RoleSeer Understanding Informal Social Role Changes in MMORPGs via Visual Analytics

2025-05-03 0 0 6.1MB 17 页 10玖币

侵权投诉

RoleSeer: Understanding Informal Social Role Changes in

MMORPGs via Visual Analytics

Laixin Xie

School of Information Science and

Technology, ShanghaiTech University

Shanghai, China

xielx@shanghaitech.edu.cn

Ziming Wu

Interactive Entertainment Group,

Tencent Inc.

Shenzhen, China

zwual@connect.ust.hk

Peng Xu

User Experience Center, NetEase Inc.

Hangzhou, China

alexkdd@163.com

Wei Li

College of Creative Design, Shenzhen

Technology University

Shenzhen, China

helloweili@icloud.com

Xiaojuan Ma

Department of Computer Science and

Engineering, The Hong Kong

University of Science and Technology

Hong Kong, China

mxj@cse.ust.hk

Quan Li∗

School of Information Science and

Technology, ShanghaiTech University

Shanghai, China

liquan@shanghaitech.edu.cn

ABSTRACT

Massively multiplayer online role-playing games create virtual com-

munities that support heterogeneous “social roles” determined by

gameplay interaction behaviors under a specic social context. For

all social roles, formal roles are pre-dened, obvious, and explicitly

ascribed to the people holding the roles, whereas informal roles

are not well-dened and unspoken. Identifying the informal roles

and understanding their subtle changes are critical to designing

sociability mechanisms. However, it is nontrivial to understand the

existence and evolution of such roles due to their loosely dened,

interconvertible, and dynamic characteristics. We propose a visual

analytics system, RoleSeer, to investigate informal roles from the

perspectives of behavioral interactions and depict their dynamic

interconversions and transitions. Two cases, experts’ feedback, and

a user study suggest that RoleSeer helps interpret the identied

informal roles and explore the patterns behind role changes. We

see our approach’s potential in investigating informal roles in a

broader range of social games.

CCS CONCEPTS

•Human-centered computing →Human computer interac-

tion (HCI);Visualization; User studies.

KEYWORDS

social role, social network, graph embedding, gameplay

ACM Reference Format:

Laixin Xie, Ziming Wu, Peng Xu, Wei Li, Xiaojuan Ma, and Quan Li. 2022.

RoleSeer: Understanding Informal Social Role Changes in MMORPGs via

∗The corresponding author.

Permission to make digital or hard copies of all or part of this work for personal or

classroom use is granted without fee provided that copies are not made or distributed

for prot or commercial advantage and that copies bear this notice and the full citation

on the rst page. Copyrights for components of this work owned by others than ACM

must be honored. Abstracting with credit is permitted. To copy otherwise, or republish,

to post on servers or to redistribute to lists, requires prior specic permission and/or a

fee. Request permissions from permissions@acm.org.

CHI ’22, April 29-May 5, 2022, New Orleans, LA, USA

ACM ISBN 978-1-4503-9157-3/22/04. . . $15.00

https://doi.org/10.1145/3491102.3517712

Visual Analytics. In CHI Conference on Human Factors in Computing Systems

(CHI ’22), April 29-May 5, 2022, New Orleans, LA, USA. ACM, New York, NY,

USA, 17 pages. https://doi.org/10.1145/3491102.3517712

1 INTRODUCTION

Like in real-life communities, heterogeneous social dynamics are

also observed in virtual communities. Massively multiplayer online

role-playing games (MMORPGs) are essentially virtual communities

in the form of role-playing. As the gameplay takes place, members of

the community adopt dierent social roles (i.e., “a set of expectations

and norms that dene how people playing the roles should and would

behave in a social situation” [

]). Some roles are public and

ocial. They can be easily observed from the given positions or

titles, often labeled as warriors,mages,hunters in games like World

of Warcraft (WoW). As these roles are evidently pronounced by the

game mechanism and independent from the individual behavior

patterns, they are often recognized as formal roles [

]. However,

compared to the formal roles, some roles are less evident but can

be equally important. For instance, a member’s social status is

primarily inuenced by its contribution to the community. When

an old member helps a newcomer, it adds to the eort of building a

healthy community culture – Altruism behaviors like such fortify

the sense of belonging, which will eventually benet the community

as an entirety. Alternatively, players may also only take periphery

positions and play as bystanders. The dierence of these acts may

marginally aect the formal roles but subtly shift the community’s

social structure. Based on players’ respective inuence upon the

social attributes, roles such as dungeon leader,ghter,isolate,social

butteries, and information giver are naturally assigned based on

their explicit play behaviors [

]. Following social studies [

], we use informal roles

to represent the later, less evident roles,

which are naturally determined by players’ gameplay behaviors.

