data structure to be defined. The definition of a

model is based on the XML Schema technology in

keeping with the standard defined by the W3C. The

use of XML is appropriate to the definition of

models but involves an extensive knowledge of this

language, so various people from several teams can

participate in the process. In addition, XML Schema

does not offer a way to materialise explicitly

semantic constraints. We propose using UML

(Pilone, 2006) as a common formalism in order to

make the definition, understanding and exchange of

a model easier, and to introduce semantic constraints

into XML models.

The rest of the papers is organised as follows: the

section 2 presents our contributions presenting how

to design XML models with UML; the section 3

presents a case study focused on the application of

our works; the section 4 concludes this paper.

2 MASTER DATA MODELING

PROFILE (MDM-P)

In an Information System, the needs evolve

inevitably over the time. The complexity of a data

461

Menet L. and Lamolle M. (2008).

DESIGNING XML PIVOT MODELS FOR MASTER DATA INTEGRATION VIA UML PROFILE.

In Proceedings of the Tenth International Conference on Enterprise Information Systems - DISI, pages 461-464

DOI: 10.5220/0001695604610464

 SciTePress

model evolves in the same way. These models are

becoming more difficult to define and design due to

the diversity of existing technologies. In order to

resolve this issue, a standard notation and a standard

design method has to be used. We should keep in

mind that the quality of a model is based on three

criteria: (i) Expressiveness to exploit the information

contained by data; (ii) Expressive power in order to

meet the needs of the field; (iii) Suitable level of

abstraction, in order to be accessible by specialists

from different fields.

The first and the last points are the issues that we

want to address using the UML formalism.

In addition to its modeling abilities, UML allows

profiles to be defined. A profile specialises the UML

formalism for an application field or a particular

technology. Many profiles have been developed for

several goals: for expressing the whole semantic of

CORBA (OMG, 2002), for defining the features of

EJB (Greenfield, 2001), or the Turtle-P profile for

the formal validation of critical and distributed

systems (Apvrille, 2006). An UML profile can be

used for giving syntax to new concepts, giving a

different notation to existing symbols, or adding

semantic to existing concepts. In addition, we use

the UML formalism as a way to introduce semantic

constraints in XML models in order to optimise

some operations such as data factorisation, data

deletion, etc... We define MDM-p in three steps.

First, we proceed to the definition of the domain of

the profile, i.e. the metamodel defining the concepts

and the relations between them. Secondly, we carry

out the technical definition of the profile which

establishes the matching between UML class

diagram concepts and those defined in the profile.

Thirdly, we define the profile extending the standard

UML class diagram.

2.1 Domain Definition of MDM-p

The domain of our profile is defined by a

metamodel. The MOF (Meta Object Facility) has

been defined by the OMG (OMG, 2003) as a

standard for defining metamodels. The MOF

metamodel framework is depicted as a four layer

architecture. To be conformed to the MOF we have

defined our metamodel as the following:

Our works is focus on the layer M2, representing

the metamodel of Master Data models.

2.2 Definition of MDM-p

Our profile extends the UML class diagram for the

description and the structure of the Master Data

models. Figure 2 presents part of the UML profile

defining the relations between the different concepts

introduced by our metamodel.

Figure 1: OMG Metamodel architecture.

Figure 2: Part of the UML profile representing Master

Data Management metamodel.

Using UML extension mechanism enables us to

extend the UML formalism to our semantic. This

extension is done using stereotypes and tagged

values. Stereotypes are used to extend the semantic

of UML, to a particular domain, by the definition of

new types of element derived from existing ones.

We can see on the figure 2 that the stereotypes

(labelled <<Stereotypes>>) inherit from the element

Class of the UML metamodel. As a result, the

stereotypes will be instantiated from the metamodel

constructor in the same way as the element Class.

Tagged values specify keyword-value pairs of model

elements to set properties for existing elements or

for stereotypes. According to our profile, a model is

composed of stereotyped classes called

<<MClass>>. <<MClass>> is a generic class for the

concepts of our metamodel. The definition of these

MOF

Meta MetaModel

UML

MetaModel

Master Data

Management

MetaModel

UML Profile

Master Data

Management

Model

Data

Conform to

specialises

Defines

LAYER M3

LAYER M2

LAYER M1

LAYER M0

Meta MetaModel

MetaModel

Model

Instance

ICEIS 2008 - International Conference on Enterprise Information Systems

462

stereotypes allows more semantics to be introduced,

out of the concepts defined in UML which will

enable us to define an XML model with an UML

diagram.

3 CASE STUDY

This section illustrates part of a model defined with

MDM-p and its equivalent in XML Schema. The

figure 3 represents the UML modeling of a XML

model defining a simplified train network.

Figure 3: Definition of an XML model with UML.

In this example, we have defined the concept

Train as composed of an engine, wheels (notion of

composition), being able to have wagons (notion of

aggregation) and having properties such as a type

and a trademark. We associate a driver of type

Person with a train. The concept Person defines

properties such as firstname, lastname and birthday.

We have also defined the property email

representing the use of a redefined type. The class

Email uses the stereotype <<SimpleType>>

indicating that it is a redefined type in the XML

Schema sense. The properties of this redefined type

are expressed in an UML annotation specifying

values for SimpleType::base, SimpleType::pattern

and SimpleType::whitespace. The root of the schema

is specified by the stereotype <<Root>>, applied to

the class NetworkTrain. The classes defined in this

model are also stereotyped <<Sequence>>. This

stereotype specifies that the corresponding class

represents a complex element of the type

<xs:sequence> in the XML Schema definition.

