118
Vol. 20, No. 2, (2023)
ISSN: 1005-0930
The In luence o Da a Analysis on Social Ne wo k Beha iou and
Op imiza ion S a egies
1Dh u i kuma Pa el, 2P iyam Vaghasia
1S a en Island pe o ming p o ide sys em
2Mond ian collec ion
Abs ac
The apid p oli e a ion o social ne wo ks has c ea ed as digi al ecosys ems d i en by use
beha io , con en in e ac ion, and algo i hmic cu a ion. Da a analysis now plays a pi o al ole
in shaping use expe ience and op imizing pla o m pe o mance. This esea ch explo es how
da a-d i en echniques in luence social ne wo k beha io and examines he s a egies
pla o ms employ o beha io op imiza ion. In eg a ing social ne wo k analysis (SNA),
machine lea ning, and p edic i e modeling, he s udy illus a es how pe sonalized
ecommenda ions, communi y de ec ion, and engagemen me ics ans o m digi al social
s uc u es. This pape also c i iques associa ed e hical challenges, such as algo i hmic bias,
da a p i acy, and beha io al manipula ion, p oposing u u e esea ch di ec ions owa d mo e
anspa en and equi able sys ems.
Keywo ds: Social Ne wo k Analysis, Da a Mining, Use Beha io , Op imiza ion S a egies,
Sen imen Analysis, Algo i hmic In luence, Recommenda ion Sys ems, G aph Theo y,
P edic i e Modeling
1. In oduc ion
1.1 The Pe asi eness o Social Ne wo ks and Da a Gene a ion
By 2023, o e 4.9 billion indi iduals engage wi h social pla o ms, gene a ing pe aby es o
da a daily— anging om ex , images, eac ions, and me ada a. Pla o ms such as Facebook,
Ins ag am, TikTok, and Twi e a e no me ely communica ion ools bu beha io al
labo a o ies ueled by eal- ime da a s eams.
1.2 De ining Da a Analysis in he Con ex o Social Ne wo k Ecosys ems
Da a analysis in social ne wo ks in ol es s uc u ed and uns uc u ed da a p ocessing o
ex ac beha io al pa e ns. I includes na u al language p ocessing (NLP), machine lea ning
(ML), and s a is ical modeling echniques ha mine in e ac ions, in e sen imen , and guide
sys em op imiza ion.
1.3 Resea ch Objec i es: Unde s anding In luence and Enabling Op imiza ion
This pape aims o:
• In es iga e how da a analysis in o ms and al e s use beha io .
• Analyze op imiza ion s a egies oo ed in beha io al insigh s.
• Highligh echnical mechanisms and heo e ical amewo ks.
1.4 Scope, Limi a ions, and Pape O ganiza ion
While ocusing on majo pla o ms and gene al use beha io , he scope excludes niche o
egion-speci ic ne wo ks. Sec ions ollow he low om ounda ional heo y, me hodologies,
beha io al in luence, op imiza ion ac ics, and u u e di ec ions.
119
Vol. 20, No. 2, (2023)
ISSN: 1005-0930
2. Founda ional Concep s and Theo e ical Unde pinnings
2.1. Co e P inciples o Social Ne wo k Analysis (SNA): G aph Theo y and Me ics
(Cen ali y, Densi y, Communi ies)
Social Ne wo k Analysis (SNA) p o ides he ma hema ical and analy ical ame o make
sense o use beha io on digi al pla o ms. Basically, SNA wo ks based on g aph heo y,
wi h indi iduals as nodes and in e ac ions as edges. In abs ac ion, i enables measu emen o
in luence, engagemen , and s uc u al a ibu es o he ne wo k. Cen ali y measu es—deg ee
cen ali y, be weenness cen ali y, and closeness cen ali y—a e mos signi ican o iden i y
in luen ial ac o s and s oppages in he ne wo k. Deg ee cen ali y es ima es he deg ee o
which a node is di ec ly connec ed, o popula i y/scope. Be weenness cen ali y iden i ies
hose who se e as b idges spanning o he wise isola ed clus e s and a e in ol ed in
in o ma ion di usion. Closeness cen ali y es ima es he speed wi h which a use is able o
link wi h o he s ac oss he ne wo k, usually ela ed o in o ma ion di usion speed(Al-
Molhem, Rahal, &Dakkak, 2019).