Characterizing the collective conguration of informal roles

in a community and understanding the existence and changes of

roles within are essential ways to study MMORPG communities.

First, informal roles are greatly inuenced by existing formal roles.

They will often replace or supplement formal social roles [

https://eagle.northwestu.edu/faculty/gary-gillespie/roles-and-small-group-

communication/

arXiv:2210.10698v1 [cs.HC] 19 Oct 2022

CHI ’22, April 29-May 5, 2022, New Orleans, LA, USA Xie, L., Wu, Z., Xu, P., Li, W., Ma, X., Li, Q.

which can provide insights into the sociability design of gameplay

mechanisms that can facilitate the formation of specic informal

roles [

]. Second, analyzing how individuals’ informal roles vary

regarding their behaviors can promote the long-term operations

of a virtual community, such as drawing in lapsed gamers and

encouraging, e.g., “isolate” players to change their roles.

However, identifying informal roles and understanding the rea-

sons behind their existence and changes is nontrivial due to several

issues:

(1) Loose denition.

Existing literature has attempted to

provide descriptions of informal roles in MMORPGs. For example,

an early study on text-based multi-user dungeons (often referred

to as the predecessor of MMORPGs) categorized the typology of

roles into four groups [

]: killers who like to annoy other players;

socializers who want causal social interaction with other players;

achievers who aim to master the game, and explorers who enjoy

exploring the game world. Although they provide grounded the-

ories for later research on social roles in the context of graphical

MMORPGs, the current denitions of the limited set of informal

roles are primarily based on prior knowledge of game designers and

analysts. Consequently, existing methods for deriving meaningful

roles or patterns using pure statistical analysis approaches are insuf-

cient without a delineated denition of informal roles, especially

for those unspoken ones, making automated solutions of informal

roles mining challenging to achieve.

(2) Dynamic interconver-

sion and evolution.

Informal roles are based on human needs

for trust, support, resource sharing irrespective of whether they

operate in real life or a virtual setting [

]. Specically, these roles

result from social interaction and negotiation between the actor

and those they interact with. They are dynamically developing and

interconvertible over time, i.e., players may switch from one role to

another. Given such characteristics of informal roles, summarizing

and understanding the temporal properties of the dynamic inter-

conversion and evolution of informal roles in an MMORPG virtual

community can help reveal players’ exploration patterns and shed

light upon the eective socialization mechanism of the players from

a global perspective.

(3) Diverse paths behind role formations.

Diverse steps of social exploration (i.e., social self-discovery help-

ing an individual understand their own social needs [28]) precede

forming a particular informal role. For example, players may need

to balance their personal and social time or choose between large

but weak social ties and a few strong social ties when interacting

with other virtual companions in the game [

]. Although a

small number of players with abundant external resources (e.g.,

time and money) could take on specic informal roles through the

spending of these resources, the large majority of the members in

the game community may not have sucient external resources at

their disposal. Instead, players adopt some informal roles through

gameplay interactions. Inspecting the diverse paths behind the for-

mations of such informal roles is more important because they

can explain how people interact, collaborate, and work together to

cultivate community building and growth. It can also increase the

participant’s awareness of their social interaction [50].

Most previous work tends to study the social role from “who the

users are” and then “what the users do” rather than behavioral in-

teractions between them in the virtual community, or they assumed

a prior knowledge about the social roles (e.g., focusing on the con-

tents of the players’ posting or activities) [

]. Although such

studies have been helpful, we argue that focusing on behavioral

interactions would cast new light on social roles in virtual worlds.

Thus, to ll the gap, we propose RoleSeer, an interactive visual ana-

lytics system that helps game designers and game user experience

(UX) practitioners understand informal roles and paths behind their

formations and dynamic changes in the context of an MMORPG.

