ITSEGO: An Ontology for Game-based Intelligent Tutoring Systems

Valentina Centola

and Francesco Orciuoli

Istituto Comprensivo Gonzaga, Via Caduti di Bruxelles, Eboli (SA), Italy

DISA-MIS, University of Salerno, Via Giovanni Paolo II, Fisciano (SA), Italy

Keywords:

Intelligent Tutoring Systems, Game-based Learning, Storytelling, Ontologies.

Abstract:

This work proposes the deﬁnition of a tool supporting the transition of children from kindergarten to primary

school and, as a side effect, the development of problem solving and digital competences. The tool has

been deﬁned, by means of an ontology-driven approach, as an Intelligent Tutoring System (ITS) integrated

to a structured game-based educational environment and provides beneﬁts for both teachers and children.

The deﬁnition of a novel ontology, namely ITSEGO, providing a model (generally applicable in different

learning contexts) to build Game-based ITS and, the design of a concrete Game-based ITS for supporting the

aforementioned transition are the main results of this work.

1 INTRODUCTION

In the last twenty years, numerous research activities

emphasize that the human learning process is not se-

quential but rather it exploits a number of conceptual

interconnections (Collins and Loftus, 1975) forming a

network. In this scenario, Information and Communi-

cation Technologies (ICT) provides a signiﬁcant con-

tribution in supporting (e.g., visualizing, organizing,

personalizing, adapting and sharing learning content),

improving (e.g., increasing engagement of students)

and making possible (e.g., satisfying special needs)

learning and teaching activities with respect to differ-

ent subject matters. One of the most important usage

of ICT for supporting learning and teaching activities

is represented, without any doubt, by Intelligent Tu-

toring Systems (ITS). ITS (Polson and Richardson,

2013) are software systems providing adaptive educa-

tional experiences for both students and teachers. ITS

have a great potential to improve school education, at

the moment, unexpressed yet. Adaptivity is supported

by ITS in several ways: i) providing students’ learn-

ing activities coherently with their current knowledge

and skills in order to foster meaningful learning; ii)

providing individualized feedback able to stimulate

next learning activities and avoid frustration, demo-

tivation and disengagement due to unsuccessful per-

formances; iii) providing hints helping students dur-

ing the execution of their learning tasks. Adaptation

capabilities of ITS and the evidence related to the im-

provement of learning experiences, when game-based

approaches are used (Jackson and McNamara, 2013),

lead to the idea to integrate these two elements in or-

der to synergistically exploit the advantages of both.

In several works like (McNamara et al., 2010) ITS

are enhanced with game-based features in order to in-

crease engagement of learners. Otherwise, this paper

proposes to inject ITS features into game-based envi-

ronments for children. Thus, we introduce a novel

ontology for modeling such features. Such ontol-

ogy can be populated in order to develop a concrete

ITS-empowered educational game. Although ontolo-

gies have been studied extensively with respect to ITS

(Mizoguchi et al., 1995), at the moment there are

not ontologies, based on Semantic Web technologies,

modeling all the aspects related to an ITS and able to

achieve different objectives like those reported in Sec-

tion 3.2. It is important to underline that the main part

of this work can be generally applicable for a wide

range of targets (also, for instance, to training) and

heterogeneous domains.

2 OVERALL APPROACH

The main idea of the paper is to provide a tool for sup-

porting children in the transition from kindergarten

to primary school. In our understanding, technolog-

ical environments that blend games, stories and in-

telligent (automatic) tutoring, represent effective so-

lutions for the above mentioned aim. In this section

we will motivate this idea and provide the necessary

background knowledge. In the next sections we will

238

Centola, V. and Orciuoli, F.

ITSEGO: An Ontology for Game-based Intelligent Tutoring Systems.

In Proceedings of the 8th International Conference on Computer Supported Education (CSEDU 2016) - Volume 1, pages 238-245

ISBN: 978-989-758-179-3

deﬁne a speciﬁc Game-based ITS, by which it is pos-

sible to achieve our goal, through two main phases.

