IMPROVED DEVELOPMENT OF THE DOMAIN ONTOLOGY

FOR DIFFERENT USER PROFILES

Application Domain: Conformity Checking in Construction

Anastasiya Yurchyshyna

, Alain Zarli

CSTB, 290 Route des Lucioles, BP 209, 06904 Sophia Antipolis, France

Catherine Faron-Zucker, Nhan Le Thanh

I3S, UNSA-CNRS, 930 route des Colles, BP 145, 06903, Sophia Antipolis, France

Keywords: Development of the domain ontology, Conformity-checking modelling, User profiles, Semantic search,

Validation by usage.

Abstract: This paper presents a formal method for the development of the domain ontology for different user profiles

in the context of conformity checking in construction, which is developed to enrich our conformity-

checking model. We start by describing our research domain: the conformity-checking in construction. Then

we discuss some ontology-based approaches for formalising domain knowledge and particularly focus on

methods for the ontology development for different user profiles. In order to efficiently adapt our

conformity-checking ontology for different user profiles, we suggest a semantic approach for the improved

development of the domain ontology that takes into account the domain knowledge. This semantic approach

is based on three main ideas. First, we adapt the knowledge acquisition method developed for our

conformity-checking model. Second, we integrate a method of the context modelling of the domain

ontology applied by end users by integrating the results of the semantic search. Third, we develop an

approach for the adaptation of the domain ontology for different user profiles. Finally, we describe a web-

based prototype, the C3R (Conformity Checking in Construction: Reasoning) prototype that integrates our

semantic method of the improved development of the domain ontology for different user profiles.

1 INTRODUCTION

This paper presents a formal method for the

development of the domain ontology for different

user profiles. This work continues and extends our

research on the modelling of the conformity

checking process in construction, and particularly

focuses on the conceptual modelling of domain

knowledge and the usage-based validation by end

users.

The complexity of the conformity checking

problem can be explained by the following factors:

(i) the multidisciplinary of the components defining

the conformity checking (e.g. modeling of

construction regulations, reasoning on conformity),

(ii) the interdependence of various actors of the

construction domain; (iii) the large amount of the

non formalised expert knowledge guiding the

process, (iv) the great volumes of construction data

to be retrieved and maintained.

The central problem of the conformity checking

in construction is to automate the process of

checking whether a construction project (e.g. a

private house, a public building, a non-building

installation) is conform to a set of conformity

requirements described by regulation texts. The

semantic complexity of this problem requires an

expressive formalism for representing the

knowledge of the checking process. Recently,

multiple approaches for the development of

building-oriented ontologies have been developed:

e-COGNOS (El-Diraby et al. 2003), ifcOWL (Gehre

and Katranuschkov, 2007), buildingSMART (Bell

and Bjorkhaug, 2006). Despite the variety of these

approaches, these generic ontologies can be hardly

used for a specific aim of conformity-checking.

175

Yurchyshyna A., Zarli A., Faron-Zucker C. and Thanh N.

IMPROVED DEVELOPMENT OF THE DOMAIN ONTOLOGY FOR DIFFERENT USER PROFILES - Application Domain: Conformity Checking in Construction.

DOI: 10.5220/0001838901750183

In Proceedings of the Fifth International Conference on Web Information Systems and Technologies (WEBIST 2009), page

ISBN: 978-989-8111-81-4

To address these limitations, on our previous

research (Yurchyshyna et al., 2008a), we developed

a conformity-checking ontology, which integrates

not only building-related knowledge, but also the

knowledge on conformity regulation texts and the

expert knowledge on checking procedures.

Developed with the help of domain experts, mostly

from CSTB (Centre Scientifique et Technique du

Bâtiment), our conformity-checking ontology was a

key component of our conformity-checking model.

Complex and multidisciplinary, the construction

industry is a field of collaborative work and

communication of multiple actors, the so-called “key

players of the building-oriented market”, who form

the target audience for the development of the

construction sector and their needs define the

innovation process of the industry. The main key

players of the building-oriented market are as

follows:

• architects generating data related to

different aspects of a building;

• engineers responsible for generating data

related to a specified facility’s system of a

building;

• contructors dealing with process-related

characteristics of a building (scheduling,

cost analysis, project management, etc.);

• consumers of a building product;

• building product manufacturers generating

supplementary data related to a building

product (e.g. physical and functional

characteristics, cost);

• legal authorities formulating performance-

oriented rules of the development of a

building.

