Sharing Knowledge in Daily Activity: Application in Bio-Imaging

Cong Cuong Pham

, Nada Matta

, Alexandre Durupt

, Benoit Eynard

, Marianne Allanic

1,4

Guillaume Ducellier

, Marc Joliot

and Philippe Boutinaud

Sorbonne University, University of Technology of Compiegne, Department of Mechanical Systems Engineering,

UMR CNRS 7337 Roberval, CS 60319, 60203 Compiegne Cedex, France

University of Technology of Troyes, 12 Rue Marie Curie, 10010 Troyes, France

GIN UMR 5296, CNRS, CEA, Bordeaux University, Case 71,146 rue Lo-Saignat, 33076 Bordeaux Cedex, France

Cadesis, 37 rue Adam Ledoux, 92400 Courbevoie, France

Keywords:

Knowledge Sharing, Ontology Product Life Cycle Management, Bio-Imaging.

Abstract:

Our approach uses the ontology to facilitate the data querying of users in the domain Bio-Imaging where

the data resources are heterogeneous and complex. The dependencies among data and the evolution of data

resources challenge users (especially for non-technician users) in querying the right data. Ontology can beused

to share the users’ understanding about data relationships to all community as well as to trace the database

evolution. As consequence, using ontology is a promising solution to facilitate the user’s query making process

and to enhance the query’s results.

1 INTRODUCTION

The evolution of data resources (new added data, new

generated relationships among data) challenges users

in data querying. In the context of distributed and het-

erogeneous data resources, since the new data com-

monly belong to an individual or group of individuals,

the others don’t understand the meaning of them as

well as the relationships between them and the exist-

ing data. The data querying becomes more and more

difﬁcult.

This inconvenience can be overcome by using on-

tology as a method to assimilate the understanding

about data and their relationships to all users. In

this paper, we present an ontology-based query sys-

tem which facilitates the query making process of

scientists in the domain of Bio-Imaging where the

heterogeneity and complexity of data resources have

been handled by using Product Lifecycle Manage-

ment (PLM) solutions (Allanic et al., 2013). Our

query system added a semantic layer between the sci-

entists and PLM system, which helps each user to

query the database without helps of database techni-

cians.

The rest of paper is organized as follows: The sec-

tion 2 and 3 deal with some related works in knowl-

edge sharing using ontology and the data querying in

PLM system. Then, in section 4 and 5, we are going

to present our ontology-based approach and its appli-

cation. The section 6 is reserved for discussion and

conclusion.

2 DATA QUERYING IN

BIO-IMAGING

During quotidian activities, researchers in Bio-

Imaging domain manipulate heterogeneous data like

human information, brain images (2D, 3D), diag-

nostic information ... They produce then new im-

ages, statistics data, diseases description. Informa-

tion grows quickly and they need a support that han-

dles the data evolving more efﬁciently than the ex-

isting database. At Gin (Groupe dImagerie Neuro-

fonctionnelle UMR CNRS 5296) Lab (Bio-Imaging

lab) researchers use Product Life-cycle Management

(PLM) solutions. By deﬁnition, a PLM system con-

sists of tools that enable the management of the

whole product data and related information thought

all phases of product lifecycle (Eynard et al., 2004).

It has been proven as an efﬁcient solution to tackle the

heterogeneity, complexity and the growth of data re-

sources to tackle the complexity, variety, heterogene-

242

Pham, C., Matta, N., Durupt, A., Eynard, B., Allanic, M., Ducellier, G., Joliot, M. and Boutinaud, P..

Sharing Knowledge in Daily Activity: Application in Bio-Imaging.

In Proceedings of the 7th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2015) - Volume 3: KMIS, pages 242-247

ISBN: 978-989-758-158-8

ity and growth of data resources.

In Bio-Imaging ﬁeld, for a new study, a scientist

needs not only the original data but also the processed

data of others in order to enhance the study results. It

is important to help the scientists to query himself the

database. However, the traditional PLM systems are

not enough ﬂexible, the data querying is only reserved

for database technicians.