The ﬁrst phase is the deﬁnition of ITSEGO (Intelli-

gent Tutoring System for Educational Games Ontol-

ogy), an ontology for Game-based ITS. This ontology

can be generally applicable to a wide range of learn-

ing contexts. The second phase is the adoption of IT-

SEGO (its instantiation and the provisioning of addi-

tional contents) to produce such speciﬁc Game-based

ITS. Thus, the second phase proposes an ontology-

driven process to build Game-based ITS.

2.1 Motivating ITS and Game-based

Learning Environment

This work is focused on children during their last year

of kindergarten. They face a critical period with re-

spect to their personal development and they are go-

ing to face the ﬁrst class of primary school. In the con-

text of such class they will develop also competences

needed for future learning. Despite of the importance

of the above mentioned issues, there are no concrete

recommendations or guidelines, coming from the Ital-

ian Ministry of Education, which support teachers in

evaluating the development of these competences. On

the other hand, since 2004, the required age to be ad-

mitted to attend the ﬁrst class of primary school has

been decreased. In fact, children born until 30 April

(of the school year) can begin such school. This kind

of procedure is executed, on demand of the parents,

after discussing with the teachers of their children.

On the basis of these considerations, this work

starts from the idea of ITaG (McNamara et al., 2010)

and proposes a concrete framework to adopt systems,

which integrate intelligent tutoring mechanisms and

game-based learning environments, for supporting the

above mentioned transition.

The main beneﬁts of using such systems are: i)

ITS allow children to develop basic competences con-

cerning their personal growth in a personalized and

incremental way, meanwhile, the game-based compo-

nent of the whole system allows to increase engage-

ment and motivation, makes smooth the transition be-

tween the two considered types of school and, lastly,

to implement the learning-by-doing approach; ii) ITS

could trace all the child’s actions and provide struc-

tured data to be analyzed by teachers in order to ex-

ecute better evaluations; iii) ITS and (digital) game-

based learning environments help children to improve

digital competences, which are included in the key

competences recommended by the European Parlia-

ment in 2006

http://eur-lex.europa.eu/legal-content/EN/TXT/?uri=

URISERV:c11090

2.2 Fundamentals and Background

Knowledge

This section provides the reader with key concepts of

ITS and Game-based Learning Environments. These

two concepts will be the foundation for this work.

2.2.1 Intelligent Tutoring Systems

An ITS (Polson and Richardson, 2013) is a soft-

ware agent whose behavior can be, informally, de-

scribed by an outer (external) loop and an inner loop

(du Boulay, 2006). The external loop proposes to the

learners a sequence of tasks (problem solving) of dif-

ferent difﬁculties. The default behavior is that the

next task has a difﬁculty greater than the previous one

(mastery learning). However, if the learners’ results

are negative the ITS can propose, for instance, a task

with a lower difﬁculty or alternative learning content.

Moreover, there is an inner loop for each task where

a sequence of steps have to be executed by the learn-

ers in order to achieve the task objectives and provide

a solution for the associated problem. ITS can pro-

vide adaptive feedback (positive, negative, etc.) in re-

sponse to the learners’ answers for the current steps

or hints to anticipate the next step of the same task.

The set of all possible actions that can be carried out

by the ITS is called tutoring actions. The selection of

an effective tutoring action can be accomplished on

the basis of pedagogical strategies, learners’ proﬁles,

context, domain and so on.

2.2.2 Game-based Learning Environments

A Game-based Learning Environment is mainly char-

acterized by the capability to maintain constant play-

ers/learners’ attention and to keep high levels of en-

gagement for them. One of the main challenges of

the educational games is to provide a game design en-

abling players/learners to face tasks and steps accord-

ing to their real ability (Kiili, 2005). The main idea in

Game-based Learning is to always guarantee a game

level a bit greater than the ability she demonstrated by

considering a zone of proximal development (Vygot-

sky et al., 2012). This aspect can be sustained by the

behaviour of an ITS by controlling and adapting both

internal and external loops.