A part of the corresponding XML Schema model

is shown below:

<xs:element name="train" type="trainType"

minOccurs="0" maxOccurs="unbounded"/>

</xs:sequence>

</xs:complexType>

</xs:element>

<xs:complexType name="trainType">

<xs:sequence>

<xs:element name="trademark"

type="xs:string"/>

<xs:element name="type"

type="xs:string"/>

<xs:element name="wagon" type="wagonType"

maxOccurs="unbounded">

<xs:annotation>

<xs:appinfo>

<osd:aggregation/>

</xs:appinfo>

</xs:annotation>

</xs:element>

<xs:element name="engine"

type="engineType">

<xs:annotation>

<xs:appinfo>

<osd:composition/>

</xs:appinfo>

</xs:annotation>

</xs:element>

<xs:element name="wheel" type="wheelType"

minOccurs="4" maxOccurs="unbounded">

<xs:annotation>

<xs:appinfo>

<osd:composition/>

</xs:appinfo>

</xs:annotation>

</xs:element>

<xs:element name="drive"

type="personType"/>

…

Using the UML formalism we are able to

introduce semantic constraints to XML models,

bringing more signification to relationships inside

models. By semantic constraints, we talk about

object features such as aggregation, composition,

generalisation and specialisation. It is not possible

explicitly with XML Schema to express such

constraints. Indeed, an aggregation can be

materialised with the XML Schema keyref

mechanism but the lyfe cycle of such relation is not

evidently expressed. We have then introduced these

features throught the mechanism of extension

suggested by the W3C. With our extensions (noted

<osd:composition>, <osd:aggregation> in the

previous code example) we can describe the

DESIGNING XML PIVOT MODELS FOR MASTER DATA INTEGRATION VIA UML PROFILE

463

semantic links between concepts. More precisely,

the concept Train is a composition of wheels,

engine, and an aggregation of wagons. These

semantic links have strong impacts on optimising

data. Indeed in our previous example the

composition implies that there cannot be instances of

the concept Engine without instances of the concept

Train. Deleting an instance of concept Train implies

that all dependent instances (Engine) will be

removed. So, the deletion process is optimised. In

the other hand, the aggregated instances (Wagon)

will not be deleted in the case of aggregation

between concepts.

Generalisation and specialisation relations are

used to factorise data. In the generalisation case,

common attributes are gathered in a general concept.

For example, attributes such as firstname and

lastname are common to the concepts driver and

mecanician. These two attributes are migrated to the

concept Person to factorise data, avoiding

duplication of their definition in the concepts driver

and mecanician.

4 CONCLUSIONS

We saw in this paper that we can facilitate

communication between people using a common

formalism, in our case using UML as an abstraction

of any technical language. UML is a powerful and

flexible modeling language and XML became a

standard for data interchange on the Web. The use of

these two technologies addresses interoperability

and a standard means of exchanging and defining

models. Therefore, we have shown how UML

profiles facilitate the definition of an XML Schema

model and fill the lack of semantic expressiveness.

Our follow-up work will consist of the

enrichment of MDM-p and the creation of a

complete data modeling language for the Master

Data Management module in the following two

ways: (i) a graphical language using UML modeling,

(ii) a specialised language using UML meta-

modeling. This language will take into account the

definition of constraints and the validation of models

using OCL (Object Constraint Language) (OMG,

2002) (Bazex, 2003).

REFERENCES

Abiteboul S., Cluet S., Ferran G., Rousset M.-C., 2002.

The Xyleme Project. Computer Networks 39.

Apvrille L., De Saqui-Sannes P., Khendek F., 2006.

TURTLE-P: A UML Profile for the Formal Validation

of critical and Distributed Systems. SoSym (Software

and System Modeling) Journal, Springer, ISSN 1619-

1366, pp. 1-18, July.

Bazex P., Bodeveix J-P., Millan T ., Le Camus C.,

Percebois C., 2003. Vérification de modèles UML

fondée sur OCL. INFORSID. Nancy, France. pp. 185-

200.

Garcia-Molina H., Papakonstantinou Y., Quass D.,

Rajaraman A., Sagiv Y., Ullman J., Widom J., 1995.

The STIMMIS approach to mediation: Data Models

and Languages. NGITS (Next Generation Information

Technologies and Systems), Naharia, Isreal, June 27-

29.

Greenfield J., 2001. UML Profile For EJB. Rational

Software Corp, May.

Lamolle M., and Zerdazi A., 2005. Intégration de Bases de

données hétérogènes par une modélisation

conceptuelle XML, In Conférence sur l’Optimisation

et les Systèmes d’Information (COSI’05), pp.216-227.

Ober I., Graf S., 2005. Timed annotations in UML

accepted to STTT, Int. Journal on Software Tools for

Technology Transfer Springer Verl.

OMG/MOF, 2000. Meta Object Facility (MOF)

Specification, OMG Document formal/2000-04-03,

http://www.omg.org/technology/documents/formal/mo

f.htm.

OMG, 2002. “CORBA specifications”.

http://www.omg.org/technology/documents/formal/pr

ofile_corba.htm.

OMG, 2002. Response to the UML 2.0 OCL.

http://www.omg.org/docs/ad/02-05-09.pdf

Orchestra Networks, 2000.

http://www.orchestranetworks.com

Pilone D., Pitman N., 2006. UML 2.0 in a Nutshell,

O’Reilly.

ICEIS 2008 - International Conference on Enterprise Information Systems

464