3 KNOWLEDGE SHARING AND

PRODUCT LIFECYCLE

MANAGEMENT

3.1 Knowledge Sharing

Knowledge plays an important role in the long term

sustainability and success of organization. The need

for processes that facilitate the creation, sharing and

leveraging of individual and collective knowledge has

emerged recently for this reason. Knowledge sharing

(KS) has been introduced as one of the major activi-

ties of knowledge management and some deﬁnitions

of knowledge sharing can be found in the literature

(Small and Sage, 2006), (Hendriks, 1999). Some au-

thors (Sato et al., 2002), (Zhang et al., 2008) have

invested their efforts to construct platforms that en-

able knowledge sharing by using Information technol-

ogy ITs.(Sato et al., 2002) used XML Linking Lan-

guage (XLink) as a method of knowledge represen-

tation describing and proposed architecture for shar-

ing that knowledge among users. (Zhang et al., 2008)

tried to re-deﬁne knowledge resources in the network

by object-oriented thinking and proposed three-layer

knowledge sharing model. By using technologies on

Web 2.0, a knowledge-sharing system is built on the

Internet, allows the knowledge acquisition, sharing,

extension and retrieving.

Ontology is deﬁned as an explicit formal speciﬁ-

cation of a shared understanding. It has been used in

many knowledge sharing systems. For instance, (Yoo

and No, 2014) proposed a system based on ontol-

ogy expressing economics knowledge and Semantic

Web technologies (Domingue et al., 2011). This sys-

tem consists of ﬁve layers: registration, ontology, data

storage, reasoning and economic knowledge sharing.

Users can register economics knowledge pertaining

to a certain economics paper. They deﬁne then the

metadata and the relationships between notions dis-

cussed in the paper. According ontology model, the

system transforms this knowledge into semantic data

in a machine-understandable format. Two main func-

tions are basic search and knowledge navigation.

OntoShare (Davies et al., 2002) is another

ontology-based knowledge sharing. In his approach,

as users contribute information to the community, a

knowledge resource annotated with metadata is cre-

ated by using ontologies that have been predeﬁned

using Resources Description Framework Schema

(RDFS) and populated using RDF.

Some authors have successfully applied ontology

as a method for knowledge sharing, but they did not

consider the complexity of relationships between data

and related information, in order to enhance the data

querying. Before going to propose our approach that

uses ontology to share knowledge in PLM, let us to

describe some main features of PLM.

3.2 Product Lifecycle Management

Product Life-cycle Management (PLM) systems

integrate constantly all the information produced

throughout all phases of a product’s life-cycle to ev-

eryone in an organization at every level (managerial,

technical) (Sudarsan et al., 2005). We can ﬁgure some

key advantages of PLM systems:

• An effective PLM system reduces enormous data

resources to coherentdata ﬂows, avoids redundan-

cies.

• PLM permits the product structure management,

its evolution as well as the performed modiﬁca-

tions tracking.

• PLM enables the collaboration through virtual,

distributed and extended enterprises.

• PLM tackles the heterogeneity, complexity of the

data resources as well as the conﬁdentiality and

traceability issues.

However, along with these advantages, it also ex-

ists some issues:

• The lack of a standard between PLM systems

causes data integrity problems and limits the ac-

cess to and sharing of distributed product infor-

mation and knowledge .

• The increasing of need for product life-cycle

knowledge capitalization and reuse in order to re-

duce time and cost.

• PLM systems are not enough ﬂexible and the data

querying requires a good understanding about

data model. This issue challenges non-technicians

users in data exploitation.

The challenges for end-users (scientists for ex-

ample) come from the low-level expression of data

model as well as the evolution of database. In this

Sharing Knowledge in Daily Activity: Application in Bio-Imaging

243

paper, our approach used ontology to share the un-

derstanding about database structure and the depen-

dencies among data to all types of users in order to

enhance the data querying.

4 ONTOLOGY-BASED

APPROACH IN BIO-IMAGING

4.1 Bio-Imaging Data Management in

PLM

The bio-imaging data have been handled in PLM

database Teamcenter 9.1. We have interviewed some

researchers at GIN lab to identify their needs and dif-

ﬁculties in manipulating with information system dur-

ing their quotidian activities. The results have shown

that most of scientists had difﬁculties in data querying

due to the lack of understanding about data model and

data relationships. They almost cannot accomplish

this task without helps of database technicians. Pro-

viding an efﬁcient query interface for non-technicians

therefore becomes crucial.