3 DEFINITION OF ITSEGO

The core part of this work is the deﬁnition of an

ontology that can be used as a model to construct

ITS for Game-based Environments. Such ontology is

ITSEGO: An Ontology for Game-based Intelligent Tutoring Systems

239

called ITSEGO. ITSEGO can be adopted, extended,

instantiated and completed (with additional content)

to build speciﬁc games empowered by intelligent tu-

toring features. In order to deﬁne ITSEGO we have

selected the Web Ontology Language (OWL), pro-

vided by W3C

, that is one of the main components

of the so called Semantic Web Stack

. OWL is based

on Description Logic (Kr

otzsch et al., 2014) and pro-

vides a set of suitable capabilities like providing high

levels of interoperability and integration and a formal

inference mechanism. This is one of the main beneﬁts

of Semantic Web ontologies, they make easily to inte-

grate with existing results like, for instance, the ontol-

ogy provided by the OMNIBUS ontology (Bourdeau

et al., 2007).

3.1 Building the Ontology

In this section we will show the most relevant generic

parts of the ITSEGO that are identiﬁed with respect

to the important aspects of the Game-based ITS we

are modeling: interactions, tasks and steps, context,

storytelling, game, tutoring actions and tutoring rules.

The main design principles considered for the engi-

neering of ITSEGO are the following ones:

• Maintaining simple and general the design of

classes, properties and axioms with no specializa-

tion with respect to application domain, learning

context and content to propose. The idea is to

provide essential building blocks (with basic con-

structs and key abstractions) for developing dif-

ferent ITS in several and heterogeneous domains,

although the driver for this work comes from a

speciﬁc target with speciﬁc motivations.

• Including, in the ontology deﬁnition, all the nec-

essary to construct a speciﬁc (Game-based) ITS

to be deployed at a testing environment (e.g.

Prot

) as well as to support real-world ITS.

• Designing and including features for building

Game-based ITSs deployable in blended environ-

ments by considering game and storytelling ele-

ments.

3.1.1 Interactions

An interaction with the Game-based ITS is divided

into sessions. A single play is an individual of

class ITSEGO:Play and aggregates one or more ses-

sions. The concept of session is modelled by us-

ing the class ITSEGO:Session. Each session main-

http://www.w3.org/TR/owl2-overview/

http://www.w3.org/standards/semanticweb/

http://protege.stanford.edu/

Figure 1: The ontology part dealing with interactions.

tains a set of interactions among learner (student)

and system. Interactions are individuals of the class

ITSEGO:Interaction. The learner is represented by

means of the foaf:Person class included in the well

known FOAF ontology

(see Fig. 1). An interaction

refers to a task (individual of class ITSEGO:Task)

and, in particular, to a step (individual of class

ITSEGO:Step). Tasks and steps are mainly related by

the ITSEGO:isComposedBy property. The property

ITSEGO:firstStep indicates the ﬁrst step in a task

and the property ITSEGO:next provides an ordered

relation among the steps within a task.

Tasks have difﬁculties (ITSEGO:difficulty data

property) and can be related each others by means of

the ITSEGO:similarTo property. This is useful to

propose alternative tasks. Steps are related each oth-

ers by means of the ITSEGO:equivalentTo property

that is useful to accept multiple solutions for a spe-

ciﬁc step. Detailed information on tasks and steps will

be provided in the next sections. An interaction also

refers to a response (individual of ITSEGO:Response)

that represents data related to the answer that the

learner provides when a speciﬁc step in a speciﬁc task

is proposed to her. An interaction occurs in a given

context (i.e., a set of contextual information like, for

instance, the learner’ proﬁle). Moreover, the inter-

action includes a tutoring action (individual of class

ITSEGO:TutoringAction) that is the action the tutor

(human or automatic) has to perform in order to adapt

the experience by taking into account the speciﬁc

learner’s answer. When a learner’s answer is evalu-

ated, it is needed to check if it is correct. Some im-

plementations could foresee also the chance to have,

for instance, a fuzzy correctness value. In order to

support different kinds of metrics, ITSEGO provides

the class ITSEGO:EvaluationResult that is linked

to and provides information related to a given interac-

tion. This class can be specialized to support boolean

evaluation results, fuzzy evaluation results and addi-

tional properties that can be possibly used in the ped-

agogical rules.

http://www.foaf-project.org

CSEDU 2016 - 8th International Conference on Computer Supported Education

240

3.1.2 Tasks and Steps

The problems proposed to the learners are struc-

tured in tasks and steps. Classes ITSEGO:Task and

ITSEGO:Step must also contain all the information

needed to allow the execution of the Game-based ITS,

i.e., how the problem is presented (communicated) to

the learners. The base ontology we propose brings

with it a set of properties useful to link tasks and steps

with concrete resources (by using URIs). In this way

it is possible to deploy the system at different and het-

erogeneous learning environments.

