An Analysis Framework for Designing Declarative Knowledge Training

Games Using Roguelite Genre

enice Lemoine

, Pierre Laforcade

and S

ebastien George

LIUM Computer Science Laboratory, Le Mans Universit

e, Laval, France

Keywords:

Analysis, Serious Game, Declarative Knowledge, Training, Roguelite, Gameplay.

Abstract:

The training of declarative knowledge requires repetition and adaptation to learners’ needs. Learning games

for training purposes should then offer a wide range of adapted and varied game situations where facts are

questioned. Furthermore, answering these questions may involve many gameplays that keep the learner-

players engaged and motivated to practice again and again. This article presents and justiﬁes how the Roguelite

game genre is well-adapted to tackle these challenges. It also proposes an analysis framework to support

pluridisciplinary teams of teachers and game developers in identifying the key orientations for designing such

training games. This framework is composed of questions to consider during the preliminary design of the

training game. We have identiﬁed and used this proposition in a speciﬁc research context about multiplication

tables training. The article illustrates the ﬁrst results obtained which led to our ﬁrst playable prototype. Finally,

we outlined the major drawbacks of our ﬁrst game design (i.e., ﬁrst analysis), which brought us to carry out a

second analysis through our proposed framework.

1 INTRODUCTION

Explicit knowledge of facts, such as multiplication ta-

bles, historical dates or geographical information, are

known as declarative knowledge. Repetition is a nec-

essary task to encourage the memorization, general-

ization, and retention of declarative knowledge (Kim

et al., 2013; Roediger and Pyc, 2012). However,

repetitive or redundant serious games as well as the

ones presenting an imbalanced challenge relative to

the skills of the players can be boring for the play-

ers (Streicher and Smeddinck, 2016). Therefore, to

reduce the feeling of repetition, serious games target-

ing declarative knowledge should offer a wide variety

of adapted learning game activities.

Most of existing serious games for the training

of multiplication tables only present questions to an-

swer, surrounded by game elements that do not im-

pact the gameplay. Such training games do not pro-

pose long-term engaging activities. Poor gameplay

could quickly bore students and reduce their engage-

ment to practice (i.e., motivation, frequency, and du-

ration of sessions, etc.). It therefore seems important

https://orcid.org/0000-0002-7608-3223

https://orcid.org/0000-0001-8498-2731

https://orcid.org/0000-0003-0812-0712

to allocate as much importance to gameplay as to ed-

ucational content when designing learning game ac-

tivities (Marty and Carron, 2011).

This article proposes to highlight a speciﬁc game

genre, Roguelites. Indeed, this genre is based on sev-

eral design principles, of which we propose to dis-

cuss their compliance with the requirements for efﬁ-

cient and adapted training of declarative knowledge.

Our proposal is presented in the form of an analysis

framework in order to help the design of Roguelite-

oriented training games. This framework involves a

set of practical steps to be followed in each iteration

of a prototyping-based design approach. The aim is

to focus on design needs, from two dimensions (i.e.,

training and game), such as: technology (i.e., some

information needed for the game engine/generation

algorithm), game-gameplay (i.e., how the main game

mechanics of the Roguelite genre work), and game

structure (i.e., the game rules). To illustrate our pro-

posal, we present and discuss the application of this

analysis framework during two design iterations of

the same research context: the AdapTABLES project.

Test-based learning represents, in cognitive psy-

chology, the idea that the process of retrieving (i.e.,

remembering) concepts or facts increases their long-

term retention. Whilst tests are mainly used as sum-

mative assessment tools, they can also be formative.

276

Lemoine, B., Laforcade, P. and George, S.

An Analysis Framework for Designing Declarative Knowledge Training Games Using Roguelite Genre.

DOI: 10.5220/0011840200003470

In Proceedings of the 15th International Conference on Computer Supported Education (CSEDU 2023) - Volume 2, pages 276-287

ISBN: 978-989-758-641-5; ISSN: 2184-5026

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

One way to implement test-based learning is through

repeated retrieval (i.e., retrieval practice). Research

has shown that this method can improve long-term

retention (Brame and Biel, 2015; Roediger and Pyc,

2012). Moreover, research also suggests that the ben-

eﬁts of test-based learning are not linked to a speciﬁc

type of retrieval practice (i.e., various test formats en-

hance learning) (Brame and Biel, 2015). In this paper,

training involves providing the learner with various

forms of questions repeatedly, which is a form of re-

trieval practice.

The structure of this paper is as follows. Section 2

introduces our research context, including our AdapT-

ABLES project and its speciﬁc declarative knowledge

about multiplication tables. In addition, it presents the

video game Roguelite genre and discusses consider-

ing Roguelites as a suitable genre for designing train-

ing games. Section 3 is dedicated to the proposition of

a two-dimensional analysis framework (gaming and

training dimensions). This proposal has been applied

twice in the context of our project. Section 4 is then

devoted to the presentation of these two applications,

as well as the feedback gathered from the evaluation

of a prototype in line with the ﬁrst analysis.

2 RESEARCH CONTEXT

The AdapTABLES project aims to design and de-

velop a serious game dedicated to the individual train-

ing of multiplication tables.

From a teacher perspective, the training game to

design will be adapted, prior to its use, to reﬂect how

teachers consider the training: source facts, difﬁculty,

progress... This training structure can be set up for

the entire classroom’s students, for a group, or for in-

dividuals having speciﬁc needs. From a student per-

spective, the training game will follow the learners’

progress, proposing facts according to their previous

training sessions and results. From a player perspec-

tive, the training game will offer game levels that take

into account their preferences. From a game perspec-

tive, a same training task should be tackled through

different gameplays with different game elements. Fi-

nally, at runtime, the training game will have to gen-

erate varied training game activities, adapted, both

in terms of gameplay and educational content, to the

teachers and learners-players perspectives.

