Supporting Online Game Players by the Visualization of

Personalities and Skills Based on in-Game Statistics

Tatsuro Ide

and Hiroshi Hosobe

Graduate School of Computer and Information Sciences, Hosei University, Tokyo, Japan

Faculty of Computer and Information Sciences, Hosei University, Tokyo, Japan

Keywords: Online Cooperative Game, Team Matching, Statistical Analysis, Visualization.

Abstract: Although the COVID-19 pandemic has increased people demanding to play online cooperative games with

others, in-game random team matching has not fully supported it. Furthermore, toxic behaviors such as verbal

abuse and trolling by randomly gathered team members adversely affect user experience. Public Discord

servers and game-specific team matching services are often used to support this problem from outside the

game. However, in both services, players can obtain only a few lines of other players’ self-introductions

before playing together, and therefore their anxiety about possible mismatches is a major obstacle to the use

of these services. In this paper, we aim to support team matching in an online cooperative game from both

aspects of players’ personalities and skills. Especially, we perform team member recommendation based on

the visualization of in-game statistical information by computing players’ personalities and skills from their

game masteries and character preferences in a typical game called VALORANT.

1 INTRODUCTION

Today’s online video games are widely played as

cooperative games in which players cooperate with

others regardless of whether the type of game is

player-versus-environment (PvE) or player-versus-

player (PvP). In PvE, several players cooperate to

progress through the game; in PvP, teams consisting

of several players usually play against each other.

Such a game often has a friend search feature that

allows players to search for a player with a username,

is not possible to prepare friends to play with in

advance, the lack of team members is often filled by

random matching. Although this random matching

generally succeeds in matching players with similar

skills (Corem, Brown, & Petralia, 2013), it does not

sufficiently consider the compatibility between

players. Rather, randomly matched members might

not be able to cooperate due to the failure to divide

roles, and sometimes they might exhibit troll

behaviors such as intentional abuse and obstruction of

https://orcid.org/0000-0001-5787-0443

https://orcid.org/0000-0002-7975-052X

https://discord.com/

https://gametree.me/

allies (Ho & McLeod, 2008) (Cook, Conijn,

Schaafsma, & Antheunis, 2019) due to low

relationships and high anonymity (Kwak, Blackburn,

& Han, 2015). Although some games implement the

function for the same players to play together

continuously, it is rarely used due to the experience

mentioned above and the unknown background of the

players. Although the COVID-19 pandemic calls for

games as social spaces for people’s interaction (King,

Delfabbro, Billieux, & Potenza, 2020), a mechanism

to support it within the game has not been sufficiently

realized.

Discord

is software for communication with text

chats and voice/video calls mainly on personal

computers (PCs) and smartphones. It is characterized

by the easy preparation of a small server for a few

people as well as a large-scale community server for

thousands of people. A large-scale server is often

open to the public, and people who participate in the

server search for users with whom to play an online

game together; while cooperating in an online game,

they communicate via channels and individual calls

within the Discord server. GameTree

is a matching

Ide, T. and Hosobe, H.

Supporting Online Game Players by the Visualization of Personalities and Skills Based on in-Game Statistics.

DOI: 10.5220/0011784000003417

In Proceedings of the 18th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2023) - Volume 2: HUCAPP, pages

259-266

ISBN: 978-989-758-634-7; ISSN: 2184-4321

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

259

service specialized in finding people to play games

with. Users can search for other users to play with by

registering their simple attributes, the games that they

often play, and their self-introductions. However, in

both services, before users start playing together, the

other users’ displayed information is only a few lines

of self-introductions, and their anxiety about possible

mismatches is a major obstacle to the use of these

services.

In this paper, we aim to support the matching of

team members for an online cooperative game by

using the two axes of personalities and skills. Based

on in-game statistical information in a typical game

called VALORANT, we perform team member

recommendation based on the visualization of the

information by computing players’ personalities and

skills from their game masteries and character

preferences. We present the result of the experiment

that we conducted to evaluate our method.

2 BACKGROUND

In this section, we describe the basics of

VALORANT and Discord. First, we show the

differences and similarities between VALORANT

and other PvP online games, and explain how players

find their friends. Next, we describe the role of

Discord, how public Discord servers are used as a

communication platform, and an example of a

Discord server for VALORANT.

