Semantic Models in Web based Educational System Integration

Geraud Fokou Pelap, Catherine Faron Zucker and Fabien Gandon

University C

ote d’Azur, CNRS, INRIA, I3S, Nice, France

Keywords:

e-Education Information System, e-Education Model, Learning Environment, Education Standards, Ontology,

Semantic Web, Benchmarking.

Abstract:

Web based e-Education systems are an important kind of information systems that beneﬁted from Web stan-

dards for implementation, deployment and integration. In this paper we propose and evaluate a semantic Web

approach to support the features and interoperability of a real industrial e-Education system in production.

We show how ontology-based knowledge representation supports the required features, their extension to new

ones and the integration of external resources (e.g. ofﬁcial standards) as well as the interoperability with other

systems. We designed and implemented a proof of concept in an industrial context that was qualitatively and

quantitatively evaluated and we benchmarked different alternatives on real data and real queries. We present a

complete evaluation of the quality of service and response time in this industrial context and we show that on

a real-world tesbed Semantic Web based solutions can meet the industrial requirements, both in terms of func-

tionalities and efﬁciency compared to existing operational solutions. We also show that an ontology-oriented

modelling opens up new opportunities of advanced functionalities supporting resource recommendation and

adaptive learning.

1 INTRODUCTION

E-education systems are often at the intersection of

information systems and Web based systems. They

leverage state of the art results of information sciences

and technologies (IST) as well as the Web architecture

and resources to support educational processes and

the management of their users (learners and teachers),

pedagogical resources (courses, exercises, etc.), reg-

ulations (e.g ofﬁcial reference standards) and integra-

tion across different systems and actors in particular

to ensure compatibility and seamless user-experience.

Since education is under the responsibility of pub-

lic authorities, educational solutions developed by

public or private organizations must comply with the

public authorities speciﬁcations. Taking the example

of France, as part of the Education Code (Minist

ere

de l’

education nationale, 2018), the Ministry of Edu-

cation has deﬁned and published in the French Ofﬁ-

cial Journal a common reference base of knowledge

and skills

. It standardizes the content of courses by

specifying knowledge and skills that a student must

acquire at each step of her school curriculum. Ad-

ditionally, the French Ministry of Education speci-

original name: Socle commun de connaissance, de

comp

etences et de culture

ﬁes a format for digital pedagogical resources de-

scription called ScoLOMFR (R

eseau Canop

e, 2011).

It is based on the IEEE standard Learning Object

Metadata (LOM) (committee, 2002) and its French

version, LOMFR

. ScoLOMFR speciﬁes a descrip-

tion schema and a common vocabulary for all online

pedagogical resources for their indexing and sharing

among different e-Education actors in France. As a

result, any learning environment developed by pub-

lic institutions or private companies must meet these

standards and norms to ensure a wide dissemination,

whatever the educational context. Moreover, they

must have updating capabilities to adapt to the possi-

ble evolution of these standards. Semantic Web tech-

nologies stand as a solution to achieve these goals, of-

fering open standards for ontology-based knowledge

representation, with extensible schemata, and data in-

tegration and interoperability.

In this paper, we show beneﬁts of Web Informa-

tion systems and technologies in e-Education context.

We present the results of an ontology-based educa-

tional knowledge modelling and management expe-

rience in a real e-Education environment: the learn-

ing solution developed by the Educlever company.

We address the following questions: (1) can an in-

http://www.lom-fr.fr

Pelap, G., Zucker, C. and Gandon, F.

Semantic Models in Web based Educational System Integration.

DOI: 10.5220/0006940000780089

In Proceedings of the 14th International Conference on Web Information Systems and Technologies (WEBIST 2018), pages 78-89

ISBN: 978-989-758-324-7

dustrial educational system in production rely on se-

mantic Web technologies? (2) Does semantic Web

ontology-oriented modelling effectively support edu-

cational system integration? (3) Does a semantic Web

educational system support additional features? In or-

der to answer these questions, we provide a proof of

concept by implementing ontology-based integration

and augmentation of different systems, sources and

actors of e-Education and benchmarking them in an

industrial real-world context.

Our proposed solution relies on EduProgression

ontology (Rocha et al., 2016) which is modelling the

ofﬁcial common base of knowledge and skills, and

which we extended to meet the speciﬁc needs of the

Educlever solution. Starting from the technical so-

lution originally adopted by Educlever, mainly based

on a relational database of educational resources and

a graph database of educational concepts and skills

indexing these resources, we developed an alternative

Semantic Web based solution with (1) an ontology of

educational concepts and skills, (2) a repository of se-

mantic annotations of pedagogical resources, and (3)

a base of queries on this repository implementing the

functionalities offered by the existing solution and ad-

ditional ones. We show the feasibility of our solution

in a real industrial context by implementing it within

four off-the-shelf triplestores: Allegrograph, Corese,

GraphDB and Virtuoso. We benchmark the existing

and new solutions on real data and queries and per-

form evaluation of the quality of service and response

time. The results of our evaluation show that the se-

mantic Web based solution meets the industrial re-

quirements, both in terms of functionalities and efﬁ-

ciency. Moreover, we show that our ontology-based

modelling opens up new opportunities of advanced

functionalities supporting resource recommendation

and adaptive learning.

This paper is organized as follows: Section 2

presents state-of-the-art Educational ontologies and

triple stores. Section 3 presents our proposed Seman-

tic Web based modeling of educational systems which

meets public standards. Section 4 proposes a Seman-

tic Web architecture for educational systems and show

how it improves the Educlever solution. Section 5

evaluates and compares Web based integration propo-

sitions. We perform this evaluation in the Educlever

context, providing data and queries which implement

real industrial requirements, on different triplestores

and we compare them to each other and to the ex-

isting Educlever solution. Section 6 summarizes our

contributions and provides several perspectives.