To give detailed elaborations on the proposed idea and follow-up

implementations, we organize this work as follows. We rst ob-

serve our collaborating game team’s current gaming social role

analysis practices and identify their primary needs and concerns.

Then, we adapt a dynamic network embedding and alignment ap-

proach to the friendship network in the specied MMORPG for

facilitating potential informal role detection across multiple time

snapshots. Dierent clusters of potential informal roles and their

interconversion and evolution across these clusters are discovered

by projecting the resulting embeddings onto a low-dimensional

space. We further support the experts to explore players’ diverse

behavioral interactions that lead to their role changes. Based on

these objectives, we develop a visual analytics system to support

ne-grained informal role analysis at the overview, role cluster,

and individual levels. Lastly, we present several case studies and

interview feedback with domain experts to evaluate the ecacy of

our system. We outline the contributions of this work as follows.

•

We shadow domain experts’ daily working processes and

conduct interviews to get insight into their current practice

in understanding informal role changes.

•

We identify the potential informal roles from the perspec-

tive of behavioral interaction analysis through an adapted

dynamic network embedding and alignment model in the

context of an MMORPG.

•

We depict the interconversion and evolution of informal

roles across dierent time snapshots and explore the patterns

behind the role changes via a visual analytics system.

2 RELATED WORK

Literature that overlaps this work can be categorized into three

groups: social network analysis in MMORPGs,graph latent represen-

tations, and dynamic graph visualization.

2.1 Social Network Analysis in MMORPGs

Studies that focus on social network analysis in MMORPGs sug-

gest that players’ social behaviors and interactions considerably

inuence players’ gaming experience [

]. For exam-

ple, Szell and Thurner [

] studied the structure of friend, enemy,

and communication networks and identied that friend and enemy

networks are topologically dierent. Ducheneaut et al. [

] and

Chen et al. [

] used traditional metrics in social network analysis

such as density and centrality to analyze the properties of player

guilds in WoW. Lu et al. [

] proposed BeXplorer to explore the

dynamic interplay among multiple types of behaviors. Li et al. [

]

investigated the evolution of egocentric players and focused on

the relationship between a player (ego) and his/her directly linked

friends (alters). They also inferred how changes in an ego’s inter-

active behaviors might propagate through the friendship network.

However, their work captured the evolutionary pattern based on

ego networks at a microscopic level, a case-by-case analysis. Our

RoleSeer: Understanding Informal Social Role Changes in MMORPGs via Visual Analytics CHI ’22, April 29-May 5, 2022, New Orleans, LA, USA

work identies representative informal roles from the perspectives

of structural analysis at a global level. Furthermore, we study the

underlying patterns that drive the interconversion and transition

of informal roles held by players across multiple time snapshots.

Apart from studying the overall social relationships within a

virtual community, other studies attempted to investigate play-

ers’ social status or social roles such as leader,core members, and

newcomers [

]. For example, Williams et al. [

] conducted inter-

views with hardcore players and demonstrated the importance of

leaders and critical structural positions within a virtual commu-

nity. Canossa et al. [

] applied standard social network features to

identify “inuencers” to an online multiplayer shooter game. The

result shows that network feature-dened inuencers had an out-

sized impact on the playtime and social play of players joining their

in-game network. These studies shed light upon the social role anal-

ysis in the game community. However, the previous social roles are

pre-dened in an initial set of categories such as “inuencer” and

“hardcore player” based on experts’ prior knowledge or manually

categorized using thematic and content analysis [

], which cannot

identify more unspoken informal roles. Ang et al. [

] observed

the “chat interaction” to determine the structural characteristics of

three social roles of a guild community in WoW : densely connected

core members, loosely connected semi-periphery members, and

disconnected periphery players. However, they only focused on a

small guild in one static snapshot and verbal interactions among

players. In our work, we identify informal roles based on behav-

ioral interactions. We study the players’ non-verbal interaction

and explain the interconversion and transitions of informal roles.

Ducheneaut et al. [

] analyzed the prevalence and extent of social

activities to investigate whether (and how) a game’s “social factor”

can contribute to the game’s success. They found that playing the

game is like being “alone together,” i.e., surrounded by others but

not necessarily actively interacting with them. Their study suggests

designs for online games in which encouraging and supporting di-

rect interactions might be “less important” than designing for the

“spectator experience” and a sense of social presence, indicating dif-

ferent social interactions and roles may be equally important to the

game’s success. Motivated by their work, we study informal roles’

existence and dynamic transitions for a better user experience.