Furthermore, the ITSEGO:Interaction class

provides a link to the speciﬁc environment (platform)

used to deploy tasks and steps, receive user inputs and

provide outputs.

Now, let us focus on two important aspects: topic

and content of a task/step. Tasks and steps need to

be associated to the content that must be proposed to

learners by using generic URIs. In addition, ITSEGO

provides a conceptual layer by means of the prop-

erty ITSEGO:topic connecting ITSEGO:Task (and

ITSEGO:Step) to skos:Concept that is the main con-

struct of SKOS

. SKOS is a Semantic Web ontology

allowing the deﬁnition of thesauri, taxonomies, con-

cept maps and so on. The conceptual layer enables

reasoning on topics and allows, for instance, to clas-

sify tasks with respect to their topics.

Lastly, tasks are not pre-ordered. The sequence of

tasks, proposed to the learner, can be obtained by ap-

plying speciﬁc pedagogical rules. One of the plausi-

ble solutions is to adopt the mastery learning strategy,

i.e., proposing tasks in order of increasing difﬁculties.

3.1.3 Context

The class ITSEGO:Context includes: student model

information and contextual information like, for in-

stance, environment or situation characteristics dur-

ing learners’ interactions with the system. The stu-

dent model is composed of three parts. The ﬁrst one

includes student’s information (e.g., name, age), con-

textual information (e.g., family context, school con-

text), competency information (e.g., competencies al-

ready acquired), personal traits information. The sec-

ond one is characterized by information related to a

speciﬁc play like rewards (earned during the play),

competencies (acquired during the play), score (the

total score for the play) and performances (produced

during the play). The performances of a learner are

linked to speciﬁc game levels. Each game level has

a status (completed, not-completed, not-started) and

includes scores, rewards (that can be also badges) and

http://www.w3.org/2004/02/skos/

competencies (acquired) obtained by the player when

she faces this level. A game level is related to one or

more tasks (possibly belonging to the same difﬁculty)

that have to be successfully executed to complete the

level. The third one is characterized by information

related to affective and emotional states of the play-

er/learner. This information is dynamic and can be

“perceived” by processing raw data that comes from

additional sensors.

All the information included in the context can be

used by the pedagogical rules to adapt the experiences

and produce suitable and effective tutoring actions.

Moreover, some other data have to be gathered in ad-

dition to the previously indicated ones. For instance,

pedagogical rules can be based on number of errors

and number of consecutive errors for a speciﬁc step

and so on.

3.1.4 Tutoring Actions

Tutoring actions are the actions provided by tutors in

response to learners’ interaction with the content of a

speciﬁc step in a given task. Tutoring actions are se-

lected by considering several and heterogeneous as-

pects: context, pedagogical strategy, student’s emo-

tional state, student’s prior knowledge and so on. Tu-

toring actions can be classiﬁed in feedback, hints and

adaptations. ITSEGO provides further specializations

of the above mentioned classes. At each interaction,

only a subset of all plausible tutoring actions can be

provided.

3.1.5 Tutoring Rules

Tutoring or pedagogical rules implement the peda-

gogical strategy used to adapt the environment and

generate feedback/hint on the basis of the behavior

of the learner during her interaction with a step (in a

task) in a speciﬁc context. There are numerous ways

to implement such rules for an ITS. In this work, we

propose the development of such rules as class re-

strictions within ITSEGO. In this way it is possible

to generate the correct tutoring action (or set of tutor-

ing actions) by using standard OWL-DL reasoners.

The part of the ontology that is useful to model

and execute pedagogical rules is provided in Fig. 3.

