XML-IS: ONTOLOGY-BASED INTEGRATION ARCHITECTURE

Christophe Cruz and Christophe Nicolle

Laboratoire Le2i (UMR CNRS 5158),Université de Bourgogne, Faculté des Sciences Mirande

Aile de l’Ingénieur, BP 47870, 21078 Dijon Cedex, France

Keywords: XML, XML schema, Knowledge, Ontology, Information retrieval, Integration, Semantic.

Abstract: This paper presents an architecture that aims at integrating XML documents. Today, information exchanges

in business processes widely use XML to format data. But this is done without a common process

management. For some applications it is necessary to archive data for future processes (error detections,

logs, and knowledge enquiry data mining). For this reason our architecture allows users to retrieve

information from automatically integrated XML. This is not possible without the help of a semantic level

definition which is build upon XML schemas. To reach this goal we use the knowledge defined in XML

schemas to share a global ontology of domain. At a final step this ontology is used in the retrieval process

by end-users to search and retrieve information.

1 INTRODUCTION

One of the most impressive capacities of new

information and communication technologies is to

generate data. Every day thousands of new Web

sites and databases are available. However, it is done

without taking into account the future life of those

data. As a consequence, the reusability of data is the

weakness of these new information sites. This is the

reason why search engines are very popular.

Moreover, the search on the Web consists in using

the appropriate words that are associated with the

context. But you cannot surf on a knowledge level

that drives you directly to information looked for. In

fact, this knowledge should permit a better search

and a better reusability.

During the last seven years, XML has known an

incredible and extensive use for systems of data

exchange and data sharing. XML is used to define

most of the Web information (videos, images, 3D

scenes, domain information, office data, etc.) Many

systems using XML as database integration have a

mediation approach (Pan, 2002), (Draper, 2001),

(Carey, 2000), (Cali, 2001). The evolution of the

Web technologies changed the integration problem

of information. In fact, the XML contribution to

define not only integration schemas but also the

definition languages of the corresponding models

reduced considerably the problems related to the

structural and the syntactic heterogeneity. Moreover,

the contribution of the Web technologies related to

the service-oriented architectures solved partially the

problems of the localization and the data access,

allowing the design of interoperability architectures

on a greater scale (Aberer, 2002). Nevertheless,

during the integration data process and the

integration services there remain many problems

related to semantic heterogeneity. So, a formal

description of the semantic shared should avoid

ambiguities when it is defined in relation to a

domain of knowledge. Hence, the reuse of

information in a specified domain should be

improved.

2 RELATED WORK

The implementation of an ontology is a mapping

stage between the system elements and their

ontological "counterparts". Once this mapping has

been carried out, the representation of elements in

the ontology is regarded as a meta-data diagram. The

role of a meta-data diagram is double (Amann,

2003). On the one hand, it represents an ontology of

the knowledge shared on a domain. On the other

hand, it plays the role of a database schema which is

used for the formulation of requests structured on

meta-data or to constitute views. This principle is

applied to the XML-IS architecture to provide

integration structures and request processes to these

273

Cruz C. and Nicolle C. (2008).

XML-IS: ONTOLOGY-BASED INTEGRATION ARCHITECTURE.

In Proceedings of the Fourth International Conference on Web Information Systems and Technologies, pages 273-277

DOI: 10.5220/0001525102730277

 SciTePress

structures. According to (Cruz, 2004), (Klein, 2002),

(Lakshmannan, 2003) data integration consists in

defining rules of mapping between information

sources and the ontological level. These rules consist

in adding semantic annotation to source elements

and thus provide semantic definition to elements

compared to a consensual definition of the meaning.

Compared with the approach of SAWSML (Martin,

2007) XML-IS is also a method to annotate

<xs:element> tags contained in XML schema, but

it is used to identify the tags in an XML document

validated by this XML schema. In SAWSDL the

annotation process consists in identifying the input

and output formats of the Web Services which are

specified by an XML schema. Here XML

documents validated by annotated XML schemas are

indexed by an automatic process. Our architecture

should be able to annotate WSDL and SOAP index

documents as it is done with XML schemas and

XML documents. Actually, it was not designed for

it, as SAWSDL was not designed to index SOAP

documents.

The next section gives a general view of our

architecture based on the field of ontologies and

formal languages. This section introduces the notion

of the semantic mark which is the basis of our

architecture. Section 4 describes the implementation

of XML-IS by explaining the data model, the

handling system and the retrieval system.