Obviously, different actors of the conformity

checking in construction have different needs and

understanding of the checking process. It also means

that they may interpret and use the knowledge from

the domain ontology in a different way (e.g. an

acoustic engineer may need a very detailed

“version” of the domain sub ontology concerning

acoustics, in contrast to a final user interested in

general non-detailed conformity recommendations).

The formalisation and the integration of the

domain tacit knowledge play an important role in the

checking modelling. In our previous work

(Yurchyshyna et al., 2008b) we describe our

approach for formalising expert knowledge, which is

also base on the conformity-checking ontology.

From the other hand, it is always interesting to

enrich the initial ontology by the knowledge that is

acquired from its usage. This research objective

motivated our research on the validation of the

conformity-checking ontology by integrating the

results of the semantic search.

The interdependence of the actors of the

application domain (i.e. conformity checking in

construction) is, however, an important factor to take

into account by the application of a generic domain

ontology. It this context, it is a real challenge to

enrich the approach for the ontology development by

the knowledge of its usage by different user groups

and to refine the ontology for different user profiles.

This research objective motivates our work on the

development of our semantic method of the

improved development of the domain ontology for

different user profiles, the DOUP method.

The paper is organized as follows. In next

section, we discuss some ontology-based approaches

for formalising domain knowledge and particularly

focus on methods for the ontology development for

different user profiles. Section 3 represents our

semantic approach for the improved development of

the domain ontology for different user profiles (the

DOUP-method) and details its three levels. In

section 4, we describe the C3R prototype aiming to

illustrate the feasibility of our approach. Finally, we

describe the ongoing works and the perspectives of

our research.

2 TOWARDS THE

DEVELOPMENT OF THE

DOMAIN ONTOLOGY FOR

DIFFERENT USER PROFILES

Our research on the development of the domain

ontology for different user profiles is based on three

scientific axes. First, we study the main problems of

the development of a domain ontology, which is

characterised by a large amount of the tacit

knowledge. Second, we focus on the factors defining

the different usage of the same generic domain

ontology by different user groups. Third, we study

the methods of the ontological modelling oriented

different user profiles.

As a general rule, the development of an

industry-oriented domain ontology is characterised

by the following factors.

First, a large amount of the knowledge to be

formalised is tacit (Polanyi, 1983). For example, the

conformity-checking in construction is characterised

by: (i) the “know-what” knowledge of the

construction industry commonly known by

architects; and (ii) the “know-how” knowledge of

the checking process shared by conformity experts.

WEBIST 2009 - 5th International Conference on Web Information Systems and Technologies

176

It is thus indispensable to explicit and interpret such

tacit domain knowledge formalising it.

Second, the development of the domain ontology

is driven by its (future) application. For that reason,

the development of a domain ontology is often a part

of some more global research task (e.g. conformity-

checking modelling, semantic search, etc.).

Moreover, the formalisms used for development

should be seamlessly interconnected with the

formalisms of these research tasks and should be

based on interoperable standards. For example, the

development of the buildingSMART ontology (Bell

and Bjorkhaug, 2006) is coordinated with the

elaboration of the SMARTcodes™, the code

provisions for code compliance checking

(Smartcodes, 2008), as well as regulation-centric

knowledge representation formalisms define

conceptual architecture of the conformance

assistance framework (Kerrigan and Law, 2005).

Third, before formal representation, the domain

knowledge should be first interpreted by domain

experts. Even if the domain experts are the

professionals of the domain, it is obvious that such

interpretation remains rather subjective and/or

partial. For this reason, it is important to validate the

acquired domain ontology by usage.

Fourth, from the different point of view, the

expert interpretation sometimes differs from the

understanding of end users, who may find the

knowledge “not adequate” and “difficult to use”, but

fail to express the exact meaning of the concepts

used (e.g. a user may find it difficult to distinguish

between “main door” and “entrance door”, but does

not use these two concepts in the same way).