We followed an iterative co-design and prototyp-

ing approach, involving teachers and didacticians of

mathematics, and game experts during the design and

evaluation phases. At ﬁrst, two initial steps were nec-

essary: 1) specifying the knowledge to be trained, and

2) choosing a type of game that suits the training of

declarative knowledge. These contextual elements are

necessary to start designing at a high level the main

key concepts and rules of the training game.

2.1 Declarative Knowledge Training

We started an exploratory research (Laforcade et al.,

2022) with the help of a user group composed of

teachers and mathematics experts. The objective was

to specify the adaptations to take into account when

considering the training of multiplication tables from

a teacher perspective: what to consider (source and

targets of the adaptations) and how to realize these

adaptations. The main two results are: a model of

the training organization into training paths, and the

speciﬁcation of 5 detailed training tasks.

A training path, see Figure 1, is represented by a

set of objectives ordered by prerequisite relationships.

An objective (e.g., “Work on the table 2”) is broken

down into progressive levels of difﬁculty. Each level

is itself broken down into training tasks (e.g., “Level

1: Completion 1 with search for the result, Identiﬁ-

cation by choice of the correct facts”). A task is de-

ﬁned by its type and parameters. The levels’ achieve-

ments are considered from both a percent of encoun-

tered facts and a percent of achievement to reach.

Figure 1: Knowledge Structure.

Five types of tasks have been identiﬁed with the

teachers: Completion 1, i.e., complete an incomplete

fact that has one missing element (e.g., 3×? = 15,

15 =? × 5, 3 × 5 = ?); Completion 2, i.e., com-

plete an incomplete fact that has two missing ele-

ments (e.g., ?×? = 15 with a set of given choices

[3, 6, 5, 10], ? × 5 = ? or 3×? = ? also with sets of

given choices); Reconstruction, i.e., replace, in the

correct order, all important elements of a fact (e.g.,

?×? = ? with a set of given choices [3, 6, 5, 10, 15]);

Identiﬁcation, i.e., identify the correctness or incor-

rectness of one or several facts (e.g., 3 × 5 = 15, true

or false?); Membership Identiﬁcation, i.e., identify

the elements that share or do not share a given prop-

erty (e.g., [3, 5, 9, 14, 21] which are results of table 3?).

Considering the training tasks, their speciﬁcation

led us to deﬁne several parameters whose values de-

An Analysis Framework for Designing Declarative Knowledge Training Games Using Roguelite Genre

277

pend on teachers’ opinion and preferences to choose

and set-up these tasks for each {objective, level} pair.

For example, the parameters for the Completion 1 task

are presented in Table 1: the targeted tables, the mul-

tiplicand position, the result position, the interval (i.e.,

min/max) of the multiplier, the elements to search, the

order of the questions, the modality of response, and

the maximum response time.

As we are following a prototyping design ap-

proach, the design is carried out step by step. There-

fore, each new iteration takes into account a little

more information than the previous one. This article

presents the ﬁrst two iterations of our design, which

does not yet include the full knowledge structuring

presented. The ﬁrst prototypes aim at a parameteriza-

tion approach, i.e., teachers have to give the parameter

values (cf. Table 1) for each learner inside the game.

2.2 RogueLite Genre

Roguelike and Roguelite are video games genres

growing in popularity this last decade. They stem

from the game that pioneered this type of gameplay:

Rogue (Toy et al., 1980). Rogue was a turn-based

dungeon crawler game where you have to ﬁght your

way through levels of a dungeon, picking up items

and defeating enemies along the way. The Berlin In-

terpretation (Harris, 2020) deﬁned 8 high-value fac-

tors for Roguelikes, which includes:

• Random environment generation: different room

layouts with randomized locations for enemies

and items each time you play. This is usually re-

alized with procedural generations, not total ran-

domness, to avoid unwinnable situations.

• Permanent death: when the avatar die, all progress

is lost and the player must start over from the be-

ginning. No progress is carried over across runs.

Most of Roguelikes games fall to respect all 8 key

aspects, like the turn-based gameplay with movement

on a grid. As a result, people started considering

these games as “roguelike-like” or “roguelite”. For

now, Roguelite genre is used in relation to Roguelike

games, proposing a macro-level objective by carry-

ing over some items and progression after each

attempt. Some well-known commercial Roguelites

games are: Hades, Enter the Gungeon, The bind-

ing of Isaac, Rogue Legacy and Deadcells. They

propose different kind of gameplay, game styles and

lore as well as different features and permanent el-

ements to reach the game cross-run objective (e.g.,

weapons, currencies, upgrades, etc.). For example,

some collectible resources can persist between deaths

and players can use them to unlock permanent up-

grades and increase their chance of success.

Failure is a key-part of Roguelites. When players

start, the new mechanics, traps, difﬁcult enemies and

bosses, or the various features that they need to learn,

will lead them to fail and/or die many times before

they win their ﬁrst complete run. While losing over

and over again may not sound fun, Roguelikes typi-

cally have fast restart times and will quickly get play-

ers back into the action. Through playing more runs,

players begin to understand the underlying mechanics

and get further.

Figure 2: Roguelite Training Game Flow.

The game ﬂow (i.e., temporal representation of

game sessions) of Roguelite oriented games can be

different from one game to another. This article con-

siders the game ﬂow presented in Figure 2 where a

play session is a temporal session that begin when the

player starts the game and that ends when he/she stops

it. It is composed of complete or incomplete runs.

A run is a succession of game levels, played without

dying, with increasing difﬁculty. There are two con-

ditions for stopping the run: the end of the game is

reached or the avatar died. A game level is gener-

ally a level of a dungeon (or something similar) that

is composed of interconnected rooms or areas.

2.3 Adequacy of Declarative Knowledge

Training with RogueLite Genre

Replayability is at the heart of the Roguelite genre,

which is designed to keep players interested while re-

lying on a repetitive mechanic (i.e., permanent death).

In addition, the random environment generation fea-

ture offers a wide variety of dungeons (i.e., each gen-

erated dungeon level is different in terms of rooms,

room succession and elements). The procedurally

generated maps will always provide a different expe-

rience with each play through. Finally, the progress

feature allows for the storage of player data. All these

features are necessary elements in our research con-

text. Repetition to retain declarative knowledge, gen-

eration of varied activities to limit the impact of rep-

etition, and progression to keep track of the learner’s

progress and preferences in order to adapt the activi-

ties. Therefore, Roguelite seems to be a suitable genre

for declarative knowledge training, where the training

game activities generated are dungeon levels.

CSEDU 2023 - 15th International Conference on Computer Supported Education

278

Table 1: Examples of parameters for the ”Completion 1” training task.

Adaptable element Possible Values Some Examples

Targeted Table(s) From 1 to 12

Multiplicand Position Left ∨ Right

1 × 2, 1 × 3, 1 × 4.. ∨ 2 × 1, 3 × 1, 4 × 1..

Result Position Left ∨ Right 1 × 2 = 2 ∨ 2 = 1 × 2

Multiplier Interval Integer Min/Max in [1, 12] [1, 5] ∨ [5, 10] ∨ [1, 12]

Element to search Result ∨ Multiplicand ∨ Operand

1 × ? = 2 ∨ ? × 2 = 2 ∨ 1 × 2 = ?

Questions Order Ascending ⊕ Descending ⊕ Random

Response Modality Choice between propositions ∨ Input

Max Response Time Time in seconds

2.4 Targeted Adaptations

Our general context is about the adaptation of the gen-

erated game and learning activities. This adaptation

needs then to be characterized from both game and

learning perspective. Adaptation is often character-

ized by three concepts: the source (i.e., to what do

we adapt?), the target (i.e., what is adapted?) and the

pathways (i.e., what methods are used to adapt the tar-

get to the source?) (Vandewaetere et al., 2011).

Foremost, in our context, the adaptation targets

generated dungeon and their elements (i.e., what is

adapted). Therefore, it takes place during the gen-

eration of an activity (i.e., when it is adapted). In

the spectrum of adaptation deﬁned by (Oppermann

and Rashev, 1997), our targeted adaptation can be po-

sitioned in-between adaptivity and adaptability, as it

uses user’s data previously collected to automatically

generate an activity that is adapted to her/him.

In the literature, the gaming adaptations are

mostly based on players/personalities proﬁles (Ton-

dello et al., 2016; Nacke et al., 2014; Goldberg, 1990)

or players characteristics, such as age and genre. In

our context, adaptation from a game perspective seeks

to take into account player preferences to choose the

game elements (i.e., source). The main idea is to

represent preferences as game elements that can be

activated/deactivated by the player. From the learn-

ing perspective, our intention is to use knowledge of

the learner (e.g., actual level, previous mistakes) from

his/her learning path (source) to adapt the dungeons’

difﬁculty in terms of educational content.

Since the adaptation is an integral part of the gen-

eration, in our context, this article does not dissociate

them (i.e., the generation criterion includes adapta-

tion, see Section 3).

2.5 Research Question

Many questions need to be answered when design-

ing a Roguelite oriented game for training declara-

tive knowledge. What is generated? How and when

does the avatar die? What are the consequences?

What varies? What indicates a progression? And

so on. Furthermore, these questions need to be an-

swered from both an educational and a game perspec-

tive. Moreover, in a prototyping design approach, the

answers may change from one prototype to another.

Therefore, our research question is: How can the de-

sign of Roguelite-oriented learning be facilitated in

a prototyping design approach? Our proposal is an

analysis framework that helps designers ask the right

questions and make their choices explicit during each

design iteration.

3 ANALYSIS FRAMEWORK FOR

A ROGUELITE LEARNING

GAME

Firstly, it is important to keep in mind that we are de-

signing a game. Therefore, although the design tends

to focus on learning, the game aspect must not be ne-

glected. To this extent, the proposed framework aims

to provide a means of analyzing the design needs of

Roguelite-oriented learning games by specifying both

dimensions through speciﬁc criteria.

The ﬁrst step in designing a Roguelite is the spec-

iﬁcation of game mechanisms, such as generation

(e.g., what is generated? How?), permanent death

(e.g., when does one die?), as well as progression

(e.g., what elements are kept?). Hence, the genera-

tion mechanism involves the speciﬁcation of the vari-

ety mechanism, e.g., which elements should the gen-

eration vary? In a game, as in learning, an essential

notion is that of difﬁculty progression, e.g., how does

one increase difﬁculty? When? In our context, the

5 mechanisms (Generation, Death/Hurt, Variety,

Progress, and Difﬁculty) are criteria for analyzing

needs that must be speciﬁed from both points of view.

Each criterion consists of a series of questions that re-

late to the same mechanism. Each question should be

answered in order to clarify the design needs of the

An Analysis Framework for Designing Declarative Knowledge Training Games Using Roguelite Genre

279

learning game. The questions per criterion are:

1) Generation

Q1. Which elements are generated?

Q2. When are they generated?

Q3. On what basis are they generated? (i.e., sources

of generation)

2) Death/Hurt

Q4. When can the avatar get injured or die?

Q5. What are the consequences of being injured?

Being killed?

Q6. Where can the avatar be injured or killed?

3) Variety

Q7. Which elements vary?

Q8. How do the elements vary? (i.e., are the varia-

tions initiated by player action? Are they ran-

dom? Is it a mixture of both? Are they based

on heuristics?)

4) Progress

Q9. What is retained/preserved in-between each

death? (i.e., what elements?)

5) Difﬁculty

Q10. What are the elements that increase or decrease

the difﬁculty?

Q11. How is the difﬁculty progression designed?

(i.e., if several elements have an impact on the

difﬁculty, in what order do they occur?)

Table 2: Grid for Design Needs Analysis.

