Recommendation Algorithm in TikTok: Strengths, Dilemmas, and Possible Directions

Abstract

Recommendation algorithms are reshaping the ecology of digital video-sharing platforms and users' media usage behaviors. TikTok's recommender system is widely considered to be an outstanding representative among them. Although a large amount of research has been conducted in relation to TikTok, most of these studies pay attention to content analysis, platform features study, user behavior examination and technical aspects of platform algorithm. However, there is markedly less research into TikTok’s recommendation algorithm as well as relevant theoretical and empirical support for this. Based on a slightly simplified variant of the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) guidelines (Page et al., 2021), this paper reviews the literature on the use of recommendation algorithm in TikTok, aiming to serve as a brief primer to answer the strengths and dilemmas of the adoption of recommendation algorithm on the TikTok platform, and to propose possible directions for short-form mobile video platforms.

Full text

Recom
m
1
Centre for I
n
Corresponde
n
7AL, United
K
Received: Jul
y
doi:10.11114/
i
Abstract
Recommenda
t
behaviors. Ti
k
Although a l
a
content anal
y
However, the
r
empirical su
p
Systematic R
e
recommendat
i
adoption of
r
mobile video
p
Keywords: T
i
1. Introducti
o
As a short-f
o
p
ersonalized
v
achieved gre
a
Facebook as
COVID-19,
T
b
y download
s
those incumb
e
So far, most
Zhang, Ma,
&
in various do
m
and communi
Omar & Deq
u
algorithm.
Given the ab
o
the area of re
c
strives to clar
What challen
g
Here comes t
h
TikTok, and
recommendat
i
short-form m
o
2. Methodol
o
The systemat
i
m
endatio
n
n
terdisciplinar
y
n
ce: Pengda
W
K
ingdo
m
.
y
31, 2022
i
jsss.v10i5.56
6
t
ion algorith
m
k
Tok's reco
m
a
rge amount o
f
y
sis, platform
r
e is markedl
y
p
port for this.
e
views and
M
i
on algorithm
r
ecom
m
endati
o
p
latforms.
i
kTok, recom
m
o
n
o
rm mobile
v
video feeds
d
a
t popularity
the most do
w
T
ikTok has gr
o
s
in Q4 2021
(
e
nts in the int
e
TikTok resea
r
&
Evans, 2021
;
m
ains (Miao,
H
cation. Many
u
an, 2020).
H
o
ve literature
g
c
ommendatio
n
r
ify and answ
e
g
es and revol
u
h
e structure o
f
the next sec
t
i
on algorith
m
o
bile video pl
a
ogy
i
c literature re
v
n
Algorit
h
y
Methodolog
i
W
ang, Centre
Accepte
d
6
4
m
s are reshap
i
m
mender syste
m
f
research has
features stud
y
y
less researc
h
Based on a
M
eta-Analysis)
in TikTok, ai
m
o
n algorithm
m
endation alg
o
v
ideo platfor
m
d
riven by its
r
in quite a s
h
w
nloaded app
o
wn to over 1
(
Sensor Towe
r
e
rnational digi
t
r
ch has focus
e
;
Sun, H. Zha
n
H
uang, & Hua
n
scholars have
H
owever, only
g
ap, this essa
y
n
algorithm in
e
r the followi
n
u
tions such alg
o
f
this essay: i
t
t
ion focuses
o
m
. What outli
n
a
tforms.
v
iew was con
d
h
m in Ti
k
Di
Pe
n
i
es, Universit
y
for Interdisci
p
d
: September
2
URL: https://
d
i
ng the ecolo
g
m
is widely
been conduc
t
y
, user beha
v
h
into TikTok’
s
slightly simp
guidelines (P
a
m
ing to serve
on the TikTo
k
o
rithm, recom
m
m
, TikTok is
r
ecommendat
i
h
ort time, esp
in the Unit
e
b
illion month
l
r
, 2022). Fro
m
t
al platform m
a
e
d on the co
n
n
g, S. Zhang,
&
n
g, 2021; Vij
a
also observed
limited paper
y
focuses on t
h
TikTok as w
e
n
g questions:
W
o
rithm mecha
n
t
first identifi
e
o
n the existi
n
n
ed in the la
s
d
ucted based
o
60
k
Tok: Str
e
i
rections
n
gda Wang
1
y
of Warwic
k
,
p
linary Metho
2
, 2022
d
oi.org/10.111
g
y of digital
considered t
o
t
ed in relation
v
ior examinati
s
recommend
a
lified variant
a
ge et al., 20
2
as a brief pr
i
k
platform, a
n
m
ender syste
m
characterize
d
i
on algorithm
p
ecially for it
s
e
d States (Vij
a
l
y active user
s
m
this point,
T
arket (Gray, 2
0
n
ten
t
-
b
ased an
a
&
Luo, 2020)
a
y & Gekker,
2
user motivati
o
s have paid p
h
e discussion
e
ll as its twin
s
W
hy is the re
c
n
ism may pos
e
e
s the outstan
d
n
g and potent
i
s
t part are th
e
o
n the guideli
n
Intern
I
e
ngths,
D
Coventry, Un
i
dologies, Uni
v
Available on
l
14/ijsss.v10i5
.
