HOW TO EVOLVE ONTOLOGY AND MAINTAIN ITS

COHERENCE

A Corrective Operations-based Approach

Najla Sassi, Wassim Jaziri and Faiez Gargouri

Higher Institute of Informatics and Multimedia, Sfax University, Tunisia

Keywords: Ontology Evolution, Corrective Operations, Evolution Kits.

Abstract: An ontology is a specification of a conceptualization related to a domain of knowledge. In an evolution

context, it cannot be considered as a finite conceptualization since it must be adapted to new requirements.

This adaptation must respect the coherence of the ontology and its conformity regarding some objectives.

To update ontology while maintaining its coherence, this paper proposes an anticipatory approach based on

corrective operations. For each change to occur on the ontology, we define corrective operations to prevent

and correct potential inconsistencies likely to be generated.

1 INTRODUCTION

Changing environments require adaptable ontologies

to changes that occur over time. The adaptation of

ontology is a complex process and several evolution

problems must be treated, in particular maintaining

of the ontology coherence. The ontology coherence

is an agreement between its ontological entities with

respect of semantics of the subjacent language of

representation.

The application of a change in ontological entities is

a modification of a subset of knowledge represented

by the ontology. Changes management requires

defining mechanisms specifying how knowledge can

be changed and how to maintain the consistency of

knowledge after each change. In addition,

ontological entities are linked to each other

semantically (their semantics are complementary)

and conceptually, the application of a change in

some ontological entities may have effects on other

entities. Thus, the ontology evolution requires a

structured process to take into account all direct and

indirect effects of any change.

Two types of inconsistency may be identified

(Maedche et al., 2003):

 Structural inconsistency occurs when the

constraints of the ontology model are invalid or

if the semantics of the subjacent language of

ontology is not respected.

 Semantic inconsistency occurs when the

significance of the entities of ontology is

changed.

Maintaining the ontology coherence remains little

studied in the literature. In (Maedche et al., 2003),

ontology is considered consistent if its axioms are

respected and if it satisfies the whole of the

invariants defined in the model of ontology. The

authors defined constraints of consistency related to

the model of ontology according to the semantics of

the KAON language.

In (Schlobach et al., 2003), an algorithm of

resolution of inconsistencies is proposed based on

the identification and the elimination of incoherent

concepts. This algorithm identifies the concepts

sources of "logic contradictions" and provides

intelligent algorithms to follow and solve the sources

of inconsistencies.

Stojanovic et al. (Stojanovic et al., 2003)

proposed an approach for the management of

evolution and the maintaining of consistency for

KAON ontologies. The authors proposed the

concept of strategies of evolution which allow to the

ontologist to choose the most suitable solutions for

the resolution of inconsistencies.

Haase et al. (Haase et al., 2005) also used the

concept of strategies of resolution based on the

constraints of OWL-Lite for the detection and the

resolution of inconsistencies in OWL ontologies.

However, the resolution of inconsistencies is done

after application of changes. The resolution of

384

Sassi N., Jaziri W. and Gargouri F. (2009).

HOW TO EVOLVE ONTOLOGY AND MAINTAIN ITS COHERENCE - A Corrective Operations-based Approach.

In Proceedings of the International Conference on Knowledge Engineering and Ontology Development, pages 384-387

 SciTePress

inconsistencies is ensured in two phases: the

detection of inconsistencies which consists in

finding the parts of ontology which do not satisfy the

consistency conditions and the generation of

changes that allow ensuring the consistency of

ontology by generating additional changes. Three

types of consistency are defined: structural

consistency, logic consistency and user consistency.

Structural consistency ensures that ontology satisfies

the constraints of ontology language. Logic

consistency refers to the formal semantics of

ontology and at its satisfiability, i.e., it is

semantically correct and does not present logic

contradictions. The user consistency takes into

account the particular requirements of users.

Flouris et al. (Flouris et al., 2005) differentiate

between a consistent ontology and a coherent

ontology. Ontology is inconsistent if there is no

interpretation which satisfies all the axioms of this

ontology. It is incoherent if it does not satisfy some

predefined constraints or the related invariants. The

predefined constraints describe the consistent model

of ontology. These authors consider the

inconsistencies as sign of bad design and their

correction does not relate to the ontology evolution

but it is rather related to the ontology design.