Educational Game

Criteria

Perspective Perspective

Q2Generation

Q5Death/Hurt

Variety

Progress Q9

Q10

Difﬁculty

Q11

The table 2 presents a structure to be completed

for the needs analysis. Each row represents a criterion

and is divided into X sub-rows (i.e., one sub-row per

question). Each column represents a dimension (i.e.,

a column for the game dimension, another for the ed-

ucational dimension). If both dimensions have some

common information, it can be precised by merging

both related cells. The proposed framework is inde-

pendent of the learning ﬁeld and is suitable for the de-

sign of any learning game based on Roguelite genre.

The following section presents the application in the

context of the AdapTABLES project.

4 FRAMEWORK APPLICATION:

AdapTABLES PROJECT

This section is broken down into four subsections:

subsection 4.1 details the analysis for the design of the

ﬁrst prototype; subsection 4.2 describes the current

prototype; subsection 4.3 presents the feedbacks (col-

lected informally in real-life conditions) on the proto-

type); subsection 4.4 speciﬁes the design needs anal-

ysis for the next prototype, and subsection 4.5 dis-

cusses the proposition.

4.1 First Analysis

The initial step in our iterative process was to carry

out the design needs analysis for the ﬁrst prototype.

The initial scope was to put aside the knowledge

structure by focusing on one task only (Completion

1 for multiplication tables) that will be manually set

up into the game (persistent information through dif-

ferent training sessions). Table 3 presents an overview

of the ﬁrst design needs analysis.

Generation. The generated element (Q1) is a dun-

geon level and its elements (i.e., rooms, rooms or-

der, rooms elements, elements positions and values).

Each room has a type between question room and no-

question room. A question room is associated to a

task (i.e., the training task deﬁned by the teacher and

set-up before playing). A no-question room is a fun

room where enemies, traps, and only game elements

are present. The presence of purely gaming rooms in-

tends to avoid giving the learners-players the impres-

sion that they are simply answering a disguised ques-

tionnaire. Following the game ﬂow presented in Fig-

ure 2, a dungeon is generated when (Q2) the player

asks for a new game level.

As mentioned in Section 2.4, our game adapta-

tions target players’ preferences. To identify the pref-

erences, we ﬁrst looked at the existing Roguelites and

found out that many of them offer a mechanism for

purchasing items (e.g., equipments, upgrades, skills).

Then, by working with game designers, we identiﬁed

three kinds of preferences: 1) Content, 2) Rules and

3) Visuals & Audio (Q3). Content preferences are

additional objects present while playing or elements

that change the structure of the activity if activated.

For example, an extra life or a dungeon mode (linear

CSEDU 2023 - 15th International Conference on Computer Supported Education

280

or labyrinthine) are content preferences. Rules prefer-

ences are elements that impact the players, the avatar

or the NPCs (Non-Player Characters) behaviors. For

example, increasing the enemies speed or adding a

game goal (e.g., completing an activity without mis-

takes earns +10 coins) are rules preferences.

From the game dimension, the generation is based

on (Q3) these three types of preferences (e.g., if a

player bought and activated the labyrinthine mode,

the generation algorithm takes it into account). To

keep track of learners’ in game progress, the generator

also takes into account the last level number and state:

if the previous level was #5 and was successful, then

the next level is #6; in case of death, the next level is

reset to #1. Level number affects the dungeon length,

the quantity of rooms with questions, and the dun-

geon effects (levels greater than #4 are in dark mode).

In the educational dimension, many tasks parameters

can change depending on the level of the learners (cf.

Section 2). Therefore, all learners have their own set-

up for the task Completion 1 (or a shared one) de-

ﬁned by the teachers. These parameters values are

used to produce relevant questioned facts associated

to the rooms-with-questions (Q3). Previous encoun-

tered questioned facts and results are used to avoid

answering again a successful questioned fact.

It is worth notice that both game and train-

ing dimensions can sometimes conﬂict. For exam-

ple, if a learner-player has bought and activated the

‘labyrinthine’ mode and if the task setup requires the

learner to meet the questioned facts in an ascendant or

descendant order. In these cases, our recommendation

is to consider the training dimension as a priority.

Death/Hurt. In both cases, when the avatar is injured

(Q5), it loses a life, and when there is no life left, it

dies. The player (through the avatar) gets hurt (Q4)

when touching a foe, falling into a pit (game dimen-

sion) or when answering incorrectly to a question (ed-

ucational dimension). It also happens when the time

is over (preliminary set in the task parameters). An-

swering a question incorrectly or running out of time

can only be done in a question room (Q6), while hit-

ting the wrong enemy or falling into a pit can be done

in both types of rooms.

Variety. In terms of game variety, different elements

can be chosen by the generation algorithm. In the

context of Roguelite, the variations are mainly based

on the types of objects, their position, the shape of

objects and dungeons, and the number of elements

present (Q7). Therefore, the decorative objects po-

sitions as well as the rooms shapes are chosen “ran-

domly” (Q8) (i.e., the coherence is kept, e.g., the el-

ements are not outside the room or the avatar can ac-

cess the necessary elements). The gameplay repre-

sents the concrete way in which the learner performs

the task. Therefore, having only one form of game-

play per task can quickly lead to a sense of redun-

dancy. One way to avoid this sense is to vary the

gameplay. To that extent, 4 gameplays were identiﬁed

(Q7) for the Completion 1 task type: open the chest

wearing the correct answer, pass through the door that

has the correct answer, touch the foe wearing the cor-

rect answer, or type down the correct answer.

From the learning dimension, the facts to be

worked on are different, until each fact has at least

been seen once (Q7). Moreover, depending on the pa-

rameters (cf. Table 3), the facts shapes (e.g., missing

element, position of the equal, etc.) vary (Q8). Some

degree of randomness is always involved in the varia-

tion of elements. However, this randomness is limited

by the preferences of the game, the educational con-

straints (i.e., the teachers’ choices) as well as by the

choices made earlier (i.e., elements already selected

by the algorithm).

Progress. In order to take into account players’ pref-

erences, our main idea is to use a purchase mecha-

nism (i.e., commonly used in Roguelites) where play-

ers can buy items and then activate/deactivate them

as they wish. Therefore, progression from the game

viewpoint can be seen through the elements bought

and the number of coins (Q9). Some Roguelites only

keeps progression when the level/dungeons are ﬁn-

ished (i.e., the avatar did not die). Therefore, the coins

are only gained if the player ﬁnishes a dungeon com-

pletely. These coins are collected through the dun-

geon journey (randomly appear when opening a cor-

rect chest) or win at the end according to the activated

rules (for example, +10 if ending the dungeon level

without any mathematics error). Training progression

can be seen at the end of a dungeon (when reaching

the end or dying) with statistics presenting the mis-

takes made, the correct answers given, and what is

left to work on. All the results are persistent across

the runs.

Difﬁculty. From the game dimension, difﬁculty in-

creases during a run by increasing the number of

rooms-with-question (i.e., 5 rooms ﬁrst, then 7, then

9, then 11, etc.) (Q10). The total number of rooms

may still vary according to the generation process

that can randomly includes rooms without question

along with the dungeon structure (tortuous but lin-

ear or labyrinthine) (Q11). After 5 levels played,

without losing, a harder level is proposed where the

player is in the dark, a torch only illuminates the

avatar (Q10). The educational difﬁculty increases in

accordance with the parameters deﬁned by the teach-

ers, but while these parameters are not changed, the

questioned facts will always concern this setup (Q11).

An Analysis Framework for Designing Declarative Knowledge Training Games Using Roguelite Genre

281

These choices are mainly debatable, and intends to

deﬁne a ﬁrst version of the game’s difﬁculty progres-

sion, which is bound to evolve.

4.2 First Prototype

The prototype has been developed using the Unity

game engine, as a 2D game with C# scripts. It is

exported and deployed as a Web platform WebGL

build. The game uses an HTTP REST API (developed

in .Net Core) to persist data in a NoSQL MongoDB

database. In addition, a teacher dashboard is available

as a web application (developed in .NET with Blazor).

It allows teachers to monitor for their students: the

current settings for the multiplication parameters, cur-

rent achievement progress for the considered multi-

plication facts, current elements that has been bought.

Currently, only a French version playable with both a

gamepad or a keyboard is available.

Figure 3 shows 6 screenshots of the prototype.

The screen 3a represents part of the current ‘hub’

area with four accessible elements: statistics (“Stats

Generales”), progress (“Progression”), educational

settings (“Reglages”) and lastly the purchase panel

(“Achats”, i.e., game preferences). This ‘hub’ area

is where the player starts a run and goes back after he

died. It offers players a peaceful space where they can

check their progress or manage collectibles (purchase,

activation/deactivation). Screen 3b is an extract from

the educational settings panel (cf. Table 1). Screen 3c

is an extract from the item purchase panel. An ex-

ample of each of the currently implemented game-

plays present in the dungeon levels (i.e., choosing the

correct foe, the correct door, and the correct chest)

is illustrated in the Screens 3g, 3e, 3f. Finally, the

Screen 3h presents an example of a room where the

player has to type down his/her answer.

4.3 Experiment Feedback

The design of the prototype went through 3 iterations

where it was tested in real conditions and then im-

proved according to the feedback from teachers and

students (i.e., played in classroom by students su-

pervised by their teacher). Some empirical feedback

were about ergonomics concerns (keyboard versus

gamepad), playability experience, replayability incli-

nation, motivation to play and train the tables, etc.

The rest of the information collected were related

to the overall design, based on our previous analysis

choices, stated in the previous section. With regard

to the death/injury criterion, several problems have

been identiﬁed. Firstly, children sometimes make un-

intentional choices due to the current ‘touch’-oriented

interactions that do not require the use of a button

or key. Even if this gameplay problem is related

to ergonomics, it can lead to a feeling of unfairness

from the reward/punishment system. Secondly, some

rooms have randomly positioned foes very close to

the avatar’s entry area: swept along by the momen-

tum, children can lose hearts without having time to

avoid them. Similarly, some room-with-question also

embed holes to avoid. Teachers have pointed out to us

that these game elements can disturb children when

they should focus on the question.

Considering the variety criterion, children appre-

ciated the three gameplays for choosing an answer

(door, chest, and foe). This prototype offered 3 dif-

ferent room structures for each game, however it does

not vary enough. The gameplay about touching the

correct foe has been considered confusing by both

children and teachers. In some rooms, foes were to

be avoided, while in others the players must lead their

avatar to touch the foe with the answer they choose.

This is counter-intuitive, and teachers consider that

identifying a correct answer should not be associated

with a negative action (here, killing the foes). The

prototype also proposes several rooms-with-question

that require to directly type down the correct answer

on the keyboard. According to the correctness or not

of the answer, the right door or chest opens, or all foes

die. Any incorrect answer leads to open an empty

chest, open a door towards a dead end, or having no

effect on the touched foe, but in all three cases, in-

juring the avatar. These situations were initially de-

signed to vary the mode of response while providing

similar room content, but were ultimately confusing.

As far as the progression perspective is concerned,

we failed to provide a balanced way of collecting

coins. Only chest-oriented gameplays (choose or

write modality) can randomly contain ’+1’ or ’+3’

coins (or a ‘heart’/life). Some purchased and acti-

vated rules may provide other ways to earn coins,

but early successful dungeon levels may not result in

any coins being earned. Moreover, the ﬁrst purchased

items are mainly those of the content category (extra

hearts), preferred to rules. In addition, some teachers

were not convinced about leaving learners free to buy

and activate rules that push them to act faster, stress-

ing them to answer quickly.

As mentioned earlier, the generation process may

disable some activated rules that do not conform to the

current conﬁguration of the Completion 1 task. This

can lead to feelings of confusion when discovering the

generated dungeon levels.

CSEDU 2023 - 15th International Conference on Computer Supported Education

282

Table 3: Design Needs of the prototype #1.

Educational Game

Criteria

Perspective Perspective

Q1: What?