video-sharin
g
o
be an outst
a
to TikTok,
m
on and techn
a
tion algorith
m
of the PRIS
M
2
1), this paper
i
mer to answe
n
d to propos
e
m
, artificial int
e
d
by conten
t
-
(Klug, Qin,
E
s
excellent d
o
a
y & Gekker
,
s
(TikTok Tea
m
T
ikTok indeed
0
21).
a
lysis (Bandy
of this emerg
i
2
021; Zhu, 20
2
o
n and behavi
articular atte
n
of strengths,
d
s
ervice for th
e
c
ommendation
e
to digital so
c
d
ing advantag
e
i
al harm and
e
future patt
e
n
es suggested
i
n
ational Journal
o
Vol. 10,
I
SSN 2324-80
3
Published
URL:
D
ilemma
s
i
ted Kingdom
v
ersity of Wa
r
l
ine: Septemb
e
.
5664
g
platforms a
n
a
nding repres
e
m
ost of these s
t
ical aspects
o
m
as well as r
e
M
A (Preferre
d
reviews the l
r the strength
s
e
possible dir
e
e
lligence, digi
t
b
ased interac
t
E
vans, & Ka
u
o
wnloading p
e
,
2021). Muc
h
m
, 2021) and
r
provides sig
n
& Diakopou
l
i
ng platform,
a
2
0), such as p
o
or for TikTok
n
tion to the Ti
k
d
ilemmas, an
d
e
Chinese mar
k
algorithm so
c
iety?
e
s of recomm
e
challenges b
e
rns and the
p
i
n the PRISM
A
o
f Social Scien
c
No. 5; Septe
m
3
3 E-ISSN 2
by Redfame
P
http://ijsss.red
f
s
, and Po
s
r
wick, Coven
t
e
r 9, 2022
n
d users' me
d
e
ntative amo
n
t
udies pay att
e
o
f platform a
l
e
levant theore
t
d
Reporting I
t
iterature on t
h
s
and dilemm
a
e
ctions for s
h
t
al platfor
m
tions that de
p
u
fman, 2021)
.
e
rformance,
u
h
accelerated
r
anked 2nd w
o
n
ificant comp
e
l
os, 2020; Ch
e
a
nd its potenti
a
o
litics, econo
m
use (Klug et
a
k
Tok recom
m
d
possible dire
k
et, namely D
important for
e
ndation algo
r
rought by th
e
p
ossible dire
c
A
(Preferred
R
c
e Studies
m
be
r
2022
3
24-8041
P
ublishing
f
ame.co
m
s
sible
t
ry, CV4
i
a usage
n
g them.
e
ntion to
gorithm.
t
ical and
t
e
m
s for
h
e use of
a
s of the
h
or
t
-form
p
end on
TikTo
k
u
nseating
through
o
rldwide
e
tition to
e
n, Min,
a
l power
m
y, medi a,
a
l., 2021;
endation
c
tions in
o
uyin. It
TikTok?
r
ithm for
e
use of
c
tions of
R
eporting

International J
o
Items for Sy
s
graphically d
e
Taylor & Fra
n
After removi
n
Reference sc
r
keywords “Ti
k
ByteDance c
o
could compri
s
3. Si
g
nifican
c
The analysis
2016). With
d
recommende
d
massive user
adaptability o
f
3.1 Content
D
From the per
s
effective con
t
content drasti
TikTok is re
p
p
ersonalized
v
Recommenda
t
recommender
of user data l
hierarchical i
n
2021).