Figure 1: this figure presents the architecture of XML-IS

composed of four layers. The ontology layer and the

instance layer which define the semantic level. The factor

layer and the tag layer define the syntactic level. Semantic

marks are used to link the Semantic level to the Syntactic

level via the ontology layer and the factor layer. The tag

layer is used to index XML documents. The ontology

layer is used to request XML-IS.

3 ARCHITECTURE

Our proposal consists in connecting various levels

which are composed of the semantic and the

schematic levels. To specify the semantic of the

XML schema elements it is necessary to identify and

mark them. These marks will be used to establish

links between both levels. The syntactic level

concerns the structure of the XML document and the

semantic level concerns its semantic definition (cf.

Fig. 1).

The properties of a language of Dyck were the study

subject undertaken by J. Berstel (Berstel, 2000). By

drawing parallels between XML grammar and

languages of Dyck, J. Berstel defines the concept of

factor. According to the lemma 3.3 of J. Berstel, a

language is factorizable into an under language and

a factor of a language of Dyck is a language of

Dyck. Thus, a subtree of an XML document can be

validated by a factor of a language of Dyck. This

implies that if a factor were defined on an XML

language then this factor would correspond to a

production rule of the reduced grammar XML

generating this language. This proposal makes it

possible to introduce the concept of “semantic

mark”. A semantic mark is a mark on an XML

schema that makes it possible to identify a

production rule (e.g. a factor). This production rule

corresponds to tag

<xs:element/> in a XML

schema. A tag

<xs:element/>, which is marked, is

linked to the definition of a concept, or a relation, or

an attribute. The semantic annotation by a semantic

mark on a production rule permits to annotate

automatically all tags in XML documents which are

validated by the XML schema marked. Those

concepts, relations and attributes are used to define

the domain ontology of the XML schema. The

feature needed for our ontology is available in the

OWL specification. Consequently, we use Jena to

store the OWL ontology defined on XML schema

and the OWL instances on the corresponding XML

documents. OWL specifications make it possible to

define concepts (

owl:class and rdf:subClassOf),

relations (

owl:ObjectProperty), and attributes

(

owl:DataTypeProperty) for several domain

ontologies. For instance, the concept Heater is

common to both domain ontologies coming from

two different XML schemas. The pooling of this

concept from several domain ontologies makes it

possible to integrate several XML schemas. The

instance layer composed of instances of the ontology

layer is automatically linked to the tag layer which is

composed of the tags from XML documents

validated by annotated XML schemas.

Factor laye

Tag laye

Syntactic

Level

XML-IS architecture

Ontology laye

Instance laye

Semantic

Level

WEBIST 2008 - International Conference on Web Information Systems and Technologies

274

The next section describes the implementation of the

different layers and underlines the fact that several

semantic marks from various XML schemas can

have the same semantic defined by a common

ontology. Hence, the element properties

corresponding to the semantic marks are integrated

within the same concept defined in ontology.

4 IMPLEMENTATION

This section describes the implementation of XML-

IS architecture. The first part shows the data model

of the systems. The second part describes the

handling system. The third part presents the retrieval

model based on the data model.

The semantic level is described by an ontology that

represents the implicit knowledge on XML schemas

which are defined by the users. The syntactic level is

described by an ontology that represents the explicit

knowledge on an XML schema defined by our

architecture.

4.1 Data Model

Concerning the semantic level, the ontology layer is

made of a super class

Concept, a super

ObjectProperty

Relation, the super

DataTypeProperties

Attribute and

SimpleAttribute. An Attribute is a tag that

defines an attribute for a father tag. A

SimpleAttribute is a tag attribute. For instance, id

is an attribute of the tag

div, <div id=”1342”>. The

SimpleAttribute can be used to define the

Attribute tag that makes it possible to integrate

XML documents from different XML schemas.

Every “semantic mark concept” generates

automatically a new class which is a subclass of

Concept. Every new XML document validated by an

integrated XML schema generates automatically

new instances of the new class for every tag

referenced in the ontology. This process is identical

to every semantic mark

relation, attribute and

simpleAttribute. Concerning the syntactic level,

the factor layer represents the factors selected by the

user to define the semantic marks. Those factors are

defined by the following super DataTypeProperty

Factor_Concept, Factor_Relation, and

Factor_Attribute. We also defined a property

SchemaXML to keep a link between Factors and

Schemas. The tag layer is composed of the concept,

the relation and the attribute instances which keep

links between XML schemas, XML documents and

the ontology.