2.1 VALORANT as a Cooperative

Game

VALORANT is a 5-versus-5 cooperative character-

based tactical first-person shooter video game being

developed and published by Riot Games. According

to unofficial statistical information (Active Player,

2022), the numbers of players of VALORANT are 22

million per month and 2.3 million per day, and the

number of concurrent players at peak time exceeds

0.8 million people.

The main game modes are unrated and

competitive, where the goal is to destroy enemies or

to set or defuse bombs in the area. This rule is the

mainstream of first-person shooter games, and a

representative is the Counter-Strike series. On the

other hand, the difference from most of the other

games is that, in VALORANT, the characters used by

players have their own characteristics. These

characters can be classified into four roles (sentinels,

controllers, duelist, and initiators). This feature is

similar to that of Overwatch, a team-based action

game. In both games, players form a team with

friends before the match begins, and basically divide

their roles and fight together. Otherwise, players are

randomly assigned to teams with other players of

similar skill level for the game. It should be noted that

when forming a team of 2 or 3 players in the

competitive mode, players must have a rank

difference of about ±1 before they can line up in a

match queue. Although a 4-player team is not allowed

to play due to the game specifications, a 5-player

team (called a fully premade team) is allowed to line

up in the match queue regardless of the rank

difference. However, the enemy team also must be a

fully premade team.

2.2 Discord as a Communication

Platform

Discord is a VoIP and instant messaging platform

application. Millions of people send 4 billion

messages through the Discord platform every day

(Glen, 2022). Groups and communities in Discord are

called servers. In this paper, the term “server” is used

not to represent a central computer but to represent a

community. Servers range from small ones consisting

of closely related people to huge communities

consisting of several thousand people.

Figure 1: User interface of a Discord server. The left side

lists a part of the channels for text and voice

communications. The central part presents self-

introductions by users including their names, play styles

and short messages.

HUCAPP 2023 - 7th International Conference on Human Computer Interaction Theory and Applications

260

Figure 1 shows an example of a medium-sized

server (consisting of approximately 400 users) where

we conducted the experiment presented in this paper.

The server is further divided into channels that are

displayed in a list as shown on the left side. Text

channels are used for self-introductions, text chats,

and discussions. In the voice channel, users in the

server can freely enter and exit and directly

communicate with each other by voice. On the right

side of Figure 1, bot applications created by third

parties are shown. Over 500,000 bot applications are

used on Discord servers to help server customization,

playing games, and keeping communities safe (Cap,

2022). We implemented our method as a bot

application.

3 RELATED WORK

Delhove et al. showed that video game character

preferences correlated with personality traits such as

aggression and prosociality (Delhove & Greitemeyer,

2018). They focused on the class-based first-person

shooter game Overwatch, in which 6-versus-6 players

compete against each other. In their experiment, they

used a large sample of game players (𝑁  2323) to

evaluate the relationship between players’ in-game

role preferences and personality traits. Preference for

the aggressive role (i.e., attacker) was related to

aggressive, non-prosocial personality types. On the

other hand, this was observed only in self-reported

measures, not in a small sample of objective playtime

measures.

Corem et al. showed the relationship among

player engagement, proficiency, and intrinsic

motivation on a skill-based gaming platform (Corem,

Brown, & Petralia, 2013). Their unique rating system

matched players with similar proficiency levels.

Competing with other players with a similar skill

level improved the player’s skill, which increased the

player’s satisfaction. This indicated that engagement

increased when players felt that their playing was

improving. Also, from the viewpoint of the game

system, there was a trade-off between the speed of

matching between players and the time to wait for

players with similar abilities.

4 PILOT STUDY

As a pilot study, we verified the tendency of players’

character preferences and the possibility of clustering.

We looked at the data of 38 players searching for

friends on a public Discord server. Figure 2 shows the

dendrogram of the result of clustering by the

Euclidean distance and Ward’s method, taking the

number of times a character was selected. Since there

are 4 roles of characters in VALORANT, the

maximum number of clusters was 4, and the roles are

colored differently. The yellow group was

characterized by mainly the use of Chamber and Jett,

aggressive characters fighting at the front. Both

characters have the special skill of instantly leaving

the current place. The purple group mainly used a

character named Raze who used a special skill that

took time to learn. These groups tend to be devoted to

particular characters without using other characters.

The red group, on the other hand, picked a wide range

of characters. It can be considered that the users in

this group selected characters according to other users

or selected various characters according to their

moods.