2 RELATED WORK

2.1 Educational Ontologies

The interest of ontologies in the domain of e-

Education has been repeatedly pointed out during the

last decade. In (Jaffro, 2007), the author analyses the

reasons and ways to use ontologies in e-Education and

for which goals. Many ontologies have been proposed

and designed for dedicated applications. Among them

CURONTO (Al-Yahya et al., 2013) is an ontological

model dedicated to curriculum management and to fa-

cilitate program review and management.

In (Rani et al., 2016) the authors propose an e-

learning management system based on an ontology

modelling all the dimensions of the system. Other

works on ontology modelling deal with the produc-

tion of pedagogical resources: (Gueffaz et al., 2014)

and (Rocha et al., 2016) propose ontologies built from

French ofﬁcial texts describing curriculum and pop-

ulate such ontology. Finally, ontology engineering

can support the management of the learning process.

In (Gascue

na et al., 2006), the authors use an ontol-

ogy to describe the learning material that compose a

course, to provide adaptive e-learning environments

and reusable educational resources. In a similar way,

(Hyun-Sook and Jung-Min, 2012) and (Hyun-Sook

and Jung-Min, 2014) have as primary objective to

develop an ontology-based learning support system

which allows the learners to build adaptive learning

paths through the understanding of curriculum, syl-

labuses, and course subjects. In OntoEdu (Guangzuo

et al., 2004), the authors propose to use Semantic Web

technologies to implement a service layer which will

allow an automatic discovery, invocation, monitoring

and composition of learning paths.

(Al-Yahya et al., 2015) and (Alsultanny, 2006)

presented a review and overview of works on ontolo-

gies in the domain of e-Education. They map works

to different needs that ontologies can address. (Al-

Yahya et al., 2015) classify ontologies in E-learning

context into four categories: (1) curriculum mod-

elling and management, (2) describing learning do-

mains, (3) describing learner data and (4) describing

e-Learning services. But, to the best of our knowl-

edge, none of the ontologies reported in the literature

has been used in an industrial context, or evaluated

on the data of an EdTech company. Moreover, the

proposed ontologies do not integrate public author-

ity recommendations or standards model. This is pre-

cisely what we will focus on in this paper: We pro-

pose and evaluate an ontology-based solution mod-

eling public recommendations to answer the require-

ments of Edtech company Educlever. Our solution

Semantic Models in Web based Educational System Integration

relies on the Eduprogresion (Rocha et al., 2016) on-

tology which models the Common base of knowledge,

skills and culture published by the French ministry

of national education in 2016. It speciﬁes the set of

knowledge and skills that must be mastered by stu-

dents to build their personal and professional future

and succeed in life in society. It also speciﬁes the po-

sitioning of knowledge and skills in the different cy-

cles of primary and secondary school, and therefore

the learning progression.

Figure 1 presents the main concepts of the

Eduprogression ontology. The key concept is that of

element of knowledge and skill (EKS), which should

be acquired by a learner in his curriculum in a given

course at a given cycle. Each element has at least one

learning domain among the ﬁve deﬁned by French

ministry of education: languages for thinking and

communicate, methods and tools to learn, formation

of the person and the citizen, natural systems and

technical systems, representation of the world and the

human activities. The concept of Progression is an-

other key concept which represents the program of

study for a subject (course) at a particular level (cy-

cle). In the last version of the recommendation, a

progression is deﬁned for an EKS and a learning do-

main. Our ontologies in this paper will start from the

Eduprogression ontology and extend it to cover the

needs of a speciﬁc actor of e-eductation.

Figure 1: Ontology Eduprogression.

2.2 Off-the-shelf Triplestores

Triplestores or RDF store systems are software solu-

tions to store data represented in RDF format. These

last years, development of triple stores has ﬂourished.

Today there are more than 20 systems available

. In

order to help developers make the right choice among

all these systems, many benchmarks have been de-

signed (Wu et al., 2014; Mironov et al., 2010). But

these benchmarks have some limitations: most of

them rely on artiﬁcial data and/or hypothetical use

cases while using target data improves benchmarking

and helps for the right choice (Jean et al., 2012).

In order to conduct a comparative evaluation on

the Educlever use cases and data, we ﬁrst chose

https://fr.wikipedia.org/wiki/Triplestore and https://db-

engines.com/en/ranking/rdf+store

several triplestores by distinguishing between native

RDF triplestores, designed and dedicated to store

RDF data, and non native RDF triplestores, designed

for another type of data (e.g. relational data) but

adapted to store RDF data. Among native RDF triple-

store, we distinguished between in-memory triple-

stores and triplestores with persistent storage. As a re-

sult, we chose the four following triplestores: Corese

is an in-memory triplestore; it loads all the ontologies

and RDF data when starting the application and saves

it in an RDF ﬁle when exiting it. Allegrograph and

GraphDB (OWLIM) both are native RDF triplestores

with persistent storage capabilities. Finally, Virtuoso

is a non native RDF triplestore.

As detailed latter in the paper, for the benchmark-

ing of these triplestores we translated the Educlever

dataset into RDF, relying on a dedicated ontology and

we considered the Educlever requirements and we im-

plemented them with SPARQL. In the next section

we present our Semantic Web based modeling of the

Educlever data and needs.

3 ONTOLOGY BASED

MODELLING OF SKILLS,

KNOWLEDGE AND

PEDAGOGICAL RESOURCES

In this section, we propose an ontology-based model

to represent knowledge and skills referential and also

pedagogical resources. Beforehand, the Educlever

solution relied on relational and graph databases to

store them and had limitations to integrate heteroge-

neous data without losing information and to infer

new information from it. The ontology-based model

of skills, knowledge and pedagogical resources pre-

sented in the following has been setup in the Educle-

ver software infrastructure.