Fifth; the domain ontology is often defined in a

specified context (Hernandez et al, 2007), which

should be then validated by usage.

The second axe of our analysis is devoted to the

practical usage of the generic domain ontology by

different user groups. Such usage may cause

problems for the following reasons: (i) the

interpretation of the domain knowledge by end users

may be different from the interpretation of domain

experts; (ii) different groups of end users may have

different scope of interest (e.g. an architect and a

legal authority need different level of detailing the

conformity-related construction ontology); (iii) a

large amount of knowledge remains tacit. In other

words, it may be difficult for end users to define

how they really need to use this knowledge (e.g. in

the case of checking the conformity of a public

building, a user may interest only in checking its

accessibility, but not the acoustic requirements,

which are the part of the global conformity-

checking).

The problem of the development of a domain

ontology for different user profiles represents our

third research axe. A general approach for

personalising the user's environment and integrating

the user profiles into the development of information

services is discussed in (Sutterer et al, 2008). The

main methods for the automatic creating and

application of user profiles are discussed in (Gauch

et al, 2007). These methods allow integrating search

results tailored to individual users to more complex

systems and thus to personalise the application od

such systems. In (Sieg et al, 2007), the authors

propose a general approach for representing the user

context by assigning interest scores to existing

concepts in a domain ontology.

We focus on these three research axes aiming the

development of the domain ontology for different

user profiles to define our semantic approach for the

improved development of the domain ontology for

conformity-checking in construction, which allows

the personalisation of the domain knowledge for

different user profiles.

3 SEMANTIC APPROACH FOR

THE IMPROVED

DEVELOPMENT OF THE

DOMAIN ONTOLOGY

Our semantic approach for the improved

development of the domain ontology for different

user profiles (the DOUP-method) has three levels:

1. Our knowledge representation and

acquisition method (the KRA-method)

developed for our conformity-checking

model.

2. Our method of context modelling of the

ontology by integrating the results of

semantic search (the CMV-method).

3. Our approach for the adaptation of the

domain ontology for different user profiles

(the ECMV-method).

3.1 Knowledge Representation and

Acquisition Method

We adopt the ontological approach and the semantic

web technologies (Berners-Lee, 2001) to develop the

knowledge representation and acquisition method

(the KRA-method, cf. Figure 1) that allows us to

represent complex and multidisciplinary knowledge

characterising the conformity-checking process in

construction. In this section, we briefly describe the

IMPROVED DEVELOPMENT OF THE DOMAIN ONTOLOGY FOR DIFFERENT USER PROFILES - Application

Domain: Conformity Checking in Construction

177

main ideas of our knowledge representation and

acquisition method. A more detailed explanation and

corresponding examples could be found in

(Yurchyshyna et al, 2008a).

Figure 1: Knowledge representation and acquisition

method.

The first phase of our method aims to acquire the

formal representations of conformity requirements

expressed by technical construction norms. We have

developed a base of accessibility queries by

extracting them from the CD REEF, the electronic

encyclopaedia of construction texts and regulations,

edited by CSTB, and by formalising them as

SPARQL queries in collaboration with construction

experts from the CSTB.

The second phase aims at the semi-automatic

development of an ontology oriented conformity

checking on the basis of the concepts from the

acquired SPARQL queries. These concepts are

organized as hierarchies and described in the OWL-

Lite language. The acquired ontology is then

enriched by non-IFC concepts from formalized

conformity queries. The intervention of domain

experts is required in this case to define new non-

IFC concepts in terms of the checking ontology (e.g.

GroundFloor class is defined by a resource of type

IfcBuildingStorey situated on the level of entering

into a building).

The third phase is dedicated to the acquisition of

a construction project representation oriented

conformity checking. This representation is based on

its initial ifcXML representation and is guided by

the acquired conformity-checking ontology. We

develop an XSLT stylesheet that filters this ifcXML

to extract the data relative to the conformity

checking ontology and organizes them as RDF

triples. The acquired RDF is then enriched with non-

IFC concepts extracted from conformity queries

(e.g. a project representation is enriched by

GroundFloor concept calculated on the basis of its

initial IFC-based data (e.g. IfcDoor, IfcStair, etc.)