One task and one questioned fact per

Dungeon + rooms + entry + exit

room-with-question

Q2: When? When a new game level is required

Generation

Q3: Based

”Completion 1” set-up Previous level number and state

Current progress among possible facts Activated game elements or rules

on?

Task parameters have priority on activated game elements if conﬂict

Q4: What? Incorrect answers or time out

Being touched by foes, falling

into holes

Q5: When? Injuring causes heart lost, no more hearts causes death

Death/Hurt

Q6: Where? Question rooms Any room with foes or holes

Q7: What? Facts Rooms with gameplay and content

Variety

Q8: How? Progress and past results Random

Q9: What?

Success or failure on met Coins collected during successful

Progress

questioned facts game levels + purchased elements

Q10: What? Questioned facts

Dungeon level length

+ dark mode

Difﬁculty

Q11: How? In relation with the task parameters

According to previous level number

and state

4.4 Second Analysis

This second analysis was conducted for the design

of the next multiplication tables training prototype.

Based on teachers’ and learners’ feedback, we worked

with game designers to identify solutions and further

directions, both from a game and training dimension.

As with the functional scope of the ﬁrst prototype,

the knowledge structure is still not taken into account,

although all 5 task types are considered. The proto-

type will make it possible to manually parameterize

the current training conﬁguration for any children ac-

cording to their individual progress (correspondence

of a {objective, level} pair) of a learner’s training

path, cf. Figure 1).

Generation. The generated element (Q1) is still a

dungeon level composed of organized room of two

types: room without question, and room with one

question associated to one of the types of task in-

volved according to the training setup. Each room is

composed of various interactive elements. This dun-

geon level is generated when (Q2) the player requests

a new game level, from the hub-room (to start a new

run) or after the debrieﬁng screens following a suc-

cessful dungeon level. From the game dimension, the

generation continues to take into account the last level

number and state (Q3) as well as the purchased and

activated features. However, the purchasable (activat-

able) elements have changed. These elements are ex-

plained in the following categories. From an educa-

tional dimension, each generation considers the cur-

rent conﬁguration of the learner for all types of tasks

involved (from 1 to 5), and takes into account the pre-

vious questioned facts and the previous results.

Death/Hurt. The player still gets hurt (Q4) when

touching a foe, falling into a pit (game dimension),

when answering incorrectly to a question or lacking

of time (educational dimension). The main difference

is that the question rooms will no longer have traps

and game elements that can hurt the avatar other than

those related to the question to be answered (Q6).

Getting injured in a question room will always be

caused by a wrong answer or a time-out. The con-

sequences of an injury are not changed (Q5), it leads

to the loss of a life, or death.

Variety. In the ﬁrst prototype, the different types of

rooms merged game elements and gameplays. In or-

der to propose more varieties of rooms (Q7), we con-

sider a new approach (see. the conceptual class dia-

gram in Figure 4). On one hand, the types of tasks

are mapped to different gameplay types according to

the current tasks parameters. Every type of game-

play requires a quantiﬁed number of elements having

the speciﬁed ability. On the other hand, the types of

rooms describe some positions accepting different el-

ements having a given ability. By extension, different

types of game elements are associated to the abilities

they can manage.

Combined, these elements will drive the genera-

tion of rooms and their built-in elements (Q8). As a

result, the purchase mechanic now consists of players

choosing items that will, if equipped (i.e., activated),

An Analysis Framework for Designing Declarative Knowledge Training Games Using Roguelite Genre

283

(a) Prototype’s hub (b) Training settings panel

(e) Door Gameplay (f) Chest Gameplay

(g) Foe Gameplay (h) Answer entry

Figure 3: Screenshots of the current prototype features (text in French).

unlock new types of gameplays that can occur in cer-

tain rooms if they are consistent with the task associ-

ated to the room. By providing variants of game ele-

ments that share these abilities, game developers will

increase the possibilities for variations. Let’s take as

an example the “boxing gloves” item, that allow the

destruction of every “destroyable” elements (i.e., hav-

ing an ability). Several elements can have “destroy-

able” as an ability: walls, pots, statues, rocks, etc. If

the avatar is equipped with this item, it is possible to

create multiple variants of gameplays. The mecha-

nism remains the same, but the elements vary (e.g.,

destroy the wall, break the rock).

Progress. Equipment items are deﬁnitely purchased

across runs. The coin mechanism is retained, but will

be adjusted so the learners will earn one coin per cor-

rect answer. Questioned facts encountered and asso-

ciated results are also saved beyond death. Therefore,

the progression can be observed through: the equip-

ments available, the number of coins, and fact results,

i.e., available outside a game level (Q9).

Difﬁculty. The educational difﬁculty is maintained

as in the ﬁrst prototype, i.e., according to the param-

eters of the tasks supervised by the teachers (Q11).

From a gameplay point of view, the progression ac-

cording to the length of the dungeons is kept, but a

new gameplay mechanism is added: the curses (Q10).

The game progression will be structured in different

minimum thresholds, unlocking curses that may or

may not occur during the dungeon level to be gen-

CSEDU 2023 - 15th International Conference on Computer Supported Education

284

Table 4: Design Needs of the Future Prototype (Bold describes changes from the ﬁrst analysis).

Educational Game

Criteria

Perspective Perspective

Q1: What?

One task and one questioned fact

Dungeon structure + rooms

per room-with-question

Q2: When? When a new game level is required

Generation

Q3: Based

All tasks set-up Previous level number and state

Current progress among possible facts Equipped items

on?

Task parameters have priority on activated game elements if conﬂict

Q4: When? Incorrect answers or time out

Being touched by foes, falling

into holes

Q5: What? Injuring causes heart lost, no more hearts causes death

Death/Hurt

Q6: Where? Question rooms Only rooms with no question

Q7: What? Facts

Different types of rooms, types of

gameplays, types of elements

Variety

Q8: How?

Progress and past results

Based on the available equipments,

gameplays, elements,

and in relation to the tasks ⇐⇒ gameplays mappings

Q9: What?