Our solution relies on two linked datasets. The

ﬁrst one is called Referential, it describes and con-

tains all the elements of knowledge and skill avail-

able through the e-Education solution, Educlever

for our case study. The main concept is Cocon,

which stands for ”COmp

etences et CONnaissances”

in French (skills and knowledge). The second dataset

is called Corpus, it describes and stores all pedagogi-

cal resources available through the e-Education solu-

tion. Corpus is described using a speciﬁc vocabulary,

with OPD as key concept, which stands for ”Objet

edagogique” in French (Pedagogical Object). We

formalized this vocabulary and underlying concepts

into an ontology which reuses and extends EduPro-

gression.

WEBIST 2018 - 14th International Conference on Web Information Systems and Technologies

(a) Referential Ontology. (b) Corpus ontology.

Figure 2: Educlever Ontology.

3.1 Knowledge and Skills Modelling

The concept of Cocon is the keystone of the Ref-

erential modelling. It represents an atomic element

of knowledge or skill learnt by students on the e-

Education solution. An example of Cocon is the mul-

tiplication of two integers identiﬁed with URI educle-

ver:MultiplyTwoIntegers

in the Educlever system.

We formalize this concept as a class equivalent to EKS

from the ontology Eduprogression, thus integrating

public standards description. Figure 2a presents the

Educlever Referential ontology. Each Cocon can be

described by indicating its learning domain(s), course

and cycle using respectively properties hasLearning-

Domain, hasCourse and hasCycle deﬁned on class

EKS in ontology Eduprogression. For instance, the

learning domain of the multiplication of two integers

is the ﬁrst domain of French education standards, lan-

guages for thinking and communicate, its course is

Mathematics and its cycle is the second cycle.

There are two others classes: Knowledge and Sta-

tus. Knowledge specializes Cocon, and gathers ab-

stract elements of knowledge. For example, Arith-

metic is an instance of Knowledge. Status speciﬁes

the current state of an instance of Cocon in its life

cycle in an e-Education solution; its instances are in

creation, in updating or deleted.

Referential comprises two mains properties: has-

Status to associate a status to a cocon, and isRelat-

edTo to link two cocons. The latter is specialized into

ﬁve properties specifying the nature of the relation:

skos:broader (in particular any instance of Knowl-

edge is related to other cocons representing more spe-

ciﬁc elements of knowledge or skill), isComplexiﬁca-

tionOf states that a cocon goes more in depth than an-

other, isFollowedBy expresses a progression between

two instances of Cocon, isPrerequisiteOf and isUn-

derstandingLeverOf states that a cocon helps to un-

derstand another.

educlever: http://www.educlever.fr/edumics/refeduclever#

The uses of the Referential ontology in the

Educlever platform are twofold: (1) It enables to

describe the knowledge and skills developed by the

company for learners and to link them to the stan-

dard published by the French education ministry. (2)

It is used in combination with the ontology of peda-

gogical resources described in the following, to eval-

uate the acquisition of elements of knowledge or skill

by learners and to recommend them relevant peda-

gogical resources. Moreover, by relying on semantic

Web models and technologies we can reuse, extend

and align with existing vocabularies to increase inter-

operability. The adopted solution is compliant with

linked data Web architecture and principles such as

derefenceable URIs.

3.2 Pedagogical Resources Modelling

Figure 2b presents the Corpus ontology. The concept

of pedagogical object (OPD) is the keystone of Cor-

pus. It represents a pedagogical resource created to

learn and acquire knowledge or skills. It is formal-

ized as a class which is the range of all the properties

declared in the ontology.

There are two key properties: Property worksOn

enables to link an instance of OPD and an instance of

Cocon from the Referential ontology, representing an

element of knowledge or skill tackled in the pedagog-

ical resource. It is specialized into three properties

specifying the nature of the relation, the role of the

OPD relatively to the Cocon: isLearningOf, isTrain-

ningOf, and isEvaluationOf ). The other key property

is hasOPD, linking two OPDs. It enables to repre-

sent partonomies, expressing how some pedagogical

resources are composed as a combination of other re-

sources, which may be reused for composing differ-

ent other pedagogical resources. AutonomousOPD is

the subclass of OPD gathering the resources which do

not need any other resources to be used. Three other

properties enable to associate a pedagogical resource

to a course, a learning domain and a status in the

Semantic Models in Web based Educational System Integration

life cycle of Educlever resources. Thanks to Corpus

model, e-Education company could provide pedagog-

ical resources annotated on public standards and so,

could be evaluated by the public authority. Moreover,

based to this model, private companies could share

pedagogical resources mainly when theses pedagogi-

cal resources allow to learn or evaluate many different

skills and knowledge.

4 SEMANTIC WEB BASED

ARCHITECTURE FOR

E-EDUCATIONAL SYSTEM

In this section we propose a Semantic Web based

architecture, relying on triplestores, to manage the

above described ontology-based modelling of skills,

knowledge and pedagogical resources. We use this

architecture to upgrade the existing software archi-

tecture of the Educlever solution. We ﬁrst brieﬂy

describe the initial industrial architecture before ex-

plaining the proposed evolution.

4.1 Case of a Real e-Education

Information System in Production:

The Educlever Solution

The ﬁrst version of the Educlever system was built on

top of a relational database storing the pedagogical

resources. Two tables were used: the ﬁrst one storing

OPD’s attributes like status, title, author and type; the

second one storing the course and cycle of each OPD

and the partonomic relations between them. Based

on this relational database, the three main functional-

ities implemented are: (i) ﬁnd OPDs relative to a par-

ticular course and/or cycle, (ii) ﬁnd OPDs contained

in a given OPD and (iii) ﬁnd OPDs by combining

the two previous criteria. The tree structure storing

the partonomy of OPDs is also useful for interactive

exploration of the dataset of pedagogical objects by

users through a dedicated web interface.