Data querying becomes more and more complex

since each user introduces his own view when he/she

adds new data. Knowledge sharing techniques (on-

tology for example) we presented above help to de-

scribe the meaning of new data and relationships (an-

notations, descriptions) on the one hand to respect the

logic of a user and on the other hand to share knowl-

edge. We believe that our ontology-based approach

and inference engine will improve the data querying

of users at GIN (Figure 1).

Figure 1: Ontology-based knowledge sharing architecture

in PLM.

4.2 Ontology for Bio-Imaging

The main objective of using ontology is to facilitate

the data querying of Bio-Imaging scientists at GIN.

So, the concepts in ontology model need to refer

to concepts in the data model, and the lowest-level

concepts have to be linked to real data in the PLM

database.

4.2.1 BMI Data Model and Classiﬁcation

Teamcenter

Beside the data model, Teamcenter 10 provides a tool

to classify data into categories. Some queries can be

executed by using this classiﬁcation. Our ontology

has been built based on both data model and this clas-

siﬁcation.

Figure 2 presents the BMI-LM (Bio-Medical

Imaging - Lifecycle Management) data model used

in the PLM TeamCenter 9.1 (Allanic et al., 2013).

By adopting PLM solutions in the context of Bio-

Imaging, this PLM-oriented data model covered the

whole stages of a BMI study from speciﬁcations to

publications and enabled the ﬂexibility in data man-

agement. It contains three types of objects:

• Deﬁnition objects (Exam Deﬁnition Data Unit

Deﬁnition, Acquisition Deﬁnition, Processing

Deﬁnition,) represent the deﬁnition that are used

to keep the traceability of data provenance during

the whole lifecycle of a study.

• Result objects: Acquisition Result, Data Unit Re-

sult, Exam Result, Processing Result.

• Result objects (Acquisition, Data Unit, Exam,

Processing) consist all acquired data during a par-

ticular study.

• Reference objects (Bibliographical, Reference

Data) concern data that can be deﬁned and whose

can be represented device or system in the time.

Figure 2: BMI-LM DM implemented in Teamcenter 9.1.

Deﬁnition concepts have been created for the pur-

pose of data reuse. Some data have been acquired

under some conditions or by following a special pro-

tocol. These conditions and protocol are deﬁned in

a deﬁnition concept to trace the provenance of data.

For example, all the Processing results computed by

using the same Acquisition device and the Processing

parameter can be attached to the same corresponding

Processing deﬁnition.

The classiﬁcation (Figure 3) has been built based

on the data model. From that, BMI data have been

KMIS 2015 - 7th International Conference on Knowledge Management and Information Sharing

244

classiﬁed into branch, classes and sub-classes. The

classiﬁcation allows a speciﬁc class to be added to a

generic item (object in the data model). At lowest-

level, data in PLM database have been organized in

tables based on the data model. The structure of clas-

siﬁcation, the classes’ attributes can be modiﬁed due

to the evolution of data resource. It provide the ﬂex-

ibility capacity to users when they manipulate with

data (Allanic et al., 2013).

Figure 3: Classiﬁcation in Teamcenter 9.1 corresponding

with the BMI-LM data model.

However, the nature of data in this classiﬁcation

can be eventually repeated. It means some data can be

classiﬁed in two difference classes. This issue make

users confusing when they make a query. Further-

more, when the number of classes, sub-classes grows,

the relationships among these classes become com-

plex and can be not handled with out annotation.

To overcome this issue, we build an ontology,

which bases on both of data model and classiﬁcation.

This ontology shows the logic of information in Bio-

Imaging and it provides an overview of all concepts

and relationships in the data model as well as in the

classiﬁcation.