What’s more,
(Zhao, 2021)
.
learning for f
e
o
urnal of Social
s
tematic Revi
e
e
picted in Fi
g
n
cis, and CN
K
n
g duplicates
,
r
eening was a
l
k
To
k
, Douyin
,
o
mpany and i
n
s
e a total of 4
8
Figu
r
c
e of the Use
o
of a
p
latform
d
irect networ
k
d
videos that a
r
data availab
f
the tailored
v
D
istribution
s
pective of co
m
t
ent distributi
o
cally (X. Lu,
p
rograming us
v
ideo feed wit
h
t
ion recallin
g
systems. The
ibrary, while
t
n
terest label s
e
the multiple
.
For exampl
e
e
ature learnin
g
Science Studie
s
e
ws and Met
a
g
ure 1. Indivi
d
K
I. Only texts
i
,
the screeni
n
l
so used to r
e
,
and recomm
e
n
ternational m
e
8
studies and 1
r
e 1. Literatur
e
o
f Recommen
case starts w
i
k
s effects, Ti
k
r
e 15 seconds
ility, making
v
ideo feed.
m
munication t
h
o
n strategies.
V
Z. Lu, & Liu
,
with the alg
o
h
out initiative
,
g
and recom
m
former requir
e
t
he latter is e
x
e
t, which can
c
machine lear
n
e
, DeepFM, a
g
, gains great
s
a
-Analysis) 2
0
d
ual databases
i
n English or
C
n
g process in
c
e
duce the risk
e
ndation algor
i
e
dia) were als
1 reports.
e
selection wo
n
dation Al
g
or
i
ith determini
n
k
Tok is an al
or less in dur
a
TikTok eligi
b
h
eory, the AI
r
V
ideo sharing
,
2020). As a
o
rithm techno
l
,
reducing the
m
endation ra
n
e
s the algorith
m
x
pected to det
e
c
learly show t
h
n
ing algorith
m
model that c
o
performance
f
61
0
20 statemen
t
searched inc
l
C
hinese that h
a
c
luding attrib
u
of overlooki
n
i
th
m
”. In addi
t
o included in
rking flow ba
s
i
thm for Tik
T
n
g the platfor
m
gorith
m
-drive
n
a
tion. The pot
e
b
le to enhan
c
r
ecommendati
o
platforms ha
v
new interfac
e
l
ogy. It is a f
u
time cost for i
n
n
king are tw
o
m
to extract a
ermine the or
d
h
e inclusion a
n
m
s enable bett
e
o
mbines fact
o
f
or high- and
l
t
(Page et al.,
l
uded: Web o
f
a
d been publi
s
u
te, title, abs
t
n
g important
l
t
ion, several
w
the literatu
r
e
s
s
ed on the PR
I
T
o
k
m
characterist
i
n
, conten
t
-ori
e
e
ntial of reco
m
c
e the effecti
v
o
n algorithm i
s
v
e transforme
d
e
culture, also
u
rther benefit
n
formation se
a
o
main matt
e
fraction of co
n
d
er of recom
m
n
d hierarchica
l
e
r feature inte
r
o
rization mac
h
l
ow-order fea
t
2021). The
a
f
Science, Sc
o
s
hed online af
t
t
ract, and ful
l
l
iterature that
w
ebsite report
s
s
election. The
I
SM
A
method
i
cs (
N
ooren,
v
e
nted produc
t
m
mendation al
g
v
eness of co
n
s
of crucial si
g
d
the way that
a new digital
that TikTok
c
a
rching to a l
a
e
rs waiting f
o
n
tent f
r
om th
o
m
endation list.
l
relationship
b
r
action disclo
s
h
ines for reco
m
t
ure interactio
n
Vol. 10, N
o
a
dopted fram
e
o
pus, Elsevie
r
t
er 2016 were
l
text was c
o
was not foun
d
s
(e.g., reports
final literatu
r
v
an Gorp, &
v
t
(Zhao, 2021
g
orithm is gro
u
n
tent distribu
t
g
nificance for
T
users consu
m
medium (Ha
n
c
an make use
r
a
rge extent.
o
r addressing
o
se tremendou
s
TikTok cons
t
b
etween the d
a
s
ure of user
b
m
mendation
a
n
s without the
o
. 5; 2022
e
work is
,
SAGE,
selected.
nducte
d
.
d
by the
from the
e
review
v
an Eij
k
,
)
, where
u
nded in
t
ion and
T
ikTok’s
m
e online
n
, 2017),
r
s accept
by the
s
amount
t
ructed a
a
ta (Zhao,
ehaviors
a
nd deep
premise

International Journal of Social Science Studies Vol. 10, No. 5; 2022
62
of pre-training (Guo, Tang, Ye, Li, & He, 2017).