A second version of the Educlever platform was

built to enable the implementation of new function-

alities exploiting Cocons, to support the construction

of learning paths and the evaluation of learners, e.g.

the computation of the accessiblility of a Cocon by a

learner, based on the evaluation of the acquisition of

prerequisite Cocon, or the computation of the degree

of understanding of a Cocon by a learner. To represent

property chains on Cocons a relational database was

not efﬁcient, obliging to perform joins between table

Cocon and itself. Then, Educlever upgraded its plat-

form by adding a graph database (OrientDB) to rep-

resent the relations between Cocons. Based on this

graph database, the two main functionalities imple-

mented are: (i) ﬁnd all the prerequisites of a given

Cocon and, recursively, the prerequisites of prerequi-

sites, (ii) ﬁnd all narrower Cocons of all direct prereq-

uisites of a given Cocon.

Indexer

Cocon instances

Query Search

Educlever.com

Cocon exploration

Corpus (OPD)

OPD ID

ScolomFR Doc

Ref

ScolomFR

Cocon ID

Figure 3: Existing Architecture of the Educlever Solution.

The overall architecture of the Educlever solution

is depicted in Figure 3. What this description of a

real industrial system also stresses is that there is a

need for approaches taking into account the existence

of legacy information systems and their integration,

extension and evolution.

4.2 e-Education System Architecture

based on Semantic Web

Technologies

We propose two architectures based on Semantic Web

technologies to design an e-Education system. They

are built on top of triple stores to store and process

RDF data from the Referential and Corpus datasets.

In the ﬁrst architecture (Simple architecture) we

used a triplestore to store both Referential and Cor-

pus datasets into a single graph. As depicted in Figure

4a, the Educlever solution relies on a SPARQL end-

point Web service queried with HTTP requests. Let

us note that with this architecture each functionality

is implemented by a single SPARQL query, whereas

with the current architecture (Figure 3) some func-

tionalities are implemented by combining the results

of several queries to different database systems, with

different query languages.

Indeed, in the current solution, the Educlever

data relative to Cocons and OPDs are separated in

two databases. This decision was motivated by the

fact that these two databases can support different

functionalities and are used in different processes.

The graph database on Cocons is used for learning

path design and Cocon evaluation while the relational

database on OPDs is used for OPD creation by the

pedagogical team and for learners training, learning

and evaluation. So, a failure of one database does

not affect the processes exploiting the other one which

WEBIST 2018 - 14th International Conference on Web Information Systems and Technologies

Query Search

Educlever.com

Corpus Referential Eduprogression

Cocon exploration

Cocon & OPD

Instances

(a) Simple Architecture.

Cocon instances

Query Search

OPD Instances

SPARQL Federated

Endpoint

Educlever.com

Corpus

Referential Eduprogression

Cocon exploration

(b) Federated Architecture.

Figure 4: Semantic Web based Architecture of e-Education solution.

can continue their execution. The main With this ar-

chitecture, Educlever limits the impact of a failure in

exploitation on one database. In order to add this

ﬂexibility in a semantic Web based architecture, we

proposed a architecture (Federated architecture) rely-

ing on a SPARQL federated Endpoint. As depicted

in Figure 4b this federated endpoint allows us to sep-

arate the two datasets, Referential and Corpus, thus

preventing failure while continuing to query them as

a single dataset. This context and scenario is typical

of the need to take into account legacy software, in-

formation system and organizational constraints from

real industrial contexts as well as the service quality

constraints, etc.

5 EVALUATION OF THE

SEMANTIC WEB

INTEGRATION EFFICIENCY

We led some experiments to evaluate the two pro-

posed e-Education system architectures based on Se-

mantic Web technologies. For this evaluation we im-

plemented real use cases from the Educlever com-

pany, with its real data stored in the Referential and

Corpus datasets. Here we report the results of (i) a

qualitative evaluation of the proposed semantic Web

based solution consisting in comparing the number of

use cases that can be implemented within this solu-

tion to the number of them that are implemented in

the current Educlever solution (section 5.1); and (ii)

a quantitative evaluation of the proposed solution, fo-

cusing on the execution cost time of the queries im-

plementing the use cases (section 5.2).

5.1 Qualitative Evaluation:

Implementability of the Use Sases

The existing Educlever system has been designed to

address the company use cases. Here we present these

use cases classiﬁed into four categories: (i) use cases

exploiting dataset Referential only, from C

to C

, (ii)

use cases exploiting dataset Corpus only, from C

, (iii) use cases exploiting both datasets, from C

to C

, and (iv) use cases requiring querying property

paths between cocons on dataset Referential, from C

to C

. The SPARQL queries we wrote to implement

these use cases are given in Table 3 in Appendix; each

use case C

is implemented by a query Q

1. Find All Direct Prerequisites of a Given Cocon

c: this is used to check whether a learner is ready

to work on c or if he needs to work on some pre-

requisites before.

2. Find All Direct Narrower Cocons of a Given

Cocon c: this is mainly used for the exploration

of the Referential dataset, starting with high level

Cocons and iteratively going down by following

the broader/narrower relations.

3. Find All the Cocons Such That a Given Cocon c

is in Their Prerequisites: this is used to identify

the candidate Cocons for the next learning step af-

ter working on Cocon c.

4. Find All Direct Prerequisites of a Given Cocon

c and of Its Direct Narrower Cocons: this is

used to score all these Cocons when a learner has

successfully validated c.

5. Find All Prerequisites of All the Cocons Which

Are Understanding Levers of a Cocon c

Which

is a Complexiﬁcation of a Given Cocon c: this

is used to ﬁnd alternative (longer) learning paths

to learn a Cocon c which seems to be complex.

6. Find All OPDs Which Evaluate a Given Cocon

c: this is used to build an evaluation OPD of c.