2.2 Graph Latent Representations

Graph latent representations have demonstrated their practical-

ity in many graph analysis applications and downstream machine

learning tasks such as node classication, clustering, and link pre-

diction [

]. Perozzi et al. [

] developed a random walk-based

method, DeepWalk, to learn latent representations for nodes by

generalizing neural language models to preserve the higher-order

proximity between nodes. Node2Vec [

] provides a trade-o be-

tween breadth-rst search (BFS) and depth-rst search (DFS) in the

random walk process. Van den Elzen et al. [

] derived snapshot

representations from adjacency matrices and then projected the

snapshots onto a 2D space for network state discovery. Dal Col et

al. [

] used the coecients of graph wavelets to represent nodes

for evolutionary local change discovery. Xu et al. [

] exploited

diachronic node embeddings (DNE) for better preserving structural

and temporal properties. They further designed a system for an in-

teractive exploration of dynamic networks. Particularly, its backend

model rst extracts node embeddings of dierent timestamps and

then applies embedding alignment, which shows strong capability

in capturing local information, i.e., the similarity among players.

Eren et al. [

] displayed the similarity and dierence of temporally

dynamic graphs by sorting the graph embedding vectors. Our ap-

proach is dierent from the above works on two dimensions. First,

they applied DeepWalk to the network at each time snapshot and ini-

tialized the embedding results to preserve the temporal community,

while we replaced the underlying DeepWalk with Struc2Vec [

] for

baseline model comparison. The adapted version can maintain the

structural similarity that captures the “roles" of nodes in the graph.

Second, we focus on the interaction patterns that explain informal

roles’ existence, interconversion, and transitions.

2.3 Dynamic Graph Visualization

Dynamic graph visualization has been attracting researchers’ at-

tention for a long time. Two primary visualization methods have

been proposed: animation and timeline-based methods. Particularly,

the former method simulates the graph dynamic evolution by re-

drawing the graph at each step, visualizing it with a node-link

diagram [

], showing its structural characteristics. However, anima-

tion techniques are desirable to reduce the complexity of dynamic

graphs to facilitate visualization; however, they are inadequate to

support detailed network analysis and interpretation of tempo-

ral properties [

]. Meanwhile, users must maintain a mental map

in each snapshot for comparison [

]. Instead of animating the

network as a sequence, timeline-based techniques directly draw

the network in a static image at each time step along a timeline,

providing a better temporal overview of the dynamic graph [

Small multiples such as matrices [

] and node-link diagrams [

]

are employed to study position changes of nodes within the dy-

namic graph [

]. Although these methods can capture temporal

structural changes, human cognition often limits the discovery

of temporal patterns. Furthermore, when analyzing an extensive

network, one may encounter scalability diculties when space

is limited [

]. In this work, we combine timeline visualiza-

tion with our graph latent representations. Precisely, the latent

graph representation captures the structural similarity and dynam-

ics within a short period (e.g., 6hours for one step). The timeline

aligns the corresponding informal roles with visualizing the overall

trend across the entire time steps. Thus, we could accurately cap-

ture the dynamics since we analyze the graph in a much shorter

period while still maintaining an overview of the overall changes,

achieving a balance of detail and abstraction.

3 OBSERVATIONAL STUDY

This section presents the background information of the showcased

game and the game team we collaborate with. Then, we shadow the

experts’ daily working process and summarize their conventional

practice and bottlenecks. Finally, we distill the design requirements

based on the results of expert interviews.

CHI ’22, April 29-May 5, 2022, New Orleans, LA, USA Xie, L., Wu, Z., Xu, P., Li, W., Ma, X., Li, Q.

Table 1: Event and intimacy: 𝑋indicates varying intimacy values of dierent events. 𝑁shows no upper bound.

killing monster killing player task Using props ghting chatting carbon battle

each increase 1 1 X X 25 2 X 120

weekly bound 1500 300 100 5000 350 210 N N

3.1 About the Game and the Team

The game studied in this work is a typical MMORPG that provides

various systems to facilitate social interactions among players, such

as text/voice chatting and in-game activities. Among all the systems,

the friendship system is the most popular one, allowing players to

nish specic tasks, e.g., adding others as friends, seeking help, or

nding teammates for challenging dungeons. Typically, friendship

closeness is measured by the “intimacy value” between two players,

tion between two players. Dierent interactions such as chatting,

killing monsters, and battles among players contribute to varying

intimacy values (Table 1).