3.2 User Resonance Exploration
Another strength of recommendation algorithms can be detected from the viewpoint of Human-Computer Interaction
(HCI) theory and Data Pool Marketing theory: its user-centered exploration power.
TikTok’s capability to read its users with the powerful recommendation algorithms contributes a lot to engage users.
Preference, personality, location, environment, user interests are basic variables that the AI will use to evaluate users.
Most importantly, the two systems of Natural Language Processing and Computer Vision Technology are key
techniques to help determine the success of a video (Scanlon, 2020).
The method of partitioned data buckets multilayer screening (Zhao, 2021) is used by TikTok to evaluate the content
with most audience resonance. An illustrative example is the algorithmically curated “For You” page, where a stream of
uniquely tailored videos will be delivered based on users’ past respond behavior. The mixed use of stream computing
and batch computing largely improve the time-effectiveness due to the dynamically labels updated by algorithm.
According to the “For You” algorithm, most viral videos of this page share the signal of “positive feedback loop”
(Matsakis, 2020), featuring amounts of likes and views. Furthermore, scholars claimed that TikTok’s recommendation
algorithms can not only accurately recommend videos of interest to users, but also assist them in expanding into new
intersecting areas (Zhang & Liu, 2021).
There also has been a surge of interest in understanding and improving users’ actual experiences with the app. Through
a 2×2 between-subject experimental design and subsequent data analyses, Wang (2020) suggested the potentials of
TikTok videos to persuade new technology-adoption as well as to facilitate emotional resonance by creating a sense of
Immersion, Social Presence, and Entertainment.
4. Dilemmas Caused by the Use of Recommendation Algorithm
The recommendation algorithm also has profound and subtle potential influence on all walks of our society because of
the dissemination of audio-visual information based on artificial intelligence technology. Nevertheless, there is ample
evidence from existing literature that some dilemmas, such as algorithmic transparency, digital addiction, and
information diversity, are all still unsolved issues.
4.1 Transparency, Explainability and Misattribution
The transparency and explainability of the video flows recommendation process are of vital importance. TikTok has
provided some basics about its recommender system on its official newsroom website. It simply explains several factors
that contribute the For You feed: user interactions, video information, device and account settings. On top of that, quick
guidance is also given to help users build their personalized For You feed (TikTok Team, 2020). But it’s still too general
and time-consuming for people to understand. The mathematical formula put forward by communication scholar Wilbur
Schramm (West & Turner, 2013), can help us explain people’s choice of media:
possible rewards ÷ effort required = probability of choice (1)
We can tell from the formula that the higher likelihood of meeting the need, the less effort spent, the more likely users
choose a certain channel of information. Therefore, what users need are concise innovative and interactive explanations
presented in the process while people are using the app.
TikTok also engenders a culture of misattribution. More specifically, popular formats, audio clips and even licensed
music are freely reused without any connection to the original source with impunity (Valdovinos Kaye, Rodriguez, &
Wikstrom, 2020). For example, researchers from the University of Technology Sydney (Meese & Hagedorn, 2019)
found that social media users are not actively engaging with the technicalities of copyright law even if they are aware of
its concepts. In addition, unlike online creative communities (Perkel, 2011), centralized community discussions on
authorship issues are blank on social media platforms.
4.2 Digital Addiction, Digital Dementia
Researchers have begun to investigate those social problems stemming from the overuse of social media platform.
Amongst these, digital addiction and digital dementia are the two most obvious matters.
Based on an analysis of the algorithm principle used in this platform, Zhao (2021) argued that TikTok addiction is
becoming a widespread phenomenon, and such addiction has a closed-loop relationship with algorithm optimization.
More specifically, collaborative filtering algorithm and low-cost interaction design mechanism are very powerful tools
to facilitate users’ continuance intention. But in the meanwhile, the algorithm can also be counted as a trap to lure
people to spend too much time on the app.