7. Find All the Information about a Given OPD o:

status, course and learning domain.

8. Find All OPDs Which are All Useful to Eval-

uate and Learn a Given Cocon c: recommend

evaluation OPDs for learning. The goal of this

use case is used to prepare the learners to an eval-

uation session by using evaluation OPDs during

learning stage.

Semantic Models in Web based Educational System Integration

Table 1: Implementation of the use cases depending on the tested architectures.

Referential Corpus Both datasets Path queries

R1 R2 R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R13 R14

educ-v2 4 4 4 4 6 4 4 6 6 6 6 4 6 6

all other implementations 4 4 4 4 4 4 4 4 4 4 4 4 4 4

9. Find All OPDs Useful to Evaluate Both a Given

Cocon c and All its Prerequisites: this supports

the recommendation of OPDs in order to speed up

the study.

10. Find All evaluation OPDs More Simple Than

a Given OPD o, considering the complexiﬁcation

relations between the Cocons these OPDs are re-

lated to: this is used to recommend OPDs to eval-

uate a learner.

11. Find All OPDs Useful to Understand a Given

Cocon c: these OPDs are related to c with an in-

stance of relation isTrainingOf or linked to Co-

cons c

related to c with relation isUnderstandin-

gLeverOf.

12. Recursively Find All Direct or Indirect Prereq-

uisites of a Given Cocon c: this involves evaluat-

ing learning paths of property isPrerequisiteOf.

13. Find All Cocons Within a Prerequisite Path Be-

tween Two Cocons c

and c

14. Infer Implicit Prerequisite Paths Between Two

Cocons c

and c

: ﬁnd the simplest Cocons asso-

ciated to more complex Cocons in the path.

As Table 1 shows it, the semantic Web based pro-

posed solutions implement all of the use cases while

the current version of the Educlever solution imple-

ments only eight of them. The functionalities which

are difﬁcult or impossible to be implemented in the

current solution are those requiring to jointly exploit

the two databases, and those requiring a recursive

traversal of the graph base. These can seamlessly be

implemented with semantic Web models.

5.2 Quantitative Evaluation: Analysis

of the Query Execution Times

For the evaluation of the execution times of the

queries implementing the use cases, we performed a

two-step benchmarking. First, we evaluated and com-

pared the proposed solution deployed in a local en-

vironment. This evaluation not only compare triple-

stores but also compare our proposed architectures.

Second we evaluated it when deployed in the Educle-

ver industrial environment. We compared the exe-

cution times with those of the current version of the

Educlever solution based on a relational database and

a graph database. For the deployment of the semantic

Web based solution, we compared the performances

of four triplestores. In the following, we describe the

experimental environment, protocol and results.

5.2.1 Experimental Environment and Protocol

Hardware: in the ﬁrst step of our benchmarking, we

used a MacBook Pro with processor 3,3 GHz In-

tel Core i7, 16 GB for RAM and 1 To for hard

disc. We used VirtualBox through Docker virtu-

alization. We used only one Docker container at

a time. In the second step, used a virtual Linux

server host on a remote machine. The remote

VMWare virtual machine has a processor AMD

Opteron 3.1 GHz, 6 GB of RAM and 85 GB for

hard disc.

DataSet: we used the exploitation data of Educlever

for the experiments. Table 2 summarizes the char-

acteristics of the datasets Corpus and Referential:

the number of triples and the number of instances

of Cocon Referential and of OPD in Corpus. Let

us note that the size of Corpus is much greater

than that of Referential, therefore the execution

times of queries on Corpus may be higher than

that of queries on Referential.

Queries: we implemented the Educlever use cases

by writing a base of fourteen SPARQL queries,

each one corresponding to one use case. They are

given in Table 3 in Appendix.

Triplestores: we tested four triplestores: (i) Alle-

grograph (alleg-cent), (ii) Corese (corese-cent),

(iii) GraphDB (graphdb) and (iv) Virtuoso (virt)

where we stored together the Referential and Cor-

pus datasets, as described in the ﬁrst proposed ar-

chitecture 4a. We also setup two SPARQL Feder-

ated Endpoints with Allegrograph (alleg-fed) and

Corese (corese-fed) storing Referential and Cor-

pus datasets separately as proposed in the sec-

ond proposed architecture 4b. The Allegrograph

SPARQL Federated Endpoint uses two SPARQL

Endpoints, each built with an Allegrograph repos-

itory. Similarly, the Corese SPARQL Federated

Endpoint uses a Corese server for each SPARQL

Endpoint. We compared the execution times of

the SPARQL queries implementing the Educlever

use cases with the execution times of the queries

or codes in the current Educlever Information Sys-

tem described in 3 (educ-v2).

WEBIST 2018 - 14th International Conference on Web Information Systems and Technologies

Table 2: Dataset statistics.

Dataset Number of triples Number of instances

Referential (Local vs Remote) 60 306 8 643 / 17

Corpus (Local vs Remote) 2 390 274 557 094 / 72 467

Protocol: we observed two indicators: (i) the

SPARQL query execution times and (ii) the

SPARQL query answers themselves. The ﬁrst one

measures the performance of the solution and the

second one checks its correctness. Since all the

conﬁgurations returned the same sets of answers,

in the following we focus on the evaluation of the

performance. For each tested triplestore, we exe-

cuted each query ten times and stored all the ex-

ecution times. For a deep analysis of the query

execution behaviours, we considered three indica-

tors: (i) the ﬁrst execution (1st Ex), (ii) the aver-

age execution time (Av) and (iii) the median (Med)

execution time of the next nine queries.

5.2.2 Results

Use Sases on Dataset Referential. Figure 5 shows

the query execution times of SPARQL queries on Ref-

erential for the four chosen triplestores deployed in a

local context. First, we can observe that query execu-

tion time of ﬁrst execution is greater than the average

time and the median time. This is due to the use of

cache memory for this execution.