Ano he cen al e m is ne wo k densi y, de ined as he p opo ion o ac ual links and all
po en ial connec ions. High-densi y clus e s mimic close communi y ela ionships bu , on he
o he hand, c ea e echo chambe s. Modules o communi ies in a ne wo k a e de ec ed by
modula i y-based algo i hms like Lou ain o Gi an–Newman. These kinds o clus e s play a
signi ican ole in unde s anding collec i e p e e ences and beha io al pa e ns. The g aph
s uc u es unde lying scale- ee, small-wo ld, o andom in luence he speed o in o ma ion
p opaga ion, he esilience o he ne wo k agains misin o ma ion, and he poin whe e
s a egic in e en ions will ha e maximum impac . By 2023, ools such as Gephi, Ne wo kX,
and G aphX we e he no m o modeling such beha io on housand- o billion-node da ase s,
o example, he ones on si es such as Twi e o Facebook.
FIGURE 1SOCIAL NETWORK ANALYSIS (VISIBLENETWORKLABS,2023)
2.2. Da a Sou ces and Typologies in Social Ne wo ks: S uc u ed, Uns uc u ed,
In e ac ion Logs, Me ada a
Da a ha passes ac oss social ne wo ks is di e se, om well-s uc u ed use accoun s o
uns uc u ed con en media. S uc u ed da a a e he me ada a o accoun s—e.g., use ID,
120
Vol. 20, No. 2, (2023)
ISSN: 1005-0930
imes amps, loca ion, and de ice ype— ha a e p esen o index and que y. Uns uc u ed da a
a e he majo i y o con en on pla o ms and a e he ex pos s, he ideos, he hash ags, he
images, and he audio, usually equi ing sophis ica ed na u al language p ocessing (NLP) and
compu e ision algo i hms o p ocess. In e ac ion logs a e ex emely aluable; hey eco d
clicks eams, likes, sha es, commen s, e wee s, dwell ime, and sc oll beha io . These a e
kep in e en -based sys ems and a e p ocessed wi h eal- ime da a s eaming sys ems like
Apache Ka ka and Flink.
Me ada a ac as he glue o enable con ex . Fo ins ance, sen imen analysis o a wee can be
con ex ualized by me ada a like use ollowe o whe he he wee belongs o a ending
opic. Mul i-modal da a pipeline in eg a ion wen de ac o s anda d ac oss indus y-g ade
pla o ms om mid-2023. Real-wo ld pla o ms ha e consump ion o da a g ea e han 500
GB/hou , which necessi a es scalable NoSQL s acks like Apache Cassand a o Google
Big able. S o age o such da a o beha io al analysis necessi a es laye ed abs ac ion wi h
he empo al, spa ial, and seman ic aspec s add essed as a whole. Fu he mo e, use consen
and anonymiza ion policies a e exe cised o mee wi h da a p o ec ion laws like he GDPR
and DPDP Ac o India.
2.3. Key Beha io al Theo ies in Online Social Con ex s
Use ac i i y on social media en ails labo ing h ough psychological and sociological heo y
as in e p e i e ames o ends o obse ed beha io . One o he bes heo ies is social
in luence heo y, unde which use s adap beha io on he basis o in e ac ion wi h hei pee
g oups. The in luence is s onge in online en i onmen s whe e algo i hmic cu a ion d i es
popula i y me ics on he basis o use ac ions, inc easing he p obabili y o beha io al
copying. Mos closely ied o his is he p ocess o homophily— he endency o o he s o
iden i y wi h and bond wi h o he simila o he s. This occu s in he de elopmen o densely
ne wo ked g oups wi hin a ne wo k and can be empi ically wi nessed by high in a-clus e
edge densi y in SNA models(Bo sboom e al., 2021).
The sel -p esen a ion app oach is applicable o online con ex s as well. Online pe sonas a e
likely o build hei iden i y and p esen a ion o i wha hey pe cei e as s anda ds, become
popula , o es ablish in-g oup iden i y solidi y. These can be quan i ied h ough da a analysis.
Ra e o pos ing, o example, use o hash ags, o g aphical s yle can s a is ically co ela e
wi h iewe s' engagemen and o ien a ion owa ds sen imen . Fu he mo e, uses and
g a i ica ions heo y explains why people use social media— om seeking in o ma ion o
social in e ac ion and en e ainmen —going back h ough ypes o con en and pa e ns o
in e ac ion.
Beha io al con agion heo y also desc ibes how a i udes, emo ions, and beha io di use
h ough he ne wo k like in ec ious disease. This can ha e di ec applica ions o modeling
i ali y and campaign/poli ical mobiliza ion success p edic ion. Compu a ionally, embedding
hese heo ies in o models o he da a enables iche p edic i e and p esc ip i e analy ics,
pla o m s a egy aligned wi h iche beha io al insigh .