Figure 1: The impact of the number of players’ social rela-

tions established in the rst week on the players’ retention

rate in the next week: x-axis indicates the number of social

ties found in the rst week. The top half y-axis shows the

retention rate of players in the next week, and the bottom

half y-axis is the number of the involved players. Three red

rectangles indicate the retention dierences between play-

ers with

and

social relation in terms of three intimacy

thresholds (i.e., dierent values of intimacy).

We collaborated with a team of experts from an internet game

company, including one user experience (UX) engineer (E1, female,

age: 28), one data analyst (E2, male, age: 29), and two game designers

(E3, male, age: 29, E4, male, age: 30), to study players’ socialization

in the MMORPG as mentioned above. All experts have been in

the game industry for more than ve years, and they have been

involved in this specic MMORPG since its preliminary inception.

One of the primary responsibilities of the game team is constantly

improving the game’s retention rate. According to a preliminary

analysis of the game team (Figure 1), players with certain social

relations have a 20% higher retention rate than those without social

ties (indicated by the three red rectangles). Meanwhile, specic

turning points can be witnessed. For example, players with 5–

7social relations have a considerably higher retention rate than

those with fewer social ties. The experts commented that social

ties could help improve players’ retention, thus considering 5–7

social relations the target value of players’ socialization guidance

in the early gaming stage. This initial observation also motivates

our work. The game team wanted to observe how players establish

social relations and get involved in the community, so they could

design better social mechanisms to support players’ socialization.

3.2 Experts’ Conventional Practice and

Bottlenecks

To obtain detailed information of the game experts’ current prac-

tice, with consent, we shadowed the team’s daily working process,

including videotaping how they observed players experiencing the

game, conducting testing experiments, and on-site interviews with

the players. Later, we carried out a retrospective analysis with the

game team on their conventional practices. Particularly, to tackle

the issue mentioned above, the game team attempted to use a node-

link diagram to visualize how players establish their social relations

over time. They found that at the beginning of the game, three or

four players may set up a small group, and the connections within

the group become close. Then, the small group begins to establish

social contact with external players, followed by integration with

another group. Finally, the newly formed group integrates into a

larger one. The experts identied that most players (approximately

60%) follow a similar procedure in-game socialization, and their

topological positions in the entire social network are quite dierent.

For example, in Figure 2, red dots indicate players entering the

game after the rst day of game release (marked as PAs), and black

dots indicate players entering the game on the rst day (marked as

PBs). E1 compared the top three largest communities and drew the

following three conclusions. First, in the rst network (Figure 2 (a)),

the proportion of PAs is higher compared with that in the other

two networks (Figure 2 (bc)). Second, most PAs are in the central

areas of the network (Figure 2 (a)). Third, PAs establish more social

relations in Figure 2 (a), i.e., red dots with only one social tie are few.

Many nodes have been integrated into groups or have established

multiple social relations. In Figure 2 (bc), PAs are mainly distributed

in the peripheral areas with few social relations.

Figure 2: The force-directed layout of the social networks in

three virtual communities. (a) PAs are well integrated into

the community. (b – c) PAs are mainly distributed in the pe-

ripheral areas and have fewer social relations.

Although the experts obtained some initial insights into how

players’ socialization aects the integration into the community,

they encountered the following issues when trying to make sense of

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

RoleSeer:UnderstandingInformalSocialRoleChangesinMMORPGsviaVisualAnalyticsLaixinXieSchoolofInformationScienceandTechnology,ShanghaiTechUniversityShanghai,Chinaxielx@shanghaitech.edu.cnZimingWuInteractiveEntertainmentGroup,TencentInc.Shenzhen,Chinazwual@connect.ust.hkPengXuUserExperienceCenter,NetEas...

展开>> 收起<<

RoleSeer Understanding Informal Social Role Changes in MMORPGs via Visual Analytics.pdf

共17页,预览4页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

RoleSeer Understanding Informal Social Role Changes in MMORPGs via Visual Analytics

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: