Decision Module: Adapting Learning Paths in Serious Games based on CbKST and Teachers' Decisions

DECISION MODULEAdapting Learning Paths in Serious Games

based on CbKST and teachers’ decisions

Javier Melero & Naïma El-KechaïNovember 20, 2015

OVERVIEW

2 | 16

Competence-based Knowledge Space Theory (CbKST)Requirements from Play Serious Project

for adapting Serious Games (SGs)

To propose an alternative approach to create competence structures

To propose recommendation strategies based on teachers’ requirements

To propose a solution to create well-connected competence structures

Implementation of a Decision Module and an Authoring Tool based on CbKST and adapted to teachers’ requirements

“Complete” Evaluation of the approaches based on CbKST and the recommendation strategies

Preliminary Evaluation

CONTEXT

CHALLENGES

PROPOSED SOLUTION

EVALUATION

CONTEXT

3 | 16

Competence-based Knowledge Space Theory (CbKST)Requirements from Play Serious Project

for adapting Serious Games (SGs)

CONTEXT

CbKST (Heller et al., 2006):

• Allows adaptation in SGs by modelling a knowledge domain as a formal structure of admissible and meaningful competence states on the basis of prerequisite relationsamong the competences

• Activities of SGs are related to the competences worked on

• Three main concepts: prerequisite relations, competence states (CSs) and the competence structure

e

cd

b

a{ }

{a} {d}

{a, b} {a, d}

{a, b, d}{a, b, c}

{a, b, c, d}

{a, b, c, d, e}

Play Serious Project:

• To propose adaptive tools for modifying learning paths of SGs

• Teachers’ requirements:

S1. To allow learners progressing autonomously and gradually to achieve all competences of a knowledge domain

S2. To allow teachers specifying a subset of competences to work on, as well as the degree of achievement as prerequisites to let their learners move forward in the following stages

S3. To offer teachers the possibility to choose specific competences to let their learners to progress to a higher advanced competence level

ISSUE: Modelling knowledge domains can be time-consuming and a tough task for experts, mainly because of the difficulties in defining dependencies between competences of a given domain (Falmagne et al., 2006)

CHALLENGES (1/2)

4 | 16




CHALLENGES

Three strategies proposed:

S1. “Advancing”: Considers the learner’s proficiency level and proposes activities that work on the maximum number of competences

S2. “Reinforcing”: Allows the learner reinforcing a set of competences that have not met a minimum threshold

S3. “Deepening”: Allows the learner becoming expert of certain competences that has mastered within a knowledge domain

Approach based on the Q-Matrix:

• Complementary and simplified approach to build the competence structure that considers only the competences worked on in the SG activities

• The Q-matrix (Tatsuoka, 1983) that contains the mapping between activities and their worked competences

Activities of a SG

Worked competencesCompetence

states[a]

[e] [f][b] [c]

Level 1 X [c]

Level 2 X X [b, c]

Level 3 X X X [b, c, e]



Level 6 X X X X [b, c, d, e]





Table. Example of the proposed Q-Matrix for Blockly: Maze

ISSUE: Competence structures could be not well-connected

Figure. Example of the competence structureof Cristaux d’Éhère based on the Q-Matrix

CHALLENGES (2/2)

5 | 16




CHALLENGES

Preliminary Evaluation:

• Preliminary evaluation of the strategies in the SG Cristaux d’Ehere:

• From an algorithmic point of view, The results obtained by the strategies were validated compared with the expected results

• A co-designer involved in the implementation of the SG also validated the obtained results

Lessons learnt from building the competence structures of 3 SGs:

• The competence structure obtained from the domain model can be used as an approach to validate the design of SG activities

• The competence structures derived by Q-Matrices are simpler than those generated by domain models

• Competence structures derived by domain models are more complete since they cover a larger scope than those derived by Q-Matrices

• All CSs of a competence structure built by the Q-Matrix contain activities since each CS is built by the set of competences worked on by the SG activities

• Competence structures built by domain models may contain a large number of CSs that are not linked to activities

• The competences have to be described at a suitable level of granularity

• There can be gaps between CSs in both approaches

CONTRIBUTION AT (submitted):• 23rd Conference on User Modelling, Adaptation and Personalization [UMAP 2015]

CONTRIBUTIONS AT (submitted):• 7ème Conférence sur les Environnements Informatiques pour l’Apprentissage

Humain [EIAH 2015]• 7th International Conference on Computer Supported Education [CSEDU 2015]

Current competence

state

Subset of competences(if applicable)

Activities done

System confidence

Suggested activity

Advancing Deeping Reinforcing

Initial state =ø - None - La porte gelée None None

Initial state =ø -La portegelée

-Les thermomètres

None None

{h, i}“h” (Reinforcing)“i” (Deepening)

None0.3 : “h”0.7 : “i”

La machine à cube

RemédiationLa porte gelée

{h, i}“h” (Reinforcing)“i” (Deepening)

La machine à cube

0.3 : “h”0.7 : “i”

Les deux fluides RemédiationLa portegelée

PROPOSED SOLUTION (1/10)

6 | 16


PROPOSED SOLUTION

Decision Module:

Adapted from the model proposed by Kopeinik et al. (2012). Main differences:• Recreational activities• Teachers’ decisions

Main Components of the DecisionModule:• The Domain Model of a SG• The recreational competences• The list of activities (or levels)• The Learner Model• The recommendation strategies

that the teacher can choose

Authoring Tool:

Teachers can specified the following information:• The CbKST approach to build the

competence structure • The recommendation strategies • Sorting activities by difficulty • The subset of competences to be trained

and the threshold value to suggest potential next activities

Domain ModelPedagogical competences and

prerequisite relations

Learner ModelCompetence performance values

(confidence)