Figure 2: Dendrogram of the result of clustering character

preferences in the pilot study by the Euclidean distance and

Ward’s method. The numbers along the horizontal axis

indicate users.

5 PROPOSED METHOD

In this paper, we analyze the personalities and game

skills of players based on in-game statistical

information, and support team matching among

players in a typical online cooperative game

VALORANT. In the same way as the related work

(Delhove & Greitemeyer, 2018), we use the tendency

to select characters in the game as effective statistical

information for personality analysis. There are

currently 20 characters in the game, and they can be

roughly divided into 4 roles. In our pilot study, we

classified users into groups with 4 characteristics by

clustering analysis. In the proposed method, we

additionally use competition ranks that are certified in

the competitive mode.

Supporting Online Game Players by the Visualization of Personalities and Skills Based on in-Game Statistics

261

The proposed method is implemented as a

Discord bot application. When a user gives in-game

player ID to this application, it visualizes how they

are positioned relatively to other users. According to

the visualized information, the user interacts with

other users by considering their personalities and

competition ranks. Unlike previous research, this

application enables quantitative analysis and

visualization by the bot agent from in-game statistical

information.

5.1 In-Game Statistics

In-game statistics are retrieved using the application

programming interface (API) of Riot Games, the

developer of VALORANT. We focus on characters

as a feature of personalities and on competition ranks

as a feature of players’ game skills. The Discord bot

application used in this proposed method applies

these features to visualization. The application

records the retrieved statistics in a database and

utilizes them for later analysis.

5.2 Collecting Player Data

Information about the users of this bot application is

recorded in the database, and it is also used for the

users’ visualization. To present users’ information to

the first users, we asked several people to tell us their

in-game IDs and registered them in advance. For

visualization, the VALORANT API and the Discord

API are used to collect character preferences (a list of

the numbers of times characters were selected), ranks

of the competitive mode, VALORANT IDs, and

Discord IDs.

5.3 Clustering Character Preferences

The users are clustered hierarchically based on the

numbers of times they selected characters. We use the

Euclidean distance to measure the distance between

users and Ward’s method to measure the distance

between clusters. Based on 20 characters in the game,

each user is represented as a vector 𝒖

𝑢



,𝑢



,…,𝑢



 whose components are the numbers

of times the user selected characters.

5.4 Visualization by Dendrogram

Based on the clustering results and the users’

competition ranks, a dendrogram that expresses both

personalities and skills is generated, and is used to

support the users in forming teams. Figure 3 shows a

visualization by dendrogram. Clusters are colored by

using the threshold of 12 and the maximum number 4

of clusters. Based on related work (Delhove &

Greitemeyer, 2018), we hypothesize that closeness on

the dendrogram represents the personalities of users.

The users know that closely positioned users have

similar character preferences. It also means that

distant users have different character preferences, and

therefore they can be partners by complementing each

other’s role. In addition, the color of the username

label is based on the rank of the competitive mode,

giving the users playing the competitive mode a

visual understanding of their game skills.

Figure 3: Visualization of personalities and skills by

dendrogram.

6 IMPLEMENTATION

We implemented our Discord bot application by

allowing users to interact with others through user-

friendly slash commands (Nelly, 2021). This

application is always running on a Heroku server, and

is deployed on a public Discord server with 400 users

who cooperated in our experiment. Any member of

the server can use this bot by passing the in-game ID

of VALORANT as an argument of the slash

command on the channel where the use of this bot is

permitted by the administrator.

When the system receives the in-game ID, it

retrieves the user’s most recent 86 matches by

submitting the Production API Key to Riot Games.

We use the data of the maximum of 20 matches

excluding deathmatches. The application extracts the

characters selected by the user and the rank of the

competitive mode from the data. The data are linked

with the user ID and stored in the Firestore database

of Firebase. Finally, a dendrogram based on the

recorded character preferences and competition ranks

is generated and sent as a Discord embedded message

as shown on the left side of Figure 4. The dendrogram

is generated in real time using D3.js. The text of the

embedded message contains a description of the

visualization and a list of links to the profiles of the

users. The page view that is opened with a link differs

HUCAPP 2023 - 7th International Conference on Human Computer Interaction Theory and Applications

262

between a PC and a smartphone: on the PC, the linked

user profile is displayed on the browser as in the upper

right of Figure 4; on the smartphone, the profile of the

linked user is displayed in the application as in the

lower right of the figure. In both, the user can send

friend requests to other users.