Success or failure on met Coins collected during successful

Progress

questioned facts game levels + purchased items

Q10: What? Questioned facts

Dungeon level length

+ curses

Difﬁculty

Q11: How? In relation with the task parameters

According to previous level number

and state

Figure 4: Conceptual class diagram illustrating the domain elements and relationships involved in considering a wide variety

of question pieces. Yellow concepts are to be generated; blue concepts are speciﬁcations of the game and didactic elements

available; green concepts concern each learner-player.

erated (Q11). For example, if we consider thresholds

every 3 dungeon levels, the generation of level #10

may involve at worst 3 curses, or 0 with luck.

4.5 Discussion

The ﬁrst advantage of the framework is the trace-

ability of design choices. Indeed, at each itera-

tion, the choices are explained and then summarized.

This traceability facilitates the evolution of the design

without the risk of involuntarily going backwards. On

the other hand, the visual synthesis (cf. Table 2)

makes it possible to check that none of the dimensions

has been neglected (i.e., presence of an empty box in

case of neglect). In a prototyping approach, iterations

are essential to ﬁx certain settings. However, it would

seem that the use of such a framework (i.e., allowing

traceability as well as visual veriﬁcation of the non-

neglect of a dimension) could reduce the number of

iterations required. Finally, this tool provides a sup-

port that can be understood by all stakeholders of the

design process.

However, the proposed framework takes into ac-

count a rather precise context: training or retrieval

practice (i.e., a form of test-based learning) in the con-

text of the Roguelite video game genre. Moreover,

An Analysis Framework for Designing Declarative Knowledge Training Games Using Roguelite Genre

285

the criteria used are those that we consider essential.

Consequently, other criteria might be considered es-

sential by other researchers or game designers. In

particular, some criteria depending on the application

domains might be interesting to add.

5 RELATED WORK

The literature includes many frameworks and meth-

ods for games, serious games, game-based scenario

design. However, many of them are mainly oriented

towards and used for the analysis of already existing

games (Junior and Silva, 2021).

(De Freitas and Jarvis, 2006) proposed a four-

dimensional learner-centered framework (Represen-

tation, Context, Pedagogy, Learner) to help design

game-based learning scenarios. This approach is at

a high level of design, which does not facilitate the

transition from educational content to concrete game

elements. (Winn, 2009) introduced the DPE frame-

work, an extension of the Mechanics, Dynamics, Aes-

thetics (MDA) framework (Hunicke et al., 2004) for

serious games. This framework is broken down into

three categories (design, play, experience; i.e., the de-

signer designs the game, the player plays the game,

which results in player’s experience), each described

by four criteria: Learning, Narrative, Gameplay, User

Experience. (Schell, 2008) proposed a method con-

sisting of a series of questions on different aspects to

be taken into account for the design of the game. This

method is more generic than ours (i.e., at a higher

design level) because it does not focus on a speciﬁc

type of knowledge or game genre. (Amory, 2007)

described GOM II, a framework for the design of

educational games extended from the Game Object

Model (GOM). This framework considers that an ed-

ucational game is composed of a set of elements de-

scribed by abstract and concrete interfaces (based on

the Object-Oriented paradigm). This work is theo-

retical and focuses on the general design of games

rather than their concrete implementation. (Arnab

et al., 2015) proposed the LM-GM framework that en-

ables the association of Learning Mechanics to Game

Mechanics through the use of Serious Game Mechan-

ics (SGM; “design decision that concretely realizes

the transition of a learning practice/goal into a me-

chanical element of gameplay for the sole purpose of

play and fun”). Nevertheless, this framework is more

oriented towards the analysis of games than towards

their design. (Carvalho et al., 2015) presented a con-

ceptual model, called ASTMG, based on activity the-

ory, which seeks to provide a better understanding of

the relationships between the serious games elements

and learning objectives/goals. This framework is also

more oriented towards game analysis.

Many other framework and methodologies ex-

ists (Yusoff et al., 2009; Barbosa et al., 2014; Marne

et al., 2012; Kiili, 2005; Silva, 2019). Some works

offer methods that are closer to some general game

genre (i.e., adventure games, story oriented games

for (De Lope et al., 2017)). However, these works are

mainly generic (i.e., they are not speciﬁc to a type of

knowledge or a game genre). Therefore, being in the

speciﬁc context of the Roguelite genre, none of the

existing works met our requirements. Nevertheless,

these frameworks are not incompatible and can be

used together. As examples: the framework proposed

by (Schell, 2008) could be used to describe the gen-

eral design of the game; the ASTMG (Carvalho et al.,

2015) could be used to verify the coherency of each

prototype. In short, the current frameworks that share

our goal of assisting the design of learning games deal

with different pedagogical objectives, different types

of knowledge, different game genres. They do not aim

to assist design in situations where the game genre is

already identiﬁed because of its relevance to a speciﬁc

learning objective. Our analysis framework is dedi-

cated to Roguelite games for declarative knowledge

retention: a very speciﬁc scope. Nevertheless, multi-

disciplinary teams also require speciﬁc frameworks to

guide them deeper in the design of relevant, adapted,

and balanced learning games.

6 CONCLUSION

This article 1) presented Roguelite as a suitable

game genre for declarative knowledge training, and

2) introduced a design needs analysis framework

for Roguelite-oriented learning game. The proposed

framework gives a very ﬁrst insight into the mecha-

nism and choices that beneﬁts both to the game and

training dimension. The main idea is to allow a two-

dimensional design between the play and learning di-

mensions. This design aims at a separate description

of the dimensions while allowing the veriﬁcation (and

maintenance) of both dimensions’ conformity.

This framework was used in a prototyping design

approach in the context of the AdapTABLES project.

This project aims at creating a game for multiplica-

tion tables training. This article presents the applica-

tion of the framework over two iterations: 1) the ﬁrst

iteration led to the current existing prototype, 2) the

second iteration presents the design requirements of

the future prototype.