The class ITSEGO:PedagogicalRule contains all the

admissible rules of the ITS (individuals). Each rule

is composed by a condition and an action (or more

actions) that must be executed when the condition

value is true. Conditions are individuals belong-

ing to ITSEGO:LearningCondition and are com-

posed by game state (task and step executed by the

learner), context (up to date information related to the

ITSEGO: An Ontology for Game-based Intelligent Tutoring Systems

241

learner: scores, attempts, emotions, proﬁle, etc.), re-

sponse (correctness and other information about the

learner’s response with respect to the considered task

and step) and story environment (the conﬁguration of

the story environment underlying the game). Such

elements are considered in the condition in order

to make decisions about the tutoring action to de-

liver. The idea is that if the current learning con-

dition is equal (for each element) to the learning

condition attached to a rule, such rule can be exe-

cuted to obtain the right tutoring action. Furthermore,

ITSEGO:ExpectedResponse is linked to the class

ITSEGO:Step and indicates response templates for a

step. Each response template (for a step) contains also

a score. In this way it is possible to evaluate the cur-

rent response of the learner by ﬁnding the best match

among the current response and the existing response

templates. Tutoring actions are individuals of class

ITSEGO:TutoringAction and are connected to the

rules by means of the property ITSEGO:isImplied

(and its inverse ITSEGO:hasAction).

The selection of the correct tutoring action is ob-

tained by looking for the rule generating such action.

This selection is accomplished by considering two in-

ference operations. The ﬁrst one is used to look for

the right pedagogical rule in the knowledge base by

comparing the learning conditions associated to each

existing rule and the current learning condition:

APR ≡ (PedagogicalRule uC

)

≡ (∃hasCondition.(C

u C

))

(1)

≡ (∃template.C

)

≡ (∃inverseMatchesWith.C

)

≡ (∃responseInCondition.CLC)

(2)

≡ (∃context.C

)

≡ (∃contextInCondition.CLC)

(3)

≡ (∃gameState.C

)

≡ (∃stateInCondition.CLC)

(4)

The second one is used to extract the suitable tu-

toring action from the selected pedagogical rule:

TutoringAction u (∃isImplied.APR) (5)

The above described inference operations are

graphically explained in Fig. 2 that shows how the

current learning condition matches with the condi-

tion attached to a speciﬁc rule. In particular the in-

dividual currentLearningCondition (belonging to

class ITSEGO:CurrentLearningCondition, in brief

ITSEGO:CLC) matches with learnincCondition01

Figure 2: Sample individuals related to the operation of

looking for the right rule to be applied.

Figure 3: The ontology part dealing with tutoring (pedagog-

ical) rules.

because they have all equal elements and thus the

right rule is rule01.

For the sake of clarity, and in order to provide

simple statements, we have proposed only individuals

like context01 that includes several contextual infor-

mation (the emotional state of the learner, the affec-

tive state of the learner and the tentative number for

this step in this task, etc.). This does not imply a lack

of generality because it is possible to simply provide

more details by fragmenting individuals and consider-

ing, for instance, ﬁner contextual informations. Take

care that it is possible to implement pedagogical rules

by means of other approaches enabling the deﬁnition

of more complex rules. One of such approaches could

be SWRL

3.1.6 Storytelling

In order to model storytelling aspects within the

ITSEGO, our approach is not providing “build-

ing blocks” deﬁned from-scratch but integrating an

existing language/ontology. In literature, several

Ontologies and markup languages providing mod-

els to build stories are recognizable. Taking care

our aim to maintain ITSEGO as light as possi-

ble we have selected a generic storytelling ontol-

ogy model (Nakasone and Ishizuka, 2006) based

on the Rhetorical Structure Theory (RST). We ar-

bitrary assign the namespace RST to this ontology.

In ITSEGO, a play is linked to the elements of

http://www.w3.org/Submission/SWRL/

CSEDU 2016 - 8th International Conference on Computer Supported Education

242

the story in which the play takes place. In par-

ticular, ITSEGO:Play is linked to ITSEGO:Story

by means of the ITSEGO:bgStory property. A

story is composed of story elements (individuals of

ITSEGO:StoryElement). Story elements can be

characters (individuals of ITSEGO:Character), set-

tings (individuals of ITSEGO:Setting) and scenes

(individuals of ITSEGO:Scene that is equivalent to

RST:Scene). A scene is deﬁned as a set of acts (in-

dividuals of class RST:Act) that are hierarchically

composed of nucleuses and satellite entities linked

by relations. Acts establish the minimum level of

story organization. Such entities could contain an

event (individual of class RST:Event) or another act.

Agents are actors taking part in a scene by execut-

ing or being part of one or more events. Agents are

individuals of class RST:Agent (that is equivalent to

the class ITSEGO:Character). A role is a part that

an agent plays during a scene. Roles are modelled

by means of the class RST:Role. There are several

roles: questioning role, informing role, contrasting

role and evaluating role. ITSEGO adds to RST the

class ITSEGO:Setting that provides the graphical el-

ements of a scene.

3.1.7 Game

In order to embed an ITS in a game-based environ-

ment it is needed to enrich the ITSEGO with spe-

ciﬁc elements. In particular, the class ITSEGO:Score,

that is linked to ITSEGO:ExpectedResponse, is used

to model the score earned by successfully execut-

ing the related step. The ITSEGO:Score class can

be linked also to other classes. For instance, if we

link individuals of ITSEGO:Interaction with indi-

viduals of ITSEGO:Score it is possible to sign the

score of each interaction. In addition, ITSEGO:Score

can also be linked to several parts of the context

(as reported in section 3.1.3) in order to assign

scores to play and levels for each player/learner).

Game levels can be modelled by means of the class

ITSEGO:GameLevel, whose individuals are linked

to the individuals belonging to ITSEGO:Task. The

class ITSEGO:GameLevel maintains all the informa-

tion needed to manage the game like, for instance, re-

wards (individuals of class ITSEGO:Reward) and max

time to play (individuals of class ITSEGO:Time).

3.2 Use Cases for ITSEGO

In this section, we brieﬂy describe three main use

cases in which ITSEGO can be exploited.

UC#1 Sharing Terminology. ITSEGO can be

used to share knowledge about terminology related to

ITS among the community of researchers, system de-

signers, software developers, content developers and

educators. The adoption of the Semantic Web Stack

fosters interoperability and integration.

UC#2 Building ITS. ITSEGO can be used to

build ITS. ITSEGO can be used as-is for rapid pro-

totyping by populating it and using the embedded

pedagogical rules. Populating the ontology means to

construct the domain model (by deﬁning the concept

map as indicated in Section 3.1.2 and linking those

concepts to the new inserted tasks and steps), the ex-

pert model (by providing the expected response for

each step), the tutor (pedagogical) model (by deﬁn-

ing new pedagogical rules as class restrictions or

logical rules), the student model (by ﬁlling learn-

ers’ proﬁles) and, lastly, the communication (user

interface) model (by conﬁguring stories, characters,

events, etc.). Moreover, ITSEGO can be also used

within a real-world system developed by means of a

general-purpose programming language like Java and

some GUI framework. Lastly, ITSEGO can be ex-

tended (or specialized) by subclassing existing classes

and adding new axioms and rules.

UC#3 Tracing and Learning Pedagogical

Rules. ITSEGO can be used as a knowledge base

for existing ITS. The ITSEGO:Interaction can be

populated every time a learner interacts with a step.

This allows to trace all actions occurring in the system

to perform statistics, analytics and/or machine learn-

ing. In particular, machine learning algorithms can be

used to learn pedagogical rules by using the (human)

tutoring actions applied against a speciﬁc learner’s in-

teraction.

4 INSTANTIATING A

GAME-BASED ITS

In this section we will show how the proposed on-

tology has been instantiated in order to build a tool

supporting children to their transition from kinder-

garten to primary school. In particular, an Educational

Game (in brief Edu Game) has been implemented in

the form of an Android App. Lastly, a validation and

evaluation framework is described in the second part

of this section.

4.1 The Prototype System

This part of the work aims at introducing the most im-

portant issues related to the development of the proto-

type App that has been experimented.