The acquired queries, however, contain only

conformity constraints, but have no supplementary

information, guiding the checking process: e.g. the

scheduling of queries. The forth phase of our method

aims thus at the development of semantic annotation

of conformity queries. We propose a special RDF

annotation of a query, developed according to its

tag-based context: possible values for certain tags

are concepts/properties of the conformity-checking

ontology.

To do it, we combine two main methods of

document annotation: annotation by content of the

document and annotation by its external sources

(Mokhtari and Dieng-Kuntz, 2008). First, we

annotate a query by its content. To do this, we define

a set of key concepts of this query, which describe

what is really checked by this conformity

requirement. In other words, we define keyConcept

tag in the RDF annotation of a query, which value is

a list of concepts from the conformity-checking

ontology extracted from the SPARQL representation

of this query. We remark also that there is a

semantic correspondence between different types of

knowledge used for query annotation. For example,

a conformity query defining the physical dimensions

of a door is annotated by a Door concept from our

conformity-checking ontology.

Second, we annotate a conformity query

according to external sources. Such annotation

allows representing different types of knowledge.

First, they are characteristics of the regulation text

from which the query was extracted: (i) thematic

(e.g. accessibility); (ii) regulation type (e.g.

circular); (iii) complex title composed of the title,

publication date, references, etc.; (iv) level of

application (e.g. national), (v) destination of a

building (e.g. private house). Second, they are

characteristics of extraction process: (i) article, (ii)

paragraph from which a query was extracted, (iii)

current number (e.g. 3 query of 1 paragraph of Door

article). Third, it is formalised expert knowledge:

tacit « common knowledge » on the process of

conformity-checking that is commonly applied by

domain experts: (i) knowledge on domain and sub

domain of the application of a query (e.g. Stairs); (ii)

knowledge on checking practice (e.g. if a room is

adapted, it is always accessible). Fourth, it is the

application context of a query. This group specifies

the aspects of query application for certain use cases.

For example, the requirements on the maximal

height of stairs handrail vary from 96 cm (for adults)

to 76 cm (for kids). In this case, it is important to

know the destination of a building (e.g. school).

Characteristics and possible values of the first

two groups are automatically extracted from the CD

WEBIST 2009 - 5th International Conference on Web Information Systems and Technologies

178

REEF. The knowledge described by the last two

groups is defined partially and/or has to be explicitly

formalised by domain experts.

Figure 2: Conformity-checking model.

The knowledge acquired by the KRA-method is

then used in our conformity-checking model that is

based on the analysis of matchings between the

representations of a construction project and of

conformity queries (Yurchyshyna et al, 2008a, cf.

Figure 2).

3.2 Context Modelling of the Domain

Ontology by Integrating the Results

of Semantic Search

According to the KRA-method, the conformity-

checking ontology is developed with the help of

domain experts and does not depend on the

conformity-checking process. All concepts and

relations of the ontology are defined and validated

by domain experts before the checking process and

can not be changed in the process of checking.

Domain experts also formulate rules of definition of

new concepts, context rules and, in general, they

validate the whole knowledge base of the

conformity-checking process.

In some cases, such definitions can be partial or

inadequate, and it does not represent the real usage-

driven conformity-related knowledge of the

checking process: even the definition of domain

experts is not sufficient to represent the whole

complexity of the checking knowledge. It is thus

important to propose an approach of the acquisition

of another type of the checking knowledge: the

knowledge on the checking practices, which turns

out explicit thanks to a large number of checking

operations by different end non-expert users.

To do this, we developed an approach of the

context-based modelling of the domain ontology,

which is validated by usage, the CMV-method

(Yurchyshyna et al, 2008c). In other words, we

proposed an approach for the evaluation of the

semantic proximity of different concepts/relations of

the conformity-checking ontology, according to the

interpretation of end users.

The CVM-method is based on the semantic

annotations of conformity queries. It aims to analyse

the simultaneous choice of the queries, which are

annotated by the same key concepts, and thus to

define the semantic similarity between these

concepts.