4.2 Handling System

The schema integration process is composed of four

steps, the parsing step, the schematics marks

selection step, the semantic validation step and the

covering validation step.

The Parsing Step. XML schemas are also XML

documents. Thus, they are also trees from which we

generate a special tree called Schema Tree (Fig. 2

(1)). The nodes of the Schema Tree are generated

from the tags

<xsd:element>. The attributes of the

nodes are generated from the tags

<xsd:attribute>.

The Semantic Marks Selection Step. During the

marking of the first XML schema, the user defines

the first ontology of domain. For this, the user

defines marks if the elements are concepts, relations

or attributes. During the marking of the other XML

schemas, the user defines other ontologies of

domain, but if a tag

Concept or a tag Relation or a

tag

Attribute is already defined then the user

selects the corresponding entity (Fig. 2 (3)).

The Semantic Validation Step. When the user has

finished the semantic marks selection step then it is

necessary to validate this marking. Indeed, the user

is free to define the semantic marks but some rules

must be observed for the semantic coherency. For

instance, a

Relation tag must have an Concept tag

as father tag. An entity

Relation must have a set not

equal to zero of a tag

Concept. An tag Attribute

can be linked to any entity.

The Covering Validation Step. Once the semantic

validation is done, the covering validation must be

undertaken. Indeed, every element of the element

tree is not necessarily selected by the user. For the

integration it is necessary to link every entity.

Consequently, the root node must at least be marked.

Thus, the tree is represented by clusters. The nodes

without marks are associated to the cluster of the

ancestor which handles the semantic mark (Fig. 4).

XML-IS: ONTOLOGY-BASED INTEGRATION ARCHITECTURE

275

Figure 2: This figure represents the graphic interface for the semantic definition of entities. In (1) the user can see and select

an entity. And in (2) he defines the kind of selected entity (rien~nothing). With the comboBox in (3) the user can write the

name of the new concept, relation, attribute or select a previous name (nom~name).

Figure 3: This figure shows the clustering of non-marked

nodes. The tree represents clustered entities by red lines.

The nodes without marks are associated to the cluster of

the ancestor (C2 and C3) which handles the semantic

mark.

4.3 Retrieval Model

In this section we present how to retrieve

information from XML-IS. To define our retrieval

system we use the notion of context. This notion is

inspired by the following citation of Recanati

(Recanati, 1993): “The meaning of a word like ‘I’ is

a function that takes us from a context of utterance

to the semantic value of the word in that context,

which the semantic value (the reference of ‘I’) is

what the word contributes to the proposition

expressed by the utterance.” For instance the word

“Beetle” has a different semantic value in the

context “Vehicle” than in the context “Biology”.

Indeed, in the context “Biology” the word “Beetle”

refers to the semantic value “Insect” and in the

context “Vehicule” it refers to the semantic value

“Car”. From this citation we have defined the

following function:

Y = f(X)

f: is the word

X: is the context

Y: is the semantic value of the word f in the

context X

(1)

By analogy, a word is an instance (an instance of

Concept or an instance of Relation or an instance of

Attribute). Y is the value of the attributes taken into

account in the context. The context X is defined by

the following set:

R = { r | r

∈

the set of Relation }

A = { a | a

∈

the set of Attribute }

S = { s | s

∈

the set of AttributeSimple }

X = { r, a, s | r

∈

R’ and R’

⊂

∈

A’and A’

⊂

∈

S’ and S’

⊂

S }

The context can be seen as a semantic filter because

only a subset of attribute values is kept in a

Semantic Element. Actually, the definition of a

context allows us to select a set of instances

Concepts according to a set of Relations,

Attributes and AttributeSimples. In addition, the

(2)(1) (3)

WEBIST 2008 - International Conference on Web Information Systems and Technologies

276

set of

Relations is used to select a set of instances

Concepts. Indeed, an instance of Concept is

selected if it is a son or a father of an instance of the

Relation defined in the context. Thus, if the

instance of

Concept is selected then the semantic

value of the word is a multiset (e.g. bag) of

Attribute values and a second multiset of

AttributeSimple values according to the context.