FOMO, the abbreviation of “fear of missing out”, also a new type of social anxiety, refers to worries that one may be

International Journal of Social Science Studies Vol. 10, No. 5; 2022
63
absent from having rewarding experiences (Dossey, 2014) after excessive social media use. A case in point is the
“digital dementia” phenomenon in South Korea, where the deterioration in cognitive abilities frequently seen in
netizens. Eighteen percent of heavy internet users who is between 10 and 19, even cannot recall their own phone
numbers (Ryall, 2013).
It is worth mentioning that people’s notion of the efficacy of TikTok’s recommendation algorithm plays a nonnegligible
role of the decision about whether to quit this app. For instance, a recent survey with those who have abandoned or
never adopted Douyin (Lu et al., 2020) suggested the main reasons behind this phenomenon are the fear of addiction
and those low-quality videos full of stigmatized perceptions.
4.3 Challenge of Information Diversity
Recent studies have revealed a widespread skepticism about the role of social media plays in the issue of information
diversity. One of the most concerned worries is the curated filter bubbles created by social media algorithms (Bucher,
2018), where homogenous content traps users in the world that only affirms their beliefs (Hunt & McKelvey, 2019).
In fact, discussion about the impact of advanced technology on information diversity has not formed a consensus. Some
scholars support the increased exposure to diverse perspectives that technology brings us, while others warn the
increased risk of ideological segregation (Flaxman, Goel, & Rao, 2016).
Exacerbating this, the excessively catering of users’ preferences, results in the shallow and vulgarization of the content
(Yu, 2019). In other words, such problematic content stems from the excessively pursuit of strong user stickiness. But it
is a more suitable commercial goal to expand users instead of causing people to indulge in the network space.
Most studies do not pay sufficient attention to news flows on TikTok, but it is becoming a more and more important
news carrier today. The TikTok video content is a mixed mode of UGC (User-generated Content), PGC (Professionally
Generated Content) and OGC (Occupationally Generated Content) (Yu, 2019), and therefore, it provides more
diversified content library. To name a few, information about real life of LGBTQ+ people were shared on TikTok. In a
recent study, however, Simpson and Semaan (2021) showed that TikTok was suppressing and oppressing the identities
of its growing LGBTQ+ user population through algorithmic and human moderation of LGBTQ+ creators and content
related to LGBTQ+ identity.
Overall, it deserves further empirical evidence to check whether recommendation algorithms of TikTok contribute more
to information diversity than more established platforms.
5. Recommendations for Future Directions
To enable the recommendation algorithm to play a more significant role in video sharing platforms, improvements are
urgently needed for both algorithm design and framework implementation. This paper attempts to recommend three
aspects of possible future revolutions that recommendation algorithm in short-form mobile video platforms might make.
5.1 User-based Explainable Framework
User-based and targeted explainable framework is a feasible way to deal with the transparency and explainability
problem as well as to balance the algorithm-centric view and user-centered view. Tailored explanations based on
characteristics and expectations of different user groups (Pi, 2021) should be applied among technical and non-technical
end users. It not only makes the process of individualized video recommendation pattern more transparent, but also
strengthens users’ engagement in the construction of tailored recommendation mechanism.
Additionally, a practical framework is urgently needed to assess the algorithmic systems of video-focused platforms,
and to tell whether they are compatible with core digital platform related law principles. In the year 2020, TikTok
launched Transparency Center in Los Angeles to earn the trust of policymakers and the broader public (Pappas, 2020).
A copyright reform agenda can be made through cross-faculty cooperation in the near future.
5.2 Optimizing Algorithm Collaboratively
To address the closed-loop contradiction between TikTok addiction and algorithm optimization, both users and
ByteDance can contribute a lot. Users and recommender system are influencing each other: user is the most significant
factor for training and optimizing the recommendation model, and recommendation algorithm affects netizen’s habit of
receiving information.
We may suggest here several ways to unlock the closed-loop issue. It is beneficial that users’ self-conscious to improve
new media literacy and to get a deeper understanding of algorithm mechanisms. For the AI algorithm developers of
video sharing platforms, besides notification from anti-addiction system, recommendation accuracy can also adjust with
the users’ physical environment (Zhao, 2021), giving users impetus to have a rest from the screen.