In the KRA-method, the semantic annotations of

conformity queries are developed according to the

tag-based context: possible values for certain tags of

semantic annotations of queries are

concepts/properties of the conformity-checking

ontology.

The following example illustrates an annotation

of a query by the door concept of the conformity-

checking ontology.

<rdf:RDF xmlns:ontoCC=”domain.owl#”

<Annotation rdf:ID=””> ...

<ontoCC:Door/>

</domaineApplication> ...

</Annotation>

</rdf:RDF>

Our work on modelling the domain ontology for

conformity checking in construction is conducted

under the Semantic Web vision, which guarantees

more advances capabilities for processing the

knowledge. In particular, we also interest at more

advanced search, the so called semantic search that

gives better results in comparison to traditional

search mechanisms.

The semantic annotation of conformity queries

allows us to propose a user more detailed selection

of queries to be checked. For example, for a user

interested in checking the conformity of a door, we

can propose the semantic search that will give a

semantically richer result: it will interrogate the

domain ontology to define not only the queries

annotated by Door, but also all the corresponding

ones (its subclasses Entrance, EntranceDoor,

FrontDoor, AccessibleEntrance). It also means that

a user can obtain a semantically consistent answer

about the content of the conformity query before

executing it – only by its RDF annotation – and thus

to identify what he really wishes to check.

Technically, such semantic search is based on the

execution of the following SPARQL query against a

base of RDF semantic annotations.

IMPROVED DEVELOPMENT OF THE DOMAIN ONTOLOGY FOR DIFFERENT USER PROFILES - Application

Domain: Conformity Checking in Construction

179

PREFIX a:<annotations.owl#>

PREFIX ontoCC: <domain.owl#>

SELECT ?s ?nQuery ?appValue ?cCl

WHERE { ?s direct::rdfs:subClassOf ?cCl

FILTER(?s ^ontoCC:)

?nQuery a:domaineApplication ?appValue

?appValue rdf:type ?cCl

FILTER (?cCl ~ 'door') }

In our example, the search of “door” expression

will give the list of queries, which application

domain contains “door” and classifies them

according to the conformity-checking ontology. In

comparison to the traditional search (that results

with the only answer “door”), the semantic search

will detail the application domain of found queries

and classifies them into subclasses: (i) “door”; (ii)

“entrance door”, “front door”, “entrance”; (iii)

“accessible entrance” (cf. Figure 3).

Figure 3: Door and its subclasses.

The advantage of such semantic search is that it

is defined according to the general domain

knowledge of the construction industry, formalised

in the conformity-checking ontology, which is

independent of an end user, but helps him to detail

the search of corresponding conformity requirements

and thus to refine the algorithms of their application

during this process.

Another advantage of our approach for the

semantic search of conformity queries is that the

results of the semantic search followed by the user

selection of a query can be then used to validate the

initial domain ontology.

To illustrate these ideas by an example, let us

take three subclasses of a Door class: FrontDoor,

Entrance and EntranceDoor, which are defined

equivalent in the conformity-checking ontology.

They are also used as key concepts to annotate

conformity queries (e.g. these three concepts

annotate the query “an entrance door of any

building should be accessible to disabled persons”).

According to our model, for the checking of the

conformity of an entrance door of a building, a

construction project should be checked to the queries

annotated by all these three concepts. A full list of

these queries will be thus proposed to an end user.

In some cases, this list turns out redundant when

an end user has no interest in some specific queries

(e.g. the one concerning the luminosity of an

entrance door of a school). It is, therefore, important

to evaluate the cohesion between the queries chosen

and rejected by an end user and the corresponding

key concepts annotating these queries. For example,

we can notice that queries annotated by Entrance

and EntranceDoor are chosen more frequently than

the ones annotated by Door. Intuitively, Entrance

and EntranceDoor are semantically closer than

Entrance and Door (cf. Figure 4).

Figure 4: Semantic distances between subclasses of Door.