Luong et al. (Luong et al., 2007) distinguish two

levels of consistency for the model of ontology:

structural consistency and logic consistency.

Structural consistency relates to the constraints of

consistency defined for an ontology model by

ensuring a good organization of the ontological

entities at the level of structure. Logic consistency

checks if the elements of ontology remained

"semantically correct" after their evolution. The

inconsistencies generated in ontology can be solved

automatically using strategies of resolution. These

strategies contain solutions which guide the process

of resolution for all the types of changes.

However, the majority of existing works are

interested in specific categories of ontology such as

OWL ontologies or KAON ontologies. Moreover,

the proposed approaches are based on the correction

of inconsistencies after they occur. In this paper, we

propose an anticipatory approach to manage

inconsistencies before they occur. We express the

requirements of evolution using types of changes.

For each type of change, we define corrective

operations that must be applied in conjunction with

this type of change in order to correct consistencies.

This paper is structured as follows. Section 2

proposes an anticipatory approach for ontology

consistency management. We present in sections 3

and 4 the principles of corrective operations and

evolution kits. Sections 5 concludes this work.

2 ONTOLOGY EVOLUTION

APPROACH

We propose in this work an ontology evolution

approach based on three steps to allow monitoring

the evolution of ontology by creating a new version

better adapted to the required changes (figure 1):

1. Expressing evolution changes: in changing

environment, users express new requirements

to take into account in the ontology. These

requirements are expressed informally and

sometimes in a fuzzy and ambiguous manner.

In this step, we express clearly the users’

requirements according to types of changes to

apply on the ontology.

2. Maintaining the ontology coherence: each type

of change may generate inconsistencies in all

parts of the ontology. In this step, we verify the

effects of changes on the ontology and we

define corrective operations to resolve them. In

addition, knowledge represented by the

ontological entities is complementary and

dependant, it is also necessary to identify the

direct and indirect effects (the derived effects)

of each type of change.

3. Creating a new version: after updating ontology

by applying types of changes, a new ontology

version is created. Thus, in an evolution

context, different versions of ontology should

coexist. To control these versions, it is

important to monitor the relationship between

them. However, establishing links between

versions is a complex task and requires an

investment. These links must respect the order

of versions and the changes have been occurred

[Kle02] [NK04]. We also decide on the

relevance to preserve the old version of

ontology in the ontological database or to

remove it. This choice is conditioned by the

types of implemented changes (subtractive or

not subtractive changes). In the case of a

subtractive evolution, the old version of the

ontology will be stored and added to the

ontological database. It is also important to

provide access to all versions of ontology.

HOW TO EVOLVE ONTOLOGY AND MAINTAIN ITS COHERENCE - A Corrective Operations-based Approach

385

Figure 1: The steps of an ontology evolution process.

3 CORRECTIVE OPERATIONS

The evolution of ontology can be expressed by the

update of one or more ontological entities (concept,

relationship, property, axiom). To allow updating an

ontological entity, we define primitive and

composite operators called types of changes able to

evolve ontology. These types of changes extend

these proposed in the literature (Klein et al., 2002)

(Stojanovic, 2004) to express all evolution

possibilities on the ontological entities (Sassi et al.,

2008). Each requirement of evolution can be

expressed by a primitive or a composite type of

changes.

In an evolution process, the application of types

of changes should have as consequence an ontology

which is in conformity with the whole of coherence

rules. The preservation of the ontology coherence

requires the preservation of the integrity of the

model and the constraints of ontology by preventing

the effect of each type of change on the ontology.

However, types of changes ensure only the

modification of ontology. They not guarantee that

ontology remains coherent after modifications. The

definition of types of changes must be associated

with adequate mechanisms to ensure the coherence

of ontology and its conformity after evolution. This

task is essential in an ontology evolution process

since it conditions the validation and the adoption of

the new generated version of ontology.In this work,

we develop anticipatory solutions managing the

inconsistencies upstream of their appearance to

avoid them. We identify the inconsistencies due to

each type of change in order to propose corrective

operations changes allowing correcting them. These

corrective operations are automatically applied by

the system in combination with the type of change.