5.3 Embedding Cultural Values into Recommendation Algorithm

International Journal of Social Science Studies Vol. 10, No. 5; 2022
64
To ensure that such a recommendation algorithm brings beneficial revolution to society, algorithm technology
optimization, value orientation and social requirements are equally important for the sustainable development of
short-form mobile video platform. It would be a meaningful trial to integrate the mainstream values of the society into
the design of recommendation algorithm.
More importantly, cultural values must match the users’ actual real-world society. Sun’s research team (2020) found that
the lifestyle of users in Douyin is simple and static, while TikTok users show a tendency to capture diverse items.
Traditional Confucian culture value in Chinese society and high self-expression value in English speaking countries
(Shen & Liang, 2015) can account for such differences.
The principles of decentralization are becoming one of the platform cultures of TikTok, where connection networks are
linked to each other. At the same time, AI and recommendation algorithms are replacing conventional human “Gate
Keeper” in the role of content selection and news flows control. The notion of algorithmic power is not just about the
ways in which algorithms determine the social world (Bucher, 2018). In contrast, the recommendation algorithm and
people are both involved and interact with each other in the digital world.
To mitigate the possibility of filter bubbles, the ByteDance company is purposely finding the balance between
user-relevant content and valuable potential experiences (Matsakis, 2020). Collaborative filters aim to make use of
people’s opinion to do a favor for people to make choices (Resnick, Iacovou, Suchak, Bergstrom, & Riedl, 1994). And
the collaborative filtering algorithm can help user explore their latent interests through user-based and item-based
filtering.
The market structures, governance frameworks, and infrastructures (Nieborg & Poell, 2018) of short-form mobile video
platforms make it possible to support the culture of creative practice through the recommendation designs (Zhou, 2019).
Therefore, the easiest start will be the training of effective algorithm models and the spread of civic innovation.
High-quality video input leads to prime value content output.
6. Conclusions and Limitations
The recommendation algorithm in digital platforms is a timely and effective response to the demands of today’ rapidly
evolving society.
This essay has provided an introduction to the recommendation algorithm application in TikTok. It first identified the
benefits of the use of recommendation algorithm for TikTok, featuring: powerful data processing capability, effective
content allocation, and precise user resonance exploration. Then it raised some concerns about the implementation of
recommendation algorithm, particularly for its ethical, legal and social risks. By way of further discussion, it is
suggested some possible directions for future short-form mobile video platforms: user-centered explainable and legal
framework, user-platform collaboration, and cultural values in algorithm.
There are several limitations of this paper as follows: recommendation algorithm is in early implementation phase and
the corresponding models are still immature. Given that some research may be excluded from examined range, the
scope of the reviewed social-scientific studies may not present the full picture of TikTok world under the context of
recommendation algorithm. No conventional qualitative or quantitative method used in this study. And therefore, points
in this essay are either assumptions or expectations extracted from literature review, lacking empirical support.
This essay also raises many fields worthy for future research. Subsequent studies can either take a multidisciplinary
approach to examine the technical issues or do a comparative analysis on the recommendation algorithm of TikTok’s
twin version Douyin or its counterparts like Snapchat and Kuaishou, which will be a valuable support for geographical
and cross-platform generalizability. Platform culture, user profile, and algorithm design are also key elements to
investigate for future platform-specific study.
References
Anderson, K. E. (2020). Getting acquainted with social networks and apps: it is time to talk about TikTok. Library Hi
Tech News, 37(4), 7-12. https://doi.org/10.1108/LHTN-01-2020-0001
Bandy, J., & Diakopoulos, N. (2020). # TulsaFlop: A Case Study of Algorithmically-Influenced Collective Action on
TikTok. In FAccTRec 2020 FAccTRec Workshop on Responsible Recommendation (FAccTRec ’20). arXiv preprint
arXiv:2012.07716.
Bucher T. (2018). If . . . Then: Algorithmic Power and Politics. Oxford: Oxford University Press.