ITSEGO: An Ontology for Game-based Intelligent Tutoring Systems

243

4.1.1 Architectural and Technological Issues

The system prototype has been developed by means

of LibGDX

and Java for realizing the front-end and

deploying it on Android tablets (LibGDX is multi-

platform, so it allows to deploy the software on An-

droid, iOS, Desktop, Web, etc.). In this ﬁrst prototype

the access to ITSEGO is realized by means HTTP

connections to a set of Web Services (implemented in

Java and hosted by JBoss) to query and perform infer-

ences over the knowledge base. The main idea of the

system behaviour is that the App proposes a screen

to the child in combination with the audio explana-

tion (it is probably that 5-6 years old children can-

not read text) of the associated task/step. At the end

of the audio description, the child is free to interact

with the screen. This interaction triggers an inference

over ITSEGO that, according to the pedagogical rules

modelled in the ontology, is able to ﬁnd a suitable tu-

toring action to return it to the App that executes such

action in order to provide its result (e.g. feedback)

to the child. It it important to underline that the App

implements exactly the ITSEGO model and its behav-

ior mirrors the one provided by the pedagogical rules

modelled by using the ontology constructs (as shown

in section 3.1.5). Pedagogical rules, added to the IT-

SEGO, are driven by the speciﬁc context (5-6 years

old children) in which they have to be used and for

the speciﬁc learning objectives (supporting the transi-

tion from the kindergarten to the primary school) we

have to accomplish.

4.1.2 Edu Game Content

In order to design the Edu Game content, we have se-

lected a set of characters, one for each task, who guide

the child along the steps. Each task is associated to

a story providing a problem to the child who has to

solve it by executing a sequence of steps. In partic-

ular, the ﬁrst task is associated to a speciﬁc problem

of tidying up a room. The main character for this task

is called Rughetto and he asks to the child/player to

give him help in order to allow him to go out to play

football with his friends (see Fig. 4(a)). Each step as-

sociated to this task is related to the main problem.

For instance, in the ﬁrst step the child/player has to

put toys into the basket games and the colored pencils

into the case (i.e. a problem related to classify things).

In the second step the child/player has to sort some

books from the smallest to the largest (i.e. a problem

related to the concepts of smaller and larger things).

The third step is focused on the identiﬁcation of the

right shapes and the fourth (last) step is focused on

https://libgdx.badlogicgames.com

Figure 4: Screens: (a) introduction and (b) second step.

ﬁnding correct numbers on the phone keypad in order

to call Rughetto’s friends to play football.

Feedback actions are driven by pedagogical rules

and are based on audio descriptions and colored

screens. Hint actions are enriched by audio helps

and/or images recalling suitable solutions for speciﬁc

steps. Colors are selected with respect to the emotions

that should be encouraged.

4.2 Validation and Evaluation

In order to evaluate and validate the results of this

work we have designed the activities of Table 1.

Table 1: Framework for Validation and Evaluation.

Validation Evaluation

ITSEGO

OOPS!

Methodology

Case Study

Ontology-driven

Approach

Building

of Edu Game

”Rughetto”

Asking to

Developers

Edu Game

”Rughetto”

Testing Experiments

ITSEGO. The ontology ITSEGO has been vali-

dated by means of the OOPS! tool (Poveda-Villal

et al., 2012) for detecting potential pitfalls that could

lead to modelling errors. In particular, ITSEGO has

passed the tests with respect to Functional Dimen-

sion, Consistency and Conciseness. The other tests

for Structural Dimension and Completeness advise

that there is only one important pitfall. In particular,

pitfall P11 (missing domain or range in properties)

is signaled for some properties deﬁned as inverse of

other properties. Thus, this pitfall is ﬁxed when rea-

soning is executed on ITSEGO and missing domains

and ranges are inferred for such properties. Lastly, IT-

SEGO has been evaluated by populating it with real

CSEDU 2016 - 8th International Conference on Computer Supported Education

244

Figure 5: Results of developers’ questionnaires.

data for the case study that has been designed, imple-

mented and described in Section 4. The Edu Game,

exploiting ITSEGO for generating tutor actions, has

been tested and its behavior produces correct results.