For the speciﬁc case of Q

, its execution time is

very important (2s for graphdb) because it is the ﬁrst

query of the benchmark and cache is not efﬁcient

yet. The chart also shows that graphdb and virt got

the best query execution times, and that the execu-

tion times of alleg-fed are better than those of alleg-

cent, This is because only one dataset (Referential) is

queried with alleg-fed while both datasets are stored

together and queried with alleg-cent. The same can be

observed and explained when comparing the results

of corese-cent and corese-fed. All execution times are

below 200 ms. According to (Zhou et al., 2012), this

is an acceptable response time for a Web application.

Figure 6 shows the execution times of the same

queries on Referential, for the four triplestores this

time deployed in the industrial context of Educle-

ver; it also shows the execution times of the current

Educlever solution educ-v2. It conﬁrms the results

observed on the local deployment and it shows that

the execution time of educ-v2 is greater than corese-

cent and alleg-cent for use cases C

to C

. educ-v2

does not implement C

Use Cases on Dataset Corpus. Figure 7 shows the

query execution times of SPARQL queries on Cor-

100

150

200

250

300

350

400

450

500

1st Ex Av Med 1st Ex Av Med 1st Ex Av Med 1st Ex Av Med 1st Ex Av Med

Q1 Q2 Q3 Q4 Q5

Time (ms)

Referential Queries Execution Time

corese-cent alleg-cent graphdb virt corese-fed alleg-fed

Figure 5: Execution times of SPARQL queries on Referen-

tial with a local deployment.

200

400

600

800

1000

1200

1st Ex Av Med 1st Ex Av Med 1st Ex Ave Med 1st Ex Ave Med 1st Ex Ave Med

Q1 Q2 Q3 Q4 Q5

Time (ms)

Referential Queries Execution Time Online

corese-cent educ-v2 alleg-cent

Figure 6: Execution times of SPARQL queries on Referen-

tial with a remote deployment.

100

150

200

250

1st Ex Av Med 1st Ex Av Med 1st Ex Av Med

Q6 Q7 Q8

Time (ms)

Corpus Queries Execution Time

corese-cent

corese-fed

alleg-cent

graphdb

virt

alleg-fed

Figure 7: Execution times of SPARQL queries on Corpus

with a local deployment.

pus for the four chosen triplestores deployed in a lo-

cal context. Their observation conﬁrms our previous

comparative analysis on Referential: graphdb and virt

get the best query execution times. We also get con-

ﬁrmation that, in average, a federated architecture is

better for queries on a single dataset.

In comparison to Figure 5, we can note that the

execution times of queries on Corpus are much lower

Semantic Models in Web based Educational System Integration

100

150

200

250

300

350

1st Ex Av Med 1st Ex Av Med 1st Ex Av Med

Q6 Q7 Q8

Time (ms)

Corpus Queries Execution Time Online

corese-cent educ-v2 alleg-cent

Figure 8: Execution times of SPARQL queries on Corpus

with a remote deployment.

100

120

1st Ex Av Med 1st Ex Av Med 1st Ex Av Med

Q9 Q10 Q11

Time (ms)

Referential and Corpus Queries Execution Time

corese-cent corese-fed alleg-cent graphdb virt alleg-fed

Figure 9: Execution times of SPARQL queries on Referen-

tial and Corpus with a local deployment.

than those of queries on Referential whereas the size

of the Corpus dataset is much greater than that of

the Referential dataset (see Table 2). This can be

explained by the fact that the queries on Corpus

have simple star patterns while the queries on Ref-

erential have heterogeneous and more complex pat-

terns (Arias et al., 2011). All the execution times re-

main below 200 ms which is acceptable for a response

time of a Web application (Khan and Amjad, 2016).

Figure 8 shows the execution times of the same

queries on Corpus, for the four triplestores this time

deployed in the industrial context of Educlever; it also

shows the execution times of the current Educlever

solution educ-v2. It conﬁrms our previous results,

and corese-cent and alleg-cent outperform the current

Educlever system educ-V2. Use case C

does not have

an execution time for educ-v2 because it cannot be

implemented with only one query.

Use Cases on Both Datasets. Figures 9 and 10

show the execution times of the queries on both Ref-

erential and Corpus, for the four chosen triplestores

deployed respectively in a local and remote context.

The trends are the same and the execution times does

not exceed 200 ms for all the queries on all triplestores

79,222

76,778

80,222

1st Ex Av Med 1st Ex Av Med 1st Ex Av Med

Q9 Q10 Q11

Time (ms)

Referential & Corpus Queries Execution Time Online

corese-cent educ-v2 alleg-cent

Figure 10: Execution times of SPARQL queries on Refer-

ential and Corpus with a remote deployment.

128

256

512

1024

2048

4096

1st Ex Av Med 1st Ex Av Med 1st Ex Av Med

Q12 Q13 Q14

Time (log2(ms))

Properties Path Queries ExecutionTime

corese-cent corese-fed alleg-cent graphdb virt alleg-fed

Figure 11: Execution times of SPARQL queries with prop-

erty paths on Referential with a local deployment.

549

98,444

2391

454,667

410

415

648,778

660

2000

4000

6000

8000

10000

12000

14000

16000

18000

1st Ex Av Med 1st Ex Av Med 1st Ex Av Med

Q12 Q13 Q14

Time (ms)

Properties Path Queries Execution Time Online

corese-cent educ-v2 alleg-cent

Figure 12: Execution times of SPARQL queries with prop-

erty paths on Referential with a remote deployment.

in a local context. Figure 10 does not show the exe-

cution times for educ-v2 since it does not implement

these use cases with a single query.