4.2.2 Ontological Model Construction

There is a lot of ontology deﬁned in medicine,

for instance, in oncology, in neurology... (Gibaud

et al., 2011). But little works focus on Bio-Imaging

ﬁeld. (Gibaud et al., 2014) deﬁned concepts used in

Bio-Imaging like: Dataset, Processing, Investigators,

Medical Image ﬁles, Equipment, subjects, etc. (Fig-

ure 4)

Based on the data model BMI-LM and the clas-

siﬁcation in Teamcenter, we adapt this ontology to

the logic use of information in database. Infor-

mation belongs to three major categories: Tool,

Data and Process corresponding with acquisition de-

vice/software tool, acquisition/processing results, ac-

quisition/processing deﬁnition (Figure 5). This hi-

Figure 4: Ontology for Bio-Imaging (Eynard et al., 2004).

erarchy is more logic and understandable by non-

technicians users.

We deﬁne then relationships/restrictions among

concepts and sub-concepts like Uses, Generates, In,

Refers to, Passed For example, StudySubject have

passed Acquisition Process and this process gener-

ates some Acquisition Results. These relationships

are identiﬁed from the interview with scientists and

they reﬂect their work logic.

In the next section, we are going to present the

ontology-based query system as the application of our

approach.

Figure 5: Ontology at GIN: a) Ontology tree b) Ontology

graph.

5 ONTOLOGY-BASED QUERY

SYSTEM

Using ontology facilitates users in query making pro-

cess. The ontology tree and graph help users to under-

stand the relationships among concept and to directly

deﬁne query parameters. User can also chose a query

in the list of history queries to re-execute, modify or

complete it. Figure 6 illustrates the architecture of our

query-system.

Sharing Knowledge in Daily Activity: Application in Bio-Imaging

245

Basically, to query data in PLM database, user can

do as follows:

1. User identiﬁes query concept and related query

concept in the ontology tree or graph.

2. In the description table, user sees all information

about current selected concept: Annotation, con-

cept usage, attributes and relationships (restric-

tions).

3. From the query concept, user navigates to all oth-

ers related concepts by following the predeﬁned

restrictions. At each related concept, user can de-

ﬁne the value for its attributes.

4. The step 3 is repeated until all related concepts

have been chose.

5. A mapping table between ontology concepts and

objects in data model will be used to automati-

cally generate a query which can be understand-

able and executable by the query engine.

6. The results will be displayed in the same query

interface, in a table or a graph.

We would to take here an example of query fre-

quently used by scientist at GIN lab:

Figure 6: Query making process using ontology.

Query all the results of RaventIQ test that was

performed on all male and right-handed subjects

who passed a T1 scan acquisition.

Here, RaventIQ is a test in Treatment Process

(Processing), while T1 scan is a test in Acquisition

Process. These tests have been performed on human

(StudySubject - A Subject in a predeﬁned Study)

(Figure 2). If a researcher tries to use SQL to answer

this question, he has to know in which tables these

data saved and of course the relations among tables.

This is the SQL statement made by database techni-

cians:

SELECT sujet.codesujet, test.etiquette,

passation.valeur, lateralite.main_ecrit,

fichier.nomfichier, sujet.sexe

FROM sujet, passation, lateralite, test,

examen, acquisition, fichier

WHERE sujet.id = passation.sujet

AND sujet.id = examen.sujet

AND sujet.id = lateralite.sujet

AND examen.id = acquisition.examen

AND acquisition.fichier = fichier.id

AND passation.test = test.idtest

AND passation.valeur NOT LIKE ’’

AND test.etiquette = "ravenQi"

AND lateralite.main_ecrit = ’D’

AND fichier.nomfichier LIKE ’%_t1_%’

AND sujet.sexe = ’H’

ORDER BY sujet.codesujet

However, the query construction process can be

simpliﬁed by using ontology. Each scientist knows

that he can ﬁnd the results of RaventIQ test in Treat-

ment Result, so he chooses concept Treatment Result

in Ontology Tree (or graph) as the query concept.

From Description Table, he know that this concept is

related to StudySubject concept by property: isCom-

putedFromDataOf. He chooses this concept and sets

the value Male for subject’s sex attribute. Here, he

will see that a StudySubject has passed some Acquisi-

tion processes which generate Acquisition Results. He

continues choosing related query concepts and sets

the value for its attributes (right-handed and T1).