To define corrective operations, we analyse the

direct and indirect effects of each type of change, we

detect inconsistencies likely to be generated on the

ontological entities and define additional changes for

each type of inconsistencies to resolve them (figure

2). Corrective operations depend on the type of

change.

Figure 2: The definition of corrective operations.

4 EVOLUTION KITS

We define an evolution kit as the combination of a

type of change and the corrective operations. The

evolution kits allow updating ontology while

preserving its coherence. We define for each

evolution kits: the type of change, the pre-

conditions, post-conditions, potential inconsistencies

and additional changes.

 Pre-conditions: must be checked and controlled

by the system before applying a type of change.

 Inconsistencies: potential problems can be

generated due to a type of change.

 Additional changes: to be attached to each type

of change to correct the inconsistencies that may

be generated.

 Applicative post-conditions: define what must be

true after applying the type of change,

independently of the ontology coherence.

 Coherence post-conditions: define what must be

true if the ontology is coherent.

Each type of change represents with additional

changes, a "coherent evolution kit". We define as

many evolution kits as types of changes (Sassi et al.,

2008). For an evolution requirement, the

Analyse the

effects of changes

Define

corrective operations

Detect

inconsistencies

KEOD 2009 - International Conference on Knowledge Engineering and Ontology Development

386

corresponding coherent change kit is applied

rather than only the type of change.

5 CONCLUSIONS

This paper treats the problem of ontology evolution

and the coherence maintaining. It presents corrective

operations to allow updating ontology while

maintaining its coherence and its conformity. Types

of change allow updating ontology but do not ensure

its coherence. The application of a type of change

may produce inconsistencies in ontological entities.

To correct them, corrective operations are

automatically done in addition to the type of

changes.

To implement evolution kits, we developed the

OntoChanges tool based on Protege. OntoChanges is

an ontology evolution support which allows users

updating ontologies while preserving theirs

coherences.

REFERENCES

Flouris, G., Plexousakis, D., 2005, Handling Ontology

Change: Survey and Proposal for a Future Research

Direction. Technical report FORTH-ICS/TR-362.

Haase, P., Stojanovic, L., 2005, Consistent Evolution of

OWL Ontologies. In Proceedings of the 2

European

Semantic Web Conference (ESWC-05).

Klein M., Fensel D., Kiryakov A., Ognyanov D., 2002,

Ontology versioning and change detection on the web,

International Conference on Knowledge

Engineering and Knowledge Management

(EKAW02), Spain.

Luong, P-H. Dieng-Kuntz, R. 2007, A Rule-based

Approach for Semantic Annotation Evolution, The

Computational Intelligence Journal, 23(3):320-338,

USA.

Maedche, A., Motik, B., Stojanovic, L., 2003, Managing

Multiple and Distributed Ontologies in the Semantic

Web, VLDB Journal - Special Issue on Semantic Web,

12, 286-302.

Sassi, N., Jaziri, W., Gargouri, F., 2008, Formalisation of

evolution Changes to update domain ontologies, In

Proceedings of International Arab Conference on

Information Technology (ACIT'2008), Hammamet,

Tunisia.

Sassi N., Jaziri W., Gargouri F., 2009, Using of ontology

for modeling the evolution of Tunisian higher

education system, 15

conference on Information and

Software Technologies Kaunas, Lithuania, 2009.

Schlobach, S., Cornet, R., 2003, Non-Standard Reasoning

Services for the Debugging of Description Logic

Terminologies, In Proceedings of IJCAI’03.

Stojanovic L., 2004, Methods and Tools for Ontology

Evolution, PhD thesis, University of Karlsruhe.

Stojanovic N, Hartmann J., Gonzalez J., 2003,

Ontomanager - a system for usage-based ontology

management, Proceedings of FGMLWorkshop.

HOW TO EVOLVE ONTOLOGY AND MAINTAIN ITS COHERENCE - A Corrective Operations-based Approach

387