To propose a formal definition of the validation

of the conformity-checking ontology by usage, we

first define our approach on the evaluation of the

concepts of the conformity-checking ontology. It is

based on three main criteria (Karoui et al, 2007)

adapted for the conformity-checking problematic: (i)

credibility degree: we suppose that all concepts and

properties of the conformity-checking ontology are

defined by construction experts, their definitions are

pertinent and correct with the credibility degree

equal to 1; (ii) cohesion degree: we suppose that our

conformity-checking ontology is homogeneous:

there are subclasses of a class which are declared

equivalent by domain experts (e.g. door, entrance,

entranceDoor); (iii) eligibility degree: concepts and

relations are defined by experts and added to the

conformity-checking ontology, if they are necessary

for the formalization of conformity queries.

Our approach of the context-based validation of

the conformity-checking ontology by usage is

developed according to the same criteria, in order to

keep the semantic consistency of the conformity-

checking ontology: (i) credibility degree: no

concepts or relations can be defined by end non-

expert users; (ii) cohesion degree: the distance

between the equivalent concepts is then recalculated

according to the frequency of their simultaneous

choice by end non-expert users (e.g. Entrance and

EntranceDoor are chosen more often); (iii)

eligibility degree: if some classes of semantically

close concepts are defined, it can be interesting to

identify the concept characterising the whole class,

e.g. EntranceDoor for the class containing Entrance,

AccessibleEntrance, FrontDoor, etc. By identifying

the representative concept of the class, we can refine

WEBIST 2009 - 5th International Conference on Web Information Systems and Technologies

180

the semantic annotation of the corresponding queries

(for example, annotating them only by this concept)

and, consequently, the algorithms of expert

reasoning (for example, we do not need to schedule

queries which are annotated by the concepts of the

same class).

To model the semantic distances in the

conformity-checking ontology, we base on the

calculating of the semantic similarity in content-

based retrieval systems (El Sayed et al. 2007) and by

adapting the approach of the “intelligent evaluation”

(Karoui et al, 2007) of ontological concepts.

Currently, we work on the detailed development of

the conceptual approach for the evaluation of the

concepts of the conformity-checking ontology.

3.3 Adaptation of the Domain Ontology

for Different User Profiles

Our method for the development and usage-based

validation of the domain ontology remains still

generic and not adapted to the variety of different

actors of the construction domain. For this reason, it

is a real challenge to propose an approach for

adapting it for different user profiles: e.g. architect,

electric engineer, legal authority, etc.

In order to adapt the acquired domain ontology

for different user profiles, we propose to enrich our

CMV-method by personalising it for different user

profiles: the ECMV-method.

Our ECMV-method contains two main steps.

First, we identify the groups of users and the

corresponding user profiles. For each user profile,

we create a copy of the initial generic domain

ontology: e.g. the conformity-checking ontology for

(i) architects; (ii) electric engineers; (iii) conformity-

checking experts; and (iv) end non-professional

users.

Figure 5: Semantic distances between subclasses of Door

for different user profiles.

Second, we define the scope of interest for each

user profile. To do this, we apply the CMV-method

for each group of users and modify their copy of the

domain ontology according to its usage by the

corresponding end users. As result, we generalise or

detail the domain ontology according to the scope of

interest of user profiles. For example, it is important

only for an architect to distinguish between different

types of entrances (cf. Figure 5).

It is important to underline that the ECMV-

method guarantees the coherence and semantic

consistency between the generic domain ontology its

facets developed for different user profiles. This

coherence is based on the following aspects: (i)

credibility degree: the credibility of the facets of the

initial domain ontology is 0; the users can only

refine the distances between the concepts of the

initial domain ontology, but they can not create new

concepts; (ii) cohesion degree: the distance between

the synonym concepts is recalculated according to

the frequency of their simultaneous choice by users

of the same user profile (we do not aim at

establishing correspondences between different user

profiles); (iii) eligibility degree: if users of the same

user profile define semantically close concepts, these

concepts are grouped by a representative concept

which is the closest super class of these semantically

close concepts (e.g. Entrance, EntranceDoor,

FrontDoor classes of the initial domain ontology are

grouped by Entrance class of the ontology facet for

electric engineers).