Use Cases Implemented by Queries With Property

Paths. Property paths are a key feature for imple-

menting high value use cases for Educlever. Figures

11 and 12 show the execution times of such queries

on the four triplestores deployed respectively in a lo-

cal and a remote context. For readability, we use the

logarithmic scale to draw the chart in Figure 11. Fig-

WEBIST 2018 - 14th International Conference on Web Information Systems and Technologies

ure 12 conﬁrms that with corese-cent or alleg-cent in

the Educlever industrial context, the execution time

of queries with a few property paths in the graph pat-

tern, like it is the case for Q

, remains under 200

ms in average, which is acceptable for a Web appli-

cation. But, for more complex queries, like Q

and

, the execution time can reach up to to 4000 ms

(4s), which is not acceptable in the Educlever indus-

trial context. This is among our next challenges to

ﬁnd a convenient architecture to handle such queries,

with pre-processed results.

6 CONCLUSIONS

In this paper, we reported a knowledge modelling

experience in an industrial context to propose an

e-Education solution compliant with public educa-

tion speciﬁcations based on semantic Web models

and technologies. We brieﬂy presented the ontology

Eduprogression which describes a shared conceptual-

ization of knowledge pieces and skill in the educa-

tional context and we showed how we used it and

extended it to model the speciﬁc needs of a com-

pany (Educlever) for the E-Education solution they

develop. Then we described the proof of concept

we developed and deployed in the real industrial con-

text of Educlever. It relies on two ontologies, Ref-

erential populated by all the elements of knowledge

and skill (Cocons) available on the Educlever learning

platform, and Corpus populated by all the pedagog-

ical resources available on the Educlever platform.

We developed a base of SPARQL queries to imple-

ment the Educlever uses cases and we proposed two

software architectures based on Semantic Web tech-

nologies designed for an e-Education systems. We

upgraded the Educlever software architecture follow-

ing these propositions and implemented these archi-

tectures with four triplestores Corese, Allegrograph,

GraphDB and Virtuoso in order to benchmark them

and compare them to the existing solution on real data

and real queries.

We presented a complete evaluation of the quality

of service and response time in an industrial context

with a real-world tesbed showing that the Semantic

Web based solution meets the industrial requirements,

both in terms of functionalities and efﬁciency com-

pared to existing operational solutions. Moreover, by

relying on semantic Web we can reuse, extend and

align with existing vocabularies to increase interop-

erability. We showed this by implementing the in-

troduction of the standard ScolomFR with links to

the Educlever ontologies. With our propositions, it

is also now possible to share OPDs and integrate Co-

cons with other e-Education systems, provided that

they comply with the Eduprogression modeling.

In this context we also showed that an ontology-

oriented modelling opens up new opportunities. One

of the next challenges for us is the modeling of

learner proﬁles as an additional populated ontology

integrated with Referential and Corpus and the devel-

opment of SPARQL queries and rule-based reasoning

mechanisms for resource recommendation and adap-

tive learning. We also plan to link pedagogical re-

sources from several educational organizations in or-

der to build an integrated educational solution offer-

ing the learner a coherent learning path across a set of

educational systems, based on dynamically federated

endpoints.

REFERENCES

Al-Yahya, M., Al-Faries, A., and George, R. (2013).

Curonto: An ontological model for curriculum repre-

sentation. In Proceedings of the 18th ACM Conference

on Innovation and Technology in Computer Science

Education, ITiCSE ’13, pages 358–358, New York,

NY, USA. ACM.

Al-Yahya, M., George, R., and Al-Faries, A. (2015). On-

tologies in e-learning: Review of the literature. In-

ternational Journal of Software Engineering and its

Applications, 9:67–84.

Alsultanny, Y. A. (2006). e-learning system overview

based on semantic web. The Electronic Journal of e-

Learning, 4:111–118.

Arias, M., Fern

andez, J. D., Mart

ınez-Prieto, M. A., and

de la Fuente, P. (2011). An empirical study of real-

world SPARQL queries. CoRR, abs/1103.5043.

committee, I.-L.-L. (2002). IEEE Standard for Learn-

ing Object Metadata. IEEE Standards Association.

https://standards.ieee.org/ﬁndstds/standard/1484.12.1-

2002.html.

Gascue

na, J. M., Fern

andez-Caballero, A., and Gonz

alez, P.

(2006). Domain ontology for personalized e-learning

in educational systems. In Advanced Learning Tech-

nologies, IEEE International Conference on (ICALT),

pages 456–458.

Guangzuo, C., Fei, C., Hu, C., and Shufang, L. (2004). On-

toedu: a case study of ontology-based education grid

system for e-learning. In In: GCCCE2004 Interna-

tional conference, Hong Kong.

Gueffaz, M., Deslis, J., and Moissinac, J.-C. (2014). Cur-

riculum data enrichment with ontologies. In Proceed-

ings of the 4th International Conference on Web In-

telligence, Mining and Semantics (WIMS14), WIMS

’14, pages 441–446, New York, NY, USA. ACM.

Hyun-Sook, C. and Jung-Min, K. (2012). Ontology de-

sign for creating adaptive learning path in e-learning

environment. In Proceedings of International Mul-

tiConference of Engineers and Computer Scientists,

(IMECS 2012), IMECS’2012, pages 585–5886.

Semantic Models in Web based Educational System Integration

Hyun-Sook, C. and Jung-Min, K. (2014). Semantic model

of syllabus and learning ontology for intelligent learn-

ing system. In Computational Collective Intelligence.

Technologies and Applications: 6th International

Conference, ICCCI 2014, Seoul, Korea, September

24-26, 2014. Proceedings, volume 8733 of ICCCI

2014, pages 175–183, Heidelberg. Springer Interna-

tional Publishing.

Jaffro, L. (2007). Les objets de l’

education : quelle ontolo-

gie ? Revue de m

etaphysique et de morale, 4(56):429–

448.