Ontology-driven Approach. The approach we

proposed has been validated by building the Edu

Game described in Section 4. This approach has been

evaluated by proposing a questionnaire to the three

developers of the Edu Game and analysing their an-

swers. Such questionnaire includes 9 questions re-

garding three aspects of the approach: i) usefulness

of the shared terminology formalized by ITSEGO, ii)

easiness of implementing pedagogical rules, and iii)

effort of building contents. Results are reported in

Fig. 5 that shows the usefulness and easiness of the

interaction with ITSEGO. Difﬁculties and high efforts

to build Game-based ITS contents are emphasized.

Edu Game. The Edu Game has been validated by

considering both unit testing and system testing. An

experimentation activity has been planned to evaluate

the effectiveness of the learning process realized by

means of the provided Edu Game. More in details, the

experiment has been designed to be performed in the

context of the ﬁrst class of a primary school. Three

groups will be built. The ﬁrst two groups are control

groups. The third group is the experimental group.

In the ﬁrst phase a pre-test will be dispensed to the

children of the three groups. In the second phase, the

children of the third group will interact with the game

in order to learn the subjects introduced in Section 4.

The ﬁrst two groups will deal with the same subjects

by using traditional learning content. Lastly, in the

third phase, all the groups will face the same post-test.

5 CONCLUSIONS

This paper proposes the idea of integrating ITS fea-

tures into Edu Games. Main results are: deﬁnition of

an ITS ontology, deﬁnition of an ontology-driven ap-

proach to build Game-based ITS and development of

an Android App implementing an Edu Game that ex-

ploits the reasoning capabilities of ITSEGO. Lastly, a

framework for validating and evaluating the multiple

results of this work has been proposed. In the future,

the deploy of ITSEGO on Raspberry will be investi-

gated. At the moment we are realizing the experiment

described in the last part of Section 4.

REFERENCES

Bourdeau, J., Mizoguchi, R., Hayashi, Y., Psyche, V., and

Nkambou, R. (2007). When the domain of the ontol-

ogy is education. Proc. of I2LOR07.

Collins, A. M. and Loftus, E. F. (1975). A spreading-

activation theory of semantic processing. Psychologi-

cal review, 82(6):407.

du Boulay, B. (2006). Commentary on kurt vanlehn’s” the

behavior of tutoring systems”. IJ Artiﬁcial Intelli-

gence in Education, 16(3):267–270.

Jackson, G. T. and McNamara, D. S. (2013). Motiva-

tion and performance in a game-based intelligent tu-

toring system. Journal of Educational Psychology,

105(4):1036.

Kiili, K. (2005). Digital game-based learning: Towards an

experiential gaming model. The Internet and higher

education, 8(1):13–24.

otzsch, M., Simancik, F., and Horrocks, I. (2014). De-

scription logics. IEEE Intelligent Systems, 29(1):12–

19.

McNamara, D. S., Jackson, G. T., and Graesser, A. (2010).

Intelligent tutoring and games (itag). Gaming for

classroom-based learning: Digital role-playing as a

motivator of study, pages 44–65.

Mizoguchi, R., Vanwelkenhuysen, J., and Ikeda, M. (1995).

Task ontology for reuse of problem solving knowl-

edge. Towards Very Large Knowledge Bases: Knowl-

edge Building & Knowledge Sharing, pages 46–59.

Nakasone, A. and Ishizuka, M. (2006). Storytelling

ontology model using rst. In Proceedings of the

IEEE/WIC/ACM international conference on Intelli-

gent Agent Technology, pages 163–169.

Polson, M. C. and Richardson, J. J. (2013). Foundations of

intelligent tutoring systems. Psychology Press.

Poveda-Villal

on, M., Su

arez-Figueroa, M. C., and G

omez-

erez, A. (2012). Validating ontologies with oops!

In Knowledge Engineering and Knowledge Manage-

ment, pages 267–281. Springer.

Vygotsky, L., Hanfmann, E., and Vakar, G. (2012). Thought

and language. MIT press.

ITSEGO: An Ontology for Game-based Intelligent Tutoring Systems

245