Chen, Q., Min, C., Zhang, W., Ma, X., & Evans, R. (2021). Factors Driving Citizen Engagement With Government
TikTok Accounts During the COVID-19 Pandemic: Model Development and Analysis. J Med Internet Res, 23(2),
e21463. https://doi.org/10.2196/21463

International Journal of Social Science Studies Vol. 10, No. 5; 2022
65
Dossey, L. (2014). FOMO, Digital Dementia, and Our Dangerous Experiment. EXPLORE, 10(2), 69-73.
https://doi.org/https://doi.org/10.1016/j.explore.2013.12.008
Flaxman, S., Goel, S., & Rao, J. M. (2016). Filter Bubbles, Echo Chambers, and Online News Consumption. Public
Opinion Quarterly, 80, 298-320. https://doi.org/10.1093/poq/nfw006
Gray, J. E. (2021). The geopolitics of 'platforms': the TikTok challenge. Internet Policy Review, 10(3).
https://doi.org/10.14763/2021.2.1557
Guo, H., Tang, R., Ye, Y., Li, Z., & He, X. (2017). DeepFM: a factorization-machine based neural network for CTR
prediction. arXiv preprint. https://doi.org/10.48550/arXiv.1703.04247
Han, B. (2017). In the Swarm: Digital Prospects. Cambridge, Massachusetts: MIT Press.
https://doi.org/10.7551/mitpress/11055.001.0001
Hunt, R., & McKelvey, F. (2019). Algorithmic regulation in media and cultural policy: a framework to evaluate barriers
to accountability. Journal of Information Policy, 9(1), 307-335. https://doi.org/10.5325/jinfopoli.9.2019.0307
Klug, D., Qin, Y., Evans, M., & Kaufman, G. (2021). Trick and Please. A Mixed-Method Study On User Assumptions
About the TikTok Algorithm. 13th ACM Web Science Conference 2021, Virtual Event, United Kingdom.
https://doi.org/10.1145/3447535.3462512
Lu, X., Lu, Z., & Liu, C. (2020). Exploring TikTok Use and Non-use Practices and Experiences in China. In: G.
Meiselwitz, Social Computing and Social Media. Participation, User Experience, Consumer Experience, and
Applications of Social Computing. International conference on human-computer interaction 2020.
https://doi.org/10.1007/978-3-030-49576-3_5
Matsakis, L. (2020). TikTok Finally Explains How the ‘For You’ Algorithm Works. Retrieved from
https://www.wired.com/story/tiktok-finally-explains-for-you-algorithm-works/
Meese, J., & Hagedorn, J. (2019). Mundane Content on Social Media: Creation, Circulation, and the Copyright Problem.
Social Media + Society, 5(2). https://doi.org/10.1177/2056305119839190
Miao, W., Huang, D., & Huang, Y. (2021). More than business: The de-politicisation and re-politicisation of TikTok in
the media discourses of China, America and India (2017–2020). Media International Australia.
https://doi.org/10.1177/1329878X211013919
Nieborg, D. B., & Poell, T. (2018). The platformization of cultural production: Theorizing the contingent cultural
commodity. New Media & Society, 20(11), 4275-4292. https://doi.org/10.1177/1461444818769694
Nooren, P., van Gorp, N., & van Eijk, N. (2016, September). Digital platforms: A practical framework for evaluating
policy options. Proceedings of the Internet, Policy & Politics Conference.
Omar, B., & Dequan, W. (2020). Watch, Share or Create: The Influence of Personality Traits and User Motivation on
TikTok Mobile Video Usage. International Journal of Interactive Mobile Technologies (iJIM), 14(04), 121-137.
https://doi.org/10.3991/ijim.v14i04.12429
Page, M. J., McKenzie, J. E., Bossuyt, P. M., Boutron, I., Hoffmann, T. C., Mulrow, C. D., … Moher, D. (2021). The
PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ, 372, 71.
https://doi.org/10.1136/bmj.n71
Pappas, V. (2020). TikTok to launch Transparency Center for moderation and data practices. Retrieved from
https://newsroom.tiktok.com/en-us/tiktok-to-launch-transparency-center-for-moderation-and-data-practices
Perkel, D. (2011). Making art, creating infrastructure: DeviantART and the Production of the Web. University of
California, Berkeley. Retrieved from http://www.escholarship.org/uc/item/6fg9f992
Pi, Y. (2021). Machine learning in Governments: Benefits, Challenges and Future Directions. JeDEM - EJournal of
EDemocracy and Open Government, 13(1), 203-219. https://doi.org/10.29379/jedem.v13i1.625
Resnick, P., Iacovou, N., Suchak, M., Bergstrom, P., & Riedl, J. (1994). GroupLens: an open architecture for
collaborative filtering of netnews. Proceedings of the 1994 ACM conference on Computer supported cooperative
work, Chapel Hill, North Carolina, USA. https://doi.org/10.1145/192844.192905
Ryall, J. (2013). Surgein “digital dementia”. Retrieved from
http://www.telegraph.co.uk/news/worldnews/asia/southkorea/10138403/Surge-in-digital-dementia.html
Scanlon, K. (2020). The App That Knows You Better than You Know Yourself: An Analysis of the TikTok Algorithm.