Jean, S., Bellatreche, L., Fokou, G., Baron, M., and Khouri,

S. (2012). Ontodbench: Novel benchmarking sys-

tem for ontology-based databases. In On the Move

to Meaningful Internet Systems: OTM 2012, Con-

federated International Conferences: CoopIS, DOA-

SVI, and ODBASE, Rome, Italy, Proceedings, Part II,

pages 897–914.

Khan, R. and Amjad, M. (2016). Performance testing (load)

of web applications based on test case management.

Perspectives in Science, 8:355 – 357. Recent Trends

in Engineering and Material Sciences.

Minist

ere de l’

education nationale (2018). Code de

l’

education, Version consolid

ee au 1 janvier 2018.

https://www.legifrance.gouv.fr/afﬁchCode.do?cid-

Texte=LEGITEXT000006071191.

Mironov, V., Seethappan, N., Blond

e, W., Antezana, E.,

Lindi, B., and Kuiper, M. (2010). Benchmarking triple

stores with biological data. CoRR, abs/1012.1632.

http://arxiv.org/abs/1012.1632.

Rani, M., Srivastava, K. V., and Vyas, O. P. (2016). An on-

tological learning management system. Comput. Appl.

Eng. Educ., 24(5):706–722.

eseau Canop

e (2011). ScoLomFR : Outil de

description des ressources num

eriques de

l’enseignement scolaire. https://www.reseau-canope.

fr/scolomfr/accueil.html.

Rocha, O. R., Faron-Zucker, C., and Pelap, G. F. (2016). A

formalization of the french elementary school curric-

ula. In Knowledge Engineering and Knowledge Man-

agement - EKM and Drift-an-LOD, Bologna, Italy,

Revised Selected Papers, pages 82–94.

Wu, H., Fujiwara, T., Yamamoto, Y., Bolleman, J. T., and

Yamaguchi, A. (2014). Biobenchmark toyama 2012:

an evaluation of the performance of triple stores on

biological data. In J. Biomedical Semantics.

Zhou, Q., Ye, H., and Ding, Z. (2012). Performance anal-

ysis of web applications based on user navigation.

Physics Procedia, 24:1319 – 1328. International Con-

ference on Applied Physics and Industrial Engineer-

ing 2012.

APPENDIX

Table 3 presents the SPARQL queries implement-

ing the Educlever use cases. These are templates of

queries where Cocon and OPD must be replaced by

the URI of an instance of respectively class Cocon or

class OPD.

WEBIST 2018 - 14th International Conference on Web Information Systems and Technologies

Table 3: SPARQL queries implementing the Educlever use cases.

Label SPARQL Queries

Q1 SELECT ?prerequis WHERE {?prerequis referential:isPrerequisiteOf cocon .}

Q2 SELECT ?child WHERE {cocon referential:isParentOf ?child .}

Q3 SELECT ?next WHERE {cocon referential:isPrerequisiteOf ?next. }

SELECT ?prerequisite ?child ?childPrerequisite

WHERE {?prerequisite referential:isPrerequisiteOf cocon .

cocon referential:isParentOf ?child .

?childPrerequisite referential:isPrerequisiteOf ?child .}

SELECT ?simple ?helper ?helpPrerequisite

WHERE {cocon referential:isComplexificationOf ?simple .

?helper referential:isUnderstandingLeverageOf ?simple .

?helpPrerequisite referential:isPrerequisiteOf ?helper .}

Q6 SELECT ?opd WHERE {?opd corpus:isEvaluationOf cocon .}

SELECT ?status ?course ?learningDomain

WHERE {opd corpus:hasStatus ?status . opd corpus:hasCourse ?course .

opd corpus:hasLearningDomain ?learningDomain .}

SELECT ?opd ?status

WHERE {?opd corpus:isEvaluationOf cocon . ?opd corpus:isLearningOf cocon .

?opd corpus:hasStatus ?status .}

SELECT ?opd

WHERE {?opd corpus:isEvaluationOf cocon .?opd corpus:isEvaluationOf ?prerequiste .

?prerequiste referential:isPrerequisiteOf cocon .}

Q10

SELECT ?opd

WHERE {opd corpus:isEvaluationOf ?cocon . ?opd corpus:isEvaluationOf ?simple .

?cocon referential:isComplexificationOf ?simple . }

Q11

SELECT ?opd

WHERE {{?opd corpus:isTrainningOf cocon .}

UNION

{?cocon referential:isUnderstandingLeverageOf cocon .

?opd corpus:isTrainningOf ?cocon .}}

Q12 SELECT ?prerequis WHERE {?prerequis referential:isPrerequisiteOf+ cocon .}

Q13

SELECT ?source ?dest (count(?counter) as ?edgeposition

WHERE {c

refeduclever:isPrerequisiteOf* ?counter .

?counter referential:isPrerequisiteOf* ?source .

?source referential:isPrerequisiteOf ?dest .

?dest referential:isPrerequisiteOf* c

GROUP BY ?source ?dest . ORDER BY ?edgeposition .

Q14

SELECT ?sourceSim ?destSimp (count(?counter) as ?edgeposition

WHERE {c

refeduclever:isPrerequisiteOf* ?counter .

?counter referential:isPrerequisiteOf* ?source .

?source referential:isPrerequisiteOf ?dest .

?dest referential:isPrerequisiteOf* c

?sourceSim referential:isComplexificationOf* ?source .

?destSimp referential:isComplexificationOf* ?dest .

NOT EXISTS {?sourceSim referential:isComplexificationOf ?otherS .}

NOT EXISTS {?destSim referential:isComplexificationOf ?otherD .}}

GROUP BY ?sourceSim ?destSimp . ORDER BY ?edgeposition .

Semantic Models in Web based Educational System Integration