We use TCXquery, a tool that considers PLM data

as XML documents, as the Query Processor. So when

user ﬁnishes, a query will be automatically generated

in .xquery format by using a table that maps a restric-

tion between two ontology concepts to a predeﬁned

route between two objects in data model. With this

mapping table, users don’t have to know previously

the data model. At the end, .xquery query will be sent

to and executed at server by using a web service (Fig-

ure 6).

Figure 7 illustrates our query interface where the

results obtained are represented as graph in the same

interface in order to facilitate the sharing of data and

information among users for further purpose.

Figure 7: Ontology-based query system in Bio-Imaging.

KMIS 2015 - 7th International Conference on Knowledge Management and Information Sharing

246

6 CONCLUSION

In this paper, we presented an ontology-based query

system that facilitates the query making of users in

Bio-Imaging domain. The main component of our ap-

proach is the ontological model and the table mapping

between this ontology and the data model in PLM

system. Another important factor is the relationships

deﬁnition among concepts of ontology. It reﬂects on

the one hand the logic work of users and on the other

hand, it determines the accuracy of our approach.

As future work we will focus on the test of

proposed query interface with various queries sets

(in BioImaging domain) and engineering design (in

PLM). The ontology tree and ontology graph must be

also developped to cover all concepts in BioImaging

domain. Ontology will be implemented in semantinc

web language (RDF, SPARQL) in order to more use

inference engine for information search.

ACKNOWLEDGEMENTS

The work presented in the paper is conducted

within the ANR (Agence Nationale de la Recherche)

founded project BIOMIST (noANR-13-CORD-0007)

for the matic axis no2 of the Contint 2013 Call for

Proposal: from content to knowledge and big data.

REFERENCES

Allanic, M., Durupt, A., Joliot, M., Eynard, B., and Boutin-

aud, P. (2013). Application of PLM for Bio-Medical

Imaging in Neuroscience. In Bernard, A., Rivest, L.,

and Dutta, D., editors, Product Lifecycle Management

for Society, number 409 in IFIP Advances in Informa-

tion and Communication Technology, pages 520–529.

Springer Berlin Heidelberg.

Davies, J., Duke, A., and Stonkus, A. (2002). OntoShare:

Using Ontologies for Knowledge Sharing. In Seman-

tic Web Workshop, volume 72.

Eynard, B., Gallet, T., Nowak, P., and Roucoules, L. (2004).

UML based speciﬁcations of PDM product structure

and workﬂow. Computers in Industry, 55(3):301–316.

Gibaud, B., Forestier, G., Benoit-Cattin, H., Cervenan-

sky, F., Clarysse, P., Friboulet, D., Gaignard, A.,

Hugonnard, P., Lartizien, C., Liebgott, H., Montag-

nat, J., Tabary, J., and Glatard, T. (2014). OntoVIP:

An ontology for the annotation of object models used

for medical image simulation. Journal of Biomedical

Informatics, 52:279–292.

Gibaud, B., Kassel, G., Dojat, M., Batrancourt, B., Michel,

F., Gaignard, A., and Montagnat, J. (2011). Neu-

roLOG: sharing neuroimaging data using an ontology-

based federated approach. AMIA ... Annual Sympo-

sium proceedings / AMIA Symposium. AMIA Sympo-

sium, 2011:472–480.

Hendriks, P. (1999). Why share knowledge? The inﬂu-

ence of ICT on the motivation for knowledge sharing.

Knowledge and Process Management, 6(2):91–100.

Sato, H., Otomo, K., and Masuo, T. (2002). A knowl-

edge sharing system using XML Linking Language

and peer-to-peer technology. pages 26–27.

Small, C. T. and Sage, A. P. (2006). Knowledge manage-

ment and knowledge sharing: A review. Information,

Knowledge, Systems Management, 5(3):153–169.

Sudarsan, R., Fenves, S. J., Sriram, R. D., and Wang, F.

(2005). A product information modeling framework

for product lifecycle management. Computer-Aided

Design, 37(13):1399–1411.

Zhang, J., Liu, Y., and Xiao, Y. (2008). Internet Knowledge-

Sharing System Based on Object-Oriented. In Second

International Symposium on Intelligent Information

Technology Application, 2008. IITA ’08, volume 1,

pages 239–243.

Sharing Knowledge in Daily Activity: Application in Bio-Imaging

247