Moreover, if the instance of

Concept is not linked to

an instance of

Relation, which is not an instance of

one of the

Relations defined in the context, then the

set of semantic values is empty.

These previous definitions allow us now to introduce

the notion of “Contextual Tree”. A Contextual Tree

is the result of a request to our information system

XML-IS. These requests are semantic filters that

give an adapted view of information. A request Rq is

defined in the following manner:

Ω = {x | x

∈

the set of Context }

C = {c | c

∈

the set of instance_Concept }

Rq = { x, y | x

∈

Ω’ and Ω’

∈

Ω and | Ω’| = 1,

∈

C’ and C’

⊂

(2)

Concerning the definition of a context we add a rule

for the validation of the request. If the contextual

tree resulting from the request is not a hierarchical

non-cyclic graph then the context use is not well

defined. It means that the

Relation selected for the

definition of Context generates a Contextual Tree

which is not a tree. To resolve the issue of a bad

defined Context the user has to select fewer

Relations. For instance:

R’ = { “placement” }, A’ = { “shape” }, S’ = {

“id”}

C’ = { c is the set of

Instance_Concept

which is defined by the class

Concept

“wall”, “slab” and “pipe”}

Ω

Geo

= { r, a, s | r

∈

R’ and R’

⊂

∈

A’ and A’

⊂

∈

S’ and S’

⊂

Rq = { x, y | x

∈

Ω

Geo

, y

∈

C’ and C’

⊂

(3)

The result of this request Rq is the selection of all

instances of Concept defined by the instances of

Concept called “wall”, “slab” and “pipe” which are

linked by the instance of

Relation defined by the

entity

Relation called “placement”. The semantic

value is the geometrical value of shape. The result is

an XML document that describes the shapes of walls

that belong to the building.

5 CONCLUSIONS

In this paper, we have presented our method to

integrate XML data and to retrieve XML

information through a defined context of use. The

objectives were reached with the introduction of the

semantic marks. The XML-IS system was tested on

a set of XML grammar schemas as well as on a set

of XML documents associated with each XML

schema.

REFERENCES

Aberer, K., Cudre-Mauroux, P., Hauswirth, M., 2002, A

framework for semantic gossiping. SIGMOD Record,

31(4)

Amann, B., 2003. Du Partage centralisé de ressources

Web centralisées à l’échange de documents

intensionnels, Documents de Synthèse.

Berstel, J., Boasson, L., 2000. XML Grammars, MFCS

2000: 182-191.

Cali, A., De Giacomo, G., Lenzerini, M., 2001, Models for

Information Integration: Turning Local-as-View into

Global-as-View, Proceedings of the International

Workshop on Foundations of Models for Information

Integration.

Carey, M. J., Kiernan, J., Shanmugasundaram, J., Shekita,

E. J., Subramanian, S. N., 2000. XPERANTO :

Middleware for Publishing Object-Relational Data as

XML Documents, The VLDB Journal, pp 646-648.

Cruz, I. F., Xiao, H., Hsu, F., 2004. An Ontology-based

Framework for Semantic Interoperability between

XML Sources, In Eighth International Database

Engineering & Applications Symposium (IDEAS

2004).

Guarino, N., 1994, The ontological level, in R. Casati B. S.

& White G., eds, Philosophy and the cognitive

sciences, Hölder-Pichler-Tempsky.

Klein, M., 2002. Interpreting XML via an RDF schema. In

ECAI workshop on Semantic Authoring, Annotation

& Knowledge Markup (SAAKM 2002), Lyon, France.

Lakshmannan, L. V., Sadri, F., 2003. Interoperability on

XML Data, In Proceeding of the 2nd International

Semantic Web Conference (ICSW’03).

Martin, D., Paolucci, M., Wagner, M., 2007, Towards

Semantic Annotations of Web Services: OWL-S from

the SAWSDL Perspective, In OWL-S Experiences and

Future Developments Workshop at ESWC 2007, June

2007, Innsbruck, Austria.

Pan, A., Raposo, J., Álvarez, M., Montoto, P., Orjales, V.,

Hidalgo, J., Ardao, L., Molano, A., Viña, Á., 2002,

The Denodo Data Integration Platform, VLDB, Hong

Kong, China.

Recanati. F., 1993, Direct Reference: From Language to

Thought. Blackwell Publishers, Oxford, UK.

XML-IS: ONTOLOGY-BASED INTEGRATION ARCHITECTURE

277