4 IMPLEMENTATION:

ENRICHMENT OF THE C3R

PROTOTYPE FOR DIFFERENT

USER PROFILES

In our previous work on the development of the

conformity-checking model (Yurchyshyna et al.,

2008a), we have developed the C3R (Conformity

Checking in Construction with the help of

Reasoning) system (cf. Figure 6) that implements

the algorithms of reasoning by expert rules

according to organised conformity queries. For the

checking operation, C3R relies on the semantic

search engine CORESE (Conceptual Resource

Search Engine) (Corby, Faron-Zucker, 2007), which

implements RDF, RDFS and SPARQL languages

and answers SPARQL queries asked against an

RDF/OWL Lite knowledge base (Sowa, 1984); and

SEWESE (Sewese, 2008), the JSP/Servlet/Corese

environment for building Semantic Web

applications.

The main components of the C3R prototypes are:

(i) the knowledge acquisition module (query

IMPROVED DEVELOPMENT OF THE DOMAIN ONTOLOGY FOR DIFFERENT USER PROFILES - Application

Domain: Conformity Checking in Construction

181

formaliser, ontology editor; construction project

extractor); (ii) the reasoning module (checking

reasoner enabled by the CORESE engine; query

scheduler, conformity report generator); (iii) the

module on capitalisation of context knowledge

(query base generator; annotation editor; expert

reasoning explorer; formaliser of usage-based

knowledge).

Figure 6: C3R infrastructure: general view.

For the C3R prototype, we have defined a

conformity-checking ontology that currently

comprises around 2200 concepts and 1600

properties. The conformity-checking ontology is

written in the OWL-Lite language, which is rather

expressible and, at the same time, decidable. We

also define about 50 definition rules describing new

concepts and properties with the help of the ones

from the conformity-checking ontology.

To develop a base of conformity queries for the

validation of our approach, we chose 9 regulation

texts on the accessibility of public buildings (French

regulation base). These regulation documents

represent different classes of regulation texts (e.g.

norm, circular) and describe the accessibility

constraints of different entities: doors, routes,

signalisation, etc. With the help of CSTB experts,

we have identified about 350 simple text conformity

queries that resume these 9 regulation texts, which

are partially interpreted (around 65%). Other 35% of

identified queries are classified as non interpretable

and are not formalised. For practical validation of

our approach, we currently formalised and tested

about 100 conformity queries as SPARQL queries.

To adapt the C3R prototype for different user

profiles, we define 3 user profiles: architects,

engineers, and owners/non-professional end users.

For each user profile, we create a facet of the

conformity-checking ontology and define. The

calculation of the semantic similarity between the

concepts of the conformity-checking ontology

according to the DOUP-method is not implemented

yet. It is an objective of our future work on the

incremental implementation of the C3R prototype.

5 CONCLUSIONS

We presented a formal method for the development

of the domain ontology for different user profiles in

the context of conformity checking in construction.

Our semantic approach for the improved

development of the domain ontology for different

user profiles (the DOUP-method) comprises three

components: (i) the KRA-method aiming to

represent the knowledge for the conformity-

checking modelling; (ii) the CMV-method aiming

context modelling of the ontology by integrating the

results of semantic search of a query to check; and

(iii) the ECMV-method, which adapts the CMV-

method of the development of a domain ontology for

different user profiles. We also described the

conceptual architecture of the C3R prototype and

presented the current work on its implementation, to

illustrate feasibility of our approach.

One possible limitation of our work is that we do

not establish the semantic correspondences between

different facets of the domain ontology. This very

interesting research problem is not taken into

account by our semantic approach for the improved

development of the domain ontology for different

user profiles, and can be seen as a probable axe for

future research.

Our future works focus on the further

incremental development of the conformity-

checking ontology and the C3R prototype, as well as

their evaluation by domain experts and end users. In

particular, we will detail the DOUP-method to adapt

the C3R prototype for different user profiles, as well

as to create different facets of the conformity

checking ontology.

REFERENCES

Bell, H., Bjorkhaug, L., 2006. A buildingSMART

Ontology. In ECPPM-2006, European Conference on

Product and Process Modelling, Valencia, Spain, 185-

190

Brézillon, P., 2007. Context Modeling: Task Model and

Model of Practices. In CONTEXT 07, 6th

International and Interdisciplinary Conference on

Modeling and Using Context.