Retrieved from
https://chatbotslife.com/the-app-that-knows-you-better-than-you-know-yourself-an-analysis-of-the-tiktok-algorith

International Journal of Social Science Studies Vol. 10, No. 5; 2022
66
m-be12eefaab5a
Sensor Tower. (2022). Sensor Tower's Q4 2021 Data Digest: Worldwide App Downloads Totaled 36.1 Billion Last
Quarter. Retrieved from https://sensortower.com/blog/q4-2021-data-digest
Shen, F., & Liang, H. (2015). Cultural difference, social values, or political systems? Predicting willingness to engage
in online political discussion in 75 societies. International Journal of Public Opinion Research, 27(1), 111-124.
https://doi.org/10.1093/ijpor/edu012
Simpson, E., & Semaan, B. (2021). For You, or For "You"? Everyday LGBTQ+ Encounters with TikTok. Proc. ACM
Human-Computer Interaction, 4(CSCW3), 252. https://doi.org/10.1145/3432951
Sun, L., Zhang, H., Zhang, S., & Luo, J. (2020, December 10-13). Content-based Analysis of the Cultural Differences
between TikTok and Douyin. 2020 IEEE International Conference on Big Data.
https://doi.org/10.1109/BigData50022.2020.9378032
TikTok Team. (2020). How TikTok recommends videos #ForYou. Retrieved from
https://newsroom.tiktok.com/en-us/how-tiktok-recommends-videos-for-you
TikTok Team. (2021). Thanks a billion. Retrieved from https://newsroom.tiktok.com/en-us/1-billion-people-on-tiktok
Valdovinos Kaye, D. B., Rodriguez, A., & Wikstrom, P. (2020). YOU MADE THIS? I MADE THIS: CULTURES OF
AUTOMATIC (MIS)ATTRIBUTION ON TIKTOK. AoIR Selected Papers of Internet Research, 2020.
https://doi.org/10.5210/spir.v2020i0.11354
Vijay, D., & Gekker, A. (2021). Playing Politics: How Sabarimala Played Out on TikTok. American Behavioral Scientist,
65(5), 712-734. https://doi.org/10.1177/0002764221989769
Wang, Y. (2020). Humor and camera view on mobile short-form video apps influence user experience and
technology-adoption intent, an example of TikTok (DouYin). Computers in Human Behavior, 110, 106373.
https://doi.org/https://doi.org/10.1016/j.chb.2020.106373
West, R., & Turner, L. H. (2013). Introducing communication theory: Analysis and application. New York: Marquette
University.
Yu, J. X (2019). Research on TikTok APP Based on User-Centric Theory. Applied Science and Innovative Research,
3(1), 28. https://doi.org/10.22158/asir.v3n1p28
Zhang, M., & Liu, Y. (2021). A commentary of TikTok recommendation algorithms in MIT Technology Review 2021.
Fundamental Research, 1(6), 846-847. https://doi.org/10.1016/j.fmre.2021.11.015
Zhao, Z. (2021). Analysis on the “Douyin (Tiktok) Mania” Phenomenon Based on Recommendation Algorithms. In E3S
Web of Conferences, 235, 03029. https://doi.org/10.1051/e3sconf/202123503029
Zhou, Q. (2019). Understanding User Behaviors of Creative Practice on Short Video Sharing Platforms - A Case Study
of TikTok and Bilibili (Doctoral dissertation). Available from ProQuest Dissertation & Theses: Full Text
(Publication Number 27539180).
Zhu, Y. M. (2020). The Expectation of TikTok in International Media: A Critical Discourse Analysis. Open Journal of
Social Sciences, 8(12), 136-148. https://doi.org/10.4236/jss.2020.812012
Copyrights
Copyright for this article is retained by the author(s), with first publication rights granted to the journal.
This is an open-access article distributed under the terms and conditions of the Creative Commons Attribution license
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly
cited.