Pedag. Competence 1[0,1]Pedag. Competence 2[0,1]Pedag. Competence 3[0,1]Pedag. Competence 4[0,1]

List of SG activities

Activity 1 (Pedag. competences)Activity 2 (Pedag. competences)Activity 3 (Recr. competences)Activity 4 (Pedag. and recr.

competences)

Adaptation Strategies“ADVANCING” – “REINFORCING” – “DEEPENING”

NEXT ACTIVITY

Recreational Competences

ThresholdSubset of competences Sort activities by difficulty

Teacher’s decisions


7 | 16


PROPOSED SOLUTION

Implementation of the “Advancing” Strategy:

Activity done

NEXT ACTIVITY SUGGESTED

1

{ }

{a} {d}

{a, b} {a, d}

{a, b, d}{a, b, c}

{a, b, c, d}

{a, b, c, d, e}

Current

competence state

{ }

{a} {d}

{a, b} {a, d}

{a, b, d}{a, b, c}

{a, b, c, d}

{a, b, c, d, e}

Possible next competence

states

{ }

{a} {d}

{a, b} {a, d}

{a, b, d}{a, b, c}

{a, b, c, d}

{a, b, c, d, e}

Selected

competence state

Activity

1

Activity

N

2 3


8 | 16


PROPOSED SOLUTION

Implementation of the “Reinforcing” and “Deepening” Strategies:

1

{ }

{a} {d}

{a, b} {a, d}

{a, b, d}{a, b, c}

{a, b, c, d}

{a, b, c, d, e}

Current

competence state

2

{ }

{a} {d}

{a, b} {a, d}

{a, b, d}{a, b, c}

{a, b, c, d}

{a, b, c, d, e}

Selected

competence

Possible states to

look for activities that

work the selected

competence

3

{ }

{a} {d}

{a, b} {a, d}

{a, b, d}{a, b, c}

{a, b, c, d}

{a, b, c, d, e}

Activity

1

Activity

N

NEXT

ACTIVITY

SUGGESTEDactivity done

Competence state that

contains the selected

competence and activities

to be performed

NOTE: The percentage of achievement of

competence ‘a’ is below above a certain

threshold indicated by the teacher


9 | 16


PROPOSED SOLUTION

Implementation of the competence structure based on the Q-Matrix: Overview (1/7)

Create competence states

End

Start

Build the competence structure

Is there a CS with all the competences?

Create a competence state that contains all the competences from the list of SG activities

no

yes

Do all CSs contain successors?

Do all CSs contain predecessors?

yes

yes

Look at the successors of all competence states

Look at the predecessors of all competence states

Creates new predecessors if the competence state does not have

no

no

Creates new successors if the competence state does not have


10 | 16


PROPOSED SOLUTION

Implementation of the competence structure based on the Q-Matrix : Creation of Successors (2/7)

1

CSs without successors


11 | 16


PROPOSED SOLUTION


[a], [b], [c]

[e]

Competences different

from the CS not linked:

Competences from the

CS not linked:CS not linked

Closest

successors

2 3


12 | 16


PROPOSED SOLUTION


[a], [b], [c]

[e]

Competences different from the CS not linked:

Competences from the CS not linked:

COMBINATIONS OF COMPETENCES TO

GENERATE NEW CSs

These competences are

fixed in the combinations

These competences vary

in the combinations

[b, e] [a, e] [c, e]

4


13 | 16


PROPOSED SOLUTION

Implementation of the competence structure based on the Q-Matrix : Creation of Predecessors (5/7)

CSs without predecessors

1


14 | 16


PROPOSED SOLUTION


[f], [i], [j]

[e]

Competences different from

the “closest” previous CS:

Competences from the

“closest” previous CS :

CS not linked

“Closest” previous CS in

which all its competences

are in the CS not linked

2 3


15 | 16


PROPOSED SOLUTION


4

[f], [i], [j]

[e]

Competences different from the “closest” previous CS:

Competences from the “closest” previous CS :

COMBINATIONS OF COMPETENCES TO

GENERATE NEW CSs

These competences are

fixed in the combinations

These competences vary

in the combinations

CONTRIBUTIONS

16 | 16

Competence-based Knowledge Space Theory (CbKST)Requirements from Play Serious

Project for adapting Serious Games (SGs)

To propose an alternative approach to create

competence structures

To propose recommendation strategies based on teachers’

requirements

To propose a solution to create well-connected competence

structures


“Complete” Evaluation of the approaches based on CbKST and the recommendation strategies

Preliminary Evaluation and Testing withBlockly Maze, Cristaux d’Éhère, and Refraction

CONTEXT

CHALLENGES

PROPOSED SOLUTION

EVALUATION

CONTRIBUTIONS AT (submitted):

[EIAH 2015] 7ème Conférence sur les Environnements Informatiques pour l’Apprentissage Humain

[CSEDU 2015] 7th International Conference on Computer Supported Education

[UMAP 2015]23rd Conference on User Modelling,

Adaptation and Personalization

CONTRIBUTIONS:

• Software• Technical Documentation

EXPECTED CONTRIBUTION:

• Journal Paper with the “Complete” Evaluation

THANKS!

DECISION

MODULEJavier Melero & Naïma El-Kechaï

November 20, 2015

Decision Module: Adapting Learning Paths in Serious Games based on CbKST and Teachers' Decisions

Technology

competence structuresto

competence structureecdba

competences activities

specific competences

subset of competences

meaningful competence

competence states css

serious project