Figure 4: Screenshot on the left shows the dendrogram and

its description sent to a user, where the links with blue

letters indicate online users. The screenshots on the right

show the different behavior of the bot application when a

link is opened on a PC or a smartphone.

7 EXPERIMENT

We conducted an experiment to evaluate the proposed

method. We first asked the administrator of an

existing public Discord server to install our bot

application. Next, we conducted two questionnaires

with Likert scales: we asked the users of the Discord

server to answer the first questionnaire immediately

after using the bot application and also to answer the

second questionnaire two or three weeks later. Our

analysis is based on the statistics of the recorded data

of the users and the results of the questionnaires.

7.1 Participants

The participants in the experiment were volunteer

players (18-year or older) who had searched for

VALORANT friends on public Discord servers. A

public server named “VALORANT Party”

cooperated in our experiments. A total of 16

participants (11 originally from this server and the

other 5 from the groups known to them) used our bot

application. The first questionnaire was answered by

11 participants, and the second questionnaire was

answered by 8 participants.

7.2 Procedure

The participants used the bot application and

answered the two questionnaires at the two or three-

week interval. The experiment was started by a slash

command (Nelly, 2021) on the channel for the bot

application that had been authorized by the server

administrator. As shown in Figure 5, the slash

command could be started from the suggestion list by

entering a slash character in the text box for text chat.

If a user entered his/her in-game ID separated into the

name and the tagline, VALORANT statistics search

using Riot API started. An edited reply showing the

progress of the search was sent at any time, and one

minute later, an embedded message with the

visualization result as an embedded image was sent.

It was left to the user whether to contact other users

from the link in the message. As soon as the results

were confirmed, the user was asked to answer the first

questionnaire.

Figure 5: Suggestions that appear when a slash character is

typed in the box at the bottom of the screen in a bot-enabled

text channel.

Two or three weeks after the first day of the

experiment, we asked the users to answer the second

questionnaire. Based on the visualization results, we

checked whether there were any changes. In addition,

we asked the members of the pre-existing teams to

evaluate each other, which was also examined by

using the statistics.

Supporting Online Game Players by the Visualization of Personalities and Skills Based on in-Game Statistics

263

7.3 Results of the Questionnaires

In the first questionnaire, we collected basic

information such as the device used for the bot

application and the experience of using public

Discord servers. 29% of the participants answered

that they had played with other users in the server.

Although 70% of the users participated in the server,

their use of the server was limited. Reasons for the

limited use include “I still don’t know how to use the

server” and “I found someone to play games with”.

There were many cases where people joined the

server but hesitated to talk to other users, or they did

not need to find someone to play games with for some

reason. Also, as shown in Table 1, 36.4% of the

participants answered that they could not understand

the visualization by dendrogram.

Table 1: Part of the result of the first questionnaire (%).

Question

Strongly

ree

Agree Neutral Disagree

Strongly

disa

ree

I can get the

meaning of the

visualization

9.1 45.5 9.1 36.4 0.0

Competition

ranks are

important

9.1 45.5 9.1 27.3 9.1

I want to contact

users via the

dis

ed lin

0.0 9.1 27.3 36.4 27.3

In the second questionnaire, we collected what the

participants thought about personalities and skills. As

shown in Table 2, 25% of the participants answered

that their thought changed by the visualization results.

On the other hand, more than half of the users

answered that they found inconvenience about

personalities and skills in both questions. The three

users who did not care about the ranks in competitive

mode were the members of the team of unrated

players. The reason was that there was no difference

in skills within the team, and that a full premade party

could play the competition mode together regardless

of the rank difference. Users’ interest in distance in

the dendrogram was split evenly. The three users who

were interested in closely positioned users adopted

characters of all roles in the last 20 matches.

7.4 Mutual Evaluation Within a

Pre-Existing Team

We present mutual evaluations and dendrograms of 6

users. Users 𝒖



to 𝒖



included in the dendrogram in

Figure 3 were the members of a team that existed

before the experiment. They played VALORANT’s

unrated mode together at least once a week. We

prepared a questionnaire with five questions shown in

Table 2: Result of the second questionnaire (%).