In the future, we would like to apply this frame-

work to other ﬁelds of application than mathemat-

CSEDU 2023 - 15th International Conference on Computer Supported Education

286

ics. Currently, we are working on its application to

history-geography facts (e.g., historical dates, coun-

tries of the European Union).

REFERENCES

Amory, A. (2007). Game object model version II: A theo-

retical framework for educational game development.

Educational Technology Research and Development,

55(1):51–77.

Arnab, S., Lim, T., Carvalho, M. B., Bellotti, F., de Freitas,

S., Louchart, S., Suttie, N., Berta, R., and De Gloria,

A. (2015). Mapping learning and game mechanics for

serious games analysis: Mapping learning and game

mechanics. British Journal of Educational Technol-

ogy, 46(2):391–411.

Barbosa, A. F. S., Pereira, P. N. M., Dias, J. A. F. F., and

Silva, F. G. M. (2014). A New Methodology of Design

and Development of Serious Games. International

Journal of Computer Games Technology, 2014:1–8.

Brame, C. J. and Biel, R. (2015). Test-Enhanced Learning:

The Potential for Testing to Promote Greater Learning

in Undergraduate Science Courses. CBE—Life Sci-

ences Education, 14(2).

Carvalho, M. B., Bellotti, F., Berta, R., De Gloria, A.,

Sedano, C. I., Hauge, J. B., Hu, J., and Rauterberg,

M. (2015). An activity theory-based model for serious

games analysis and conceptual design. Computers &

Education, 87:166–181.

De Freitas, S. and Jarvis, S. (2006). A Framework for devel-

oping serious games to meet learner needs. Interser-

vice/Industry Training, Simulation & Education Con-

ference, I/ITSEC.

De Lope, R. P., Medina-Medina, N., Soldado, R. M., Gar-

cia, A. M., and Gutierrez-Vela, F. L. (2017). Design-

ing educational games: Key elements and method-

ological approach. In 9th International Conference

on Virtual Worlds and Games for Serious Applications

(VS-Games), pages 63–70, Athens, Greece. IEEE.

Goldberg, L. R. (1990). An alternative” description of per-

sonality”: The big-ﬁve factor structure. Journal of

personality and social psychology, 59(6):1216–1229.

Harris, J. (2020). The Berlin Interpretation. In Exploring

Roguelike Games, pages 37–43. CRC Press.

Hunicke, R., LeBlanc, M., and Zubek, R. (2004). MDA: A

formal approach to game design and game research.

In Proceedings of the AAAI Workshop on Challenges

in Game AI, volume 4. San Jose, CA.

Junior, R. and Silva, F. (2021). Redeﬁning the MDA Frame-

work—The Pursuit of a Game Design Ontology. In-

formation, 12(10).

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

experiential gaming model. The Internet and Higher

Education, 8(1):13–24.

Kim, J. W., Ritter, F. E., and Koubek, R. J. (2013). An inte-

grated theory for improved skill acquisition and reten-

tion in the three stages of learning. Theoretical Issues

in Ergonomics Science, 14(1):22–37.

Laforcade, P., Mottier, E., Jolivet, S., and Lemoine, B.

(2022). Expressing adaptations to take into account

in generator-based exercisers: An exploratory study

about multiplication facts. In 14th International Con-

ference CSEDU, Online Streaming, France.

Marne, B., Wisdom, J., Huynh-Kim-Bang, B., and Labat,

J.-M. (2012). The Six Facets of Serious Game De-

sign: A Methodology Enhanced by Our Design Pat-

tern Library. In 21st Century Learning for 21st Cen-

tury Skills, volume 7563, pages 208–221. Springer,

Berlin, Heidelberg.

Marty, J.-C. and Carron, T. (2011). Hints for improving

motivation in game-based learning environments. In

Handbook of Research on Improving Learning and

Motivation through Educational Games: Multidisci-

plinary Approaches, pages 530–549. IGI Global.

Nacke, L. E., Bateman, C., and Mandryk, R. L. (2014).

BrainHex: A neurobiological gamer typology survey.

Entertainment Computing, 5(1):55–62.

Oppermann, R. and Rashev, R. (1997). Adaptability and

adaptivity in learning systems. Knowledge transfer,

2:173–179.

Roediger, H. L. and Pyc, M. A. (2012). Inexpensive

techniques to improve education: Applying cognitive

psychology to enhance educational practice. Jour-

nal of Applied Research in Memory and Cognition,

1(4):242–248.

Schell, J. (2008). The Art of Game Design: A Book of

Lenses. CRC press.

Silva, F. G. M. (2019). Practical Methodology for the

Design of Educational Serious Games. Information,

11(1):14.

Streicher, A. and Smeddinck, J. D. (2016). Personalized and

Adaptive Serious Games. In D

orner, R., G

obel, S.,

Kickmeier-Rust, M., Masuch, M., and Zweig, K., ed-

itors, Entertainment Computing and Serious Games,

volume 9970, pages 332–377. Springer International

Publishing, Cham.

Tondello, G. F., Wehbe, R. R., Diamond, L., Busch, M.,

Marczewski, A., and Nacke, L. E. (2016). The Gamiﬁ-

cation User Types Hexad Scale. In Proceedings of the

Annual Symposium on Computer-Human Interaction

in Play, pages 229–243, Austin Texas USA. ACM.

Toy, M., Wichman, G., Arnold, K., and Lane, J. (1980).

Rogue [digital game].

Vandewaetere, M., Desmet, P., and Clarebout, G. (2011).

The contribution of learner characteristics in the de-

velopment of computer-based adaptive learning envi-

ronments. Computers in Human Behavior, 27(1):118–

130.

Winn, B. (2009). Handbook of Research on Effective Elec-

tronic Gaming in Education:. IGI Global.

Yusoff, A., Crowder, R., Gilbert, L., and Wills, G. (2009).

A Conceptual Framework for Serious Games. In 9th

IEEE International Conference on Advanced Learn-

ing Technologies, pages 21–23, Riga, Latvia.

An Analysis Framework for Designing Declarative Knowledge Training Games Using Roguelite Genre

287