Berners-Lee, T., 2001. Reflections on Web Architecture.

Conceptual Graphs and the Semantic Web, 2001,

available at http://www.w3.org/DesignIssues/CG.html

Corby, O., Faron-Zucker, C., Implementation of SPARQL

Query Language based on Graph Homomorphism. In

15th International Conference on Conceptual

Structures (ICCS’2007).

WEBIST 2009 - 5th International Conference on Web Information Systems and Technologies

182

El-Diraby, T., Fiès, B., Lima, C. 2000. D3.6: The e-

COGNOS Ontology V1.1.0 - WP3. E-Cognos project

IST-2000-28671.

El Sayed A., Hacid H., Zighed A., 2007. A Context-

Dependent Semantic Distance Measure, In 19th

International Conference on Software Engineering

and Knowledge Engineering (SEKE), Boston, USA

Gauch, S., Speretta, M., Pretschner, A., 2007. Ontology-

Based User Profiles for Personalized Search, in A

Handbook of Principles, Concepts and Applications in

Information Systems, Integrated Series in Information

Systems (Ed: Sharman, R., Kishore, R., Ramesh, R.),

Volume 14, pp. 665-694

Gehre, A., Katranuschkov, P., 2007. InteliGrid

Deliverable D32.2 – Ontology Services,

www.inteliGrid.com

Hernandez, N., Mothe, J., Chrisment, C., Egret, D., 2007.

Modeling Context through Domain Ontologies. In

Information Retrieval 10 (2).

Karoui L., Aufaure M.-A., Bennacer N., 2007. Contextual

Concept Discovery Algorithm. In FLAIRS-20 the 20th

International FLAIRS Conference, AAAI Press.

Kerrigan, S. L., Law, K. H., 2005. Regulation-Centric,

Logic-Based Conformance Assistance Framework,

Journal of computing in civil engineering, ASCE,

19(1): 1-15, 2005.

Mokhtari, N., Dieng-Kuntz, R., 2008. Extraction et

exploitation des annotations contextuelles. In

EGC'2008, Volumes I, Sophia Antipolis, France, 7-18

Polanyi, M., 1983, The Tacit Dimension. Doubleday &

Co, 1966. Reprinted Peter Smith, Gloucester, Mass.

Sewese, 2008, available at http://www-

sop.inria.fr/teams/edelweiss/wiki/wakka.php?wiki=Se

wese

Sieg, A., Mobasher, B., Burke R., 2007. Ontological User

Profiles as the Context Model in Web Search, In

IEEE/WIC/ACM Intl Conference on Web Intelligence

and Intelligent Agent Technology, pp. 91 – 94

Smartcodes, 2008, available at

http://www.iccsafe.org/SMARTcodes/

Smart Market report, McGraw

Hill Construction, 2007, available at

http://www.iccsafe.org/SMARTcodes/SMRI_Final.pdf

Sutterer, M.; Droegehorn, O.; David, K, 2008. UPOS:

User Profile Ontology with Situation-Dependent

Preferences Support, In 1st Intl Conference on

Advances in Computer-Human Interaction, 230 – 235.

Yurchyshyna A., Faron-Zucker C., Le Thanh N., Zarli A.,

2008a. Ontological Approach for the Conformity-

Checking Modelling in Construction, In 10th

International Conference on Enterprise Information

System, ICEIS2008, Barcelone, Spain, p.139

Yurchyshyna, A., Faron-Zucker, C., Le Thanh N., Zarli,A.

2008b. Towards the Knowledge Capitalisation and

Organisation in the Model of Conformity-Checking

Process in Construction, Lecture Notes in Computer

Science; Springer Berlin / Heidelberg, Vol. 5177/2008

Yurchyshyna A., C. Faron-Zucker, N. Le Thanh, A. Zarli.

2008c. Towards an ontology-based approach for

formalising expert knowledge in the conformity-

checking model in construction, In ECPPM2008,

Sophia Antipolis, France

IMPROVED DEVELOPMENT OF THE DOMAIN ONTOLOGY FOR DIFFERENT USER PROFILES - Application

Domain: Conformity Checking in Construction

183