Question

Strongly

ree

Agree

Neutral Disagree

Strongly

disa

ree

My thought was

changed by

visualization

results

0.0 25.0 25.0 37.5 12.5

Ranks were

important in the

competitive

mode

12.5 50.0 0.0 0.0 37.5

Ranks were not

important in the

other mode

62.5 25.0 0.0 12.5 0.0

Inconvenience

was caused by

difference in

skills

25.0 12.5 25.0 0.0 37.5

I am interested

in users close to

0.0 37.5 25.0 12.5 25.0

Inconvenience

was caused by

incompatibility

ersonalities

25.0 25.0 12.5 12.5 25.0

Table 3 for the 5 users other than 𝒖



who was one of

the authors, and asked them to evaluate the others in

the team. Q1 and Q2 are about personalities, Q3 and

Q4 are about roles in the game, and Q5 is about skills.

Table 3: Questionnaire for the mutual evaluation of team

members.

Questions

Q1 My personality and this user’s personality are

distant

Q2 I often disagree with this user.

Q3 The roles that I and this user want to use are fa

Q4 The roles of mine and this user often work well

after the

ame starts.

Q5 There is a difference between my skills and this

user's skills.

Q3 and Q4 about roles had many positive answers

overall. This was because the team had already played

many times and each member had a character to

choose. For this reason, the roles in the game were

divided and effectively handled. Regarding Q5, many

users answered that they felt a difference between 𝒖



and 𝒖



, who usually played the competition mode.

Table 4 summarizes the results of Q1 and Q2

about personalities. The score for 𝒖



was particularly

high (4.25 out of 5). On the other hand, as shown in

Figure 3, the clusters were rather close to the other

users. In this way, individual cases cannot be judged

only by general character preferences. Also, 𝒖



was

located further than other members on the

dendrogram, but the similarity was not much different

from those of the other users, showing intermediate

values in the questionnaire. This was because the

users grouped into the first cluster had outlier

components that used extremely specific characters

HUCAPP 2023 - 7th International Conference on Human Computer Interaction Theory and Applications

264

(12 times out of 20). Since Ward’s method has high

classification sensitivity even for outliers,

inappropriate clustering may occur as in this case.

Table 4: Average values for questions Q1 and Q2. Larger

values indicate more different personalities. The row for

each user 𝒖



gives this user’s evaluations of the other users.

User 𝒖



, who was one of the authors, did not evaluate the

others.

User

𝒖

𝟏

𝒖

𝟐

𝒖

𝟑

𝒖

𝟒

𝒖

𝟓

𝒖

𝟔

𝒖

𝟏

- - - - - -

𝒖

𝟐

2.0 - 1.5 3.5 1.5 1.5

𝒖

𝟑

2.5 2.5 - 3.5 2.0 2.0

𝒖

𝟒

3.5 2.5 3.5 - 2.5 3.0

𝒖

𝟓

3.0 3.5 2.0 5.0 - 3.0

𝒖

𝟔

3.5 3.0 2.5 5.0 2.5 -

8 DISCUSSION

Although we used character preferences and

competition ranks as statistical information in the

game, it is also possible to use other information such

as the number of times a user played on each day of

the week and weapon preferences. An effective

indicator might be the account level of a player,

which increases as the player plays the game. It is

necessary to evaluate parameters from multiple

perspectives such as subjective evaluation and

machine learning to verify which statistical

information is useful.

We assumed that users whose personalities were

close based on their character preferences rarely

conflict while users whose personalities are distant

often conflict. We consider this from the viewpoints

of personalities and game specifications. Regarding

personalities, Lykourentzou et al. found that

personality conflicts reduced team performance while

balancing personalities significantly improved

cooperative work performance (Lykourentzou,

Antoniou, Naudet, & Dow, 2016). Regarding game

specifications, online games such as VALORANT, in

which multiple teams with multiple players compete,

are usually designed to divide roles within teams.

Since such a game currently organizes teams with

several players, it can be played without problems

even with similar roles. However, as the number of

players in a team increases, the balance might become

worse. From these two viewpoints, we can consider

recommending players with different personalities

instead of those with high similarities.

The percentage of the users who were not

comfortable with the bot application and direct

messages in Discord seems to be high. We asked 50

of the server members to cooperate in the experiment.

16 people used the bot application, 11 responded to

the first questionnaire, and 8 responded to the second

questionnaire. Although the bot application included

a link to a document containing the terms and policies

of the bot, we also used personal accounts to send

messages about cooperation in the experiment. This

was because many Discord servers including the

server that we used in our experiment prohibited bot’s

direct messages.

Half of the participants in the experiment could

not understand the dendrograms of visualization

results. Most of the participants who could

understand it had some background in computer

science. Although information visualization has

become a mainstream technology, understanding a

visualization is not always easy for the people who

see it (L'Yi, Chang, Shin, & Seo, 2019). Some of the

answers for the questionnaires showed that the

distances shown in the dendrogram were not clear to

users. It is possible to extend the representation of the

dendrogram not only with text descriptions, but also

with animations, scaling, and opacity changes.

9 CONCLUSIONS AND FUTURE

WORK

In this paper, we developed an application that

visualizes personalities and skills of players based on

the in-game statistics of the online game

VALORANT. The distances of character preferences

were close to the users’ subjective evaluations, by

which we were able to show the potential demand for

visualizing personalities and skills of users.

Our future work will promote intuitive

understanding of user relationships through scalable

and interactive information visualization, and will

support users to take the first step toward a new

experience with other users. Although currently

visualization results are embedded as images in

Discord messages, our goal is to realize interactive

visualization that runs in a browser. As the number of

users grows, the visualization should be scaled by

collapsing by clusters and filtering by ranks. We will

also improve usability by effectively using user

avatars, in-game icons, and statistical information in

pop-ups. Visualization methods need to be validated

based on construction tasks (L'Yi, Chang, Shin, &

Seo, 2019). In-game statistics and experiments will

also be used to analyze what parameters are effective

as indicators of user relationships, including win rates

and character combinations. Also, we will examine

Supporting Online Game Players by the Visualization of Personalities and Skills Based on in-Game Statistics

265

the motivations of online game players to play with

other players.

ACKNOWLEDGEMENT

We thank the administrator and the users of the public

Discord server “VALORANT Party” for their

generous cooperation in our experiment.

REFERENCES

Active Player. (2022, 11 11). Retrieved from Valorant Live

Player Counter: https://activeplayer.io/valorant/

Cap. (2022, 10 17). building-the-future-of-discord. Retrieved

from Discord BLOG -Engineering & Design:

https://discord.com/blog/building-the-future-of-discord

Cook, C., Conijn, R., Schaafsma, J., & Antheunis, M.

(2019). For Whom the Gamer Trolls: A Study of

Trolling Interactions in the Online Gaming Context.

Journal of Computer-Mediated Communication, 24(6),

293-318.

Corem, Y., Brown, N., & Petralia, J. (2013). Got Skillz?:

Player Matching, Mastery, and Engagement in Skill-

Based Games. Proc. International Conference on

Gameful Design, Research, and Applications

(Gamification), 115-118.

Delhove, M., & Greitemeyer, T. (2018). The Relationship

Between Video Game Character Preferences and

Aggressive and Prosocial Personality Traits.

Psychology of Popular Media, 9(1), 96-104.

Glen, O. (2022, 8 15). how-discord-supercharges-network-

disks-for-extreme-low-latency. Retrieved from

Discord BLOG: https://discord.com/blog/how-discord-

supercharges-network-disks-for-extreme-low-latency

Ho, S., & McLeod, D. (2008). Social-Psychological

Influences on Opinion Expression in Face-to-Face and

Computer-Mediated Communication. Communication

Research, 35(2), 190-207.

King, D., Delfabbro, P., Billieux, J., & Potenza, M. (2020).

Problematic Online Gaming and the COVID-19

Pandemic. Journal of Behavioral Addictions, 9(2), 184-

186.

Kwak, H., Blackburn, J., & Han, S. (2015). Exploring

Cyberbullying and Other Toxic Behavior in Team

Competition Online Games. Proc. ACM Conference on

Human Factors in Computing Systems (CHI), 3739-

3748.

L'Yi, S., Chang, Y., Shin, D., & Seo, J. (2019). Toward

Understanding Representation Methods in

Visualization Recommendations through Scatterplot

Construction Tasks. Computer Graphics Forum, 38(3),

201-211.

Lykourentzou, I., Antoniou, A., Naudet, Y., & Dow, S. P.

(2016). Personality Matters: Balancing for Personality

Types Leads to Better Outcomes for Crowd Teams.

Proc. ACM Conference on Computer-Supported

Cooperative Work & Social Computing (CSCW), 260-

273.

Nelly. (2021, 3 24). slash-commands-are-here. Retrieved

from Discord BLOG: https://discord.com/blog/slash-

commands-are-here.

HUCAPP 2023 - 7th International Conference on Human Computer Interaction Theory and Applications

266