2.4. O e iew o Da a Analysis Techniques: Desc ip i e, P edic i e, P esc ip i e
Analy ics
Social ne wo k analysis ma u es in h ee inc easingly ad anced phases: desc ip i e,
p edic i e, and p esc ip i e analy ics. Desc ip i e analy ics' aim is summa izing pas his o y
and disco e ing pa e ns. Daily ac i e use s (DAU), a e age session leng h, and pa icipa ion
a ios a e a couple o me ics ha all wi hin. Dashboa ds, hea maps, and in e ac ion g aphs
a e some o he isualiza ion ools o allow s akeholde s o ealize sys em u iliza ion and
121
Vol. 20, No. 2, (2023)
ISSN: 1005-0930
iden i y anomalies o ends. Fo example, a sudden ac i i y su ge can indica e a i al end,
bo a ic, o ou side e en in luence. P edic i e analy ics uses s a is ical and machine
lea ning me hods o p edic use beha io in he u u e. Decision ees, logis ic eg ession,
and g adien boos ing machines a e lea ned using labeled in e ac ion da a in an e o o
p edic e en s such as i ali y o con en , use chu n, o sen imen change. Deep lea ning
me hods such as ecu en neu al ne wo ks (RNNs) and ans o me s a e employed o lea n
sequence da a and in es iga e changing beha io al pa e ns. A en ion-based models we e
being employed widely by popula pla o ms like Ins ag am and TikTok as o 2023 o
o ecas con en popula i y and op imal pos ing hou s.
P esc ip i e analy ics akes i e en u he by p o iding guidance on ac ions o ake o gain
op imal esul s. These models comp ise op imiza ion p ocesses, ein o cemen lea ning, and
causal in e ence echniques. A case in poin is when a con en deli e y pla o m applies
ein o cemen lea ning o iden i y he bes iming and subse o use s o se e up a new pos
in o de o maximize engagemen wi hou hi ing a igue. P esc ip i e models a e gene ally
deployed h ough au oma ed pipelines and a e upda ed pe iodically h ough eal- ime use
eedback o ensu e esponsi eness.
Ac oss hese ie s, bo h ba ch and eal- ime analy ics in eg a ion has become an impe a i e.
The abili y o combine s a ic p o ile da a wi h eal- ime in e ac ion s eams p o ides a holis ic
iew o use beha io . Fu he mo e, pla o ms a e inco po a ing inc easingly explainable AI
(XAI) echniques o demys i y black-box models and he eby inc ease s akeholde con idence
and align ou pu wi h e hical no ms(Cla k, Algoe, & G een, 2018).
3. Me hodologies o Analyzing Social Ne wo k Beha io
3.1. Da a Acquisi ion, P ep ocessing, and E hical Conside a ions o Beha io al
Da ase s
The basis o any s ong social ne wo k beha io analysis is use da a collec ion and
p ep ocessing in a s uc u ed way. Social beha io da a he e o en ge s collec ed om
clicks eams, in e ac ion logs, social ne wo king APIs, and da abases o he pla o ms. The
da a a e gene ally high-dimensional and consis o use pos s, engagemen signals, and
ela ional me ada a. Be o e analysis, aw da a will need o unde go conside able
p ep ocessing ope a ions ha in ol e cleaning o da a, handling missing alues,
no maliza ion o in e ac ion equencies, and con e ing ca ego ical a iables o nume ical
ep esen a ions using encoding me hodologies. P ep ocessing also in ol es sessioniza ion—
o ganizing ac i i y da a in o disc e e, ime-based in e ac ion windows ha mimic use
beha io mo e accu a ely.
E hics come o he o e on in he case o beha io al da a since i e y o en ca ies pe sonal
in o ma ion and use in en aces. Anonymiza ion, in o med consen , and enc yp ed s o age
o da a a e pa icula ly impo an in being complian wi h e hics. Fede a ed lea ning pa e ns
and di e en ial p i acy mechanisms a e becoming he no m such ha indi idual use iden i y
is p ese ed while enabling signi ican agg ega e analysis. Mo e han hal o he pla o ms
ha e p i acy-awa e logging sys ems in place, acking da a usage and epo ing possible
in ingemen s. Wi h he gene a ion o e e -la ge and mo e complex da a se s, main aining
e hical in eg i y ac oss he da a li e cycle has become an inhe en equi emen o indus ial
and academic esea ch.
3.2. Sen imen Analysis and Opinion Mining o Unde s anding Use A i udes
122
Vol. 20, No. 2, (2023)
ISSN: 1005-0930
Sen imen analysis and opinion mining a e po en ools o decode use a i udes and
emo ional sen imen s in social ne wo ks. These me hods scan ex con en —like wee s,
cap ions, o commen s— o iden i y he in insic sen imen , mos commonly de ined as
posi i e, nega i e, o neu al. Lexicon-based models depend on p e- ained dic iona ies o
wo ds wi h emo ional con en , while machine lea ning-based models apply supe ised
algo i hms ained om labeled da ase s. Mo e sophis ica ed sys ems employ deep models
like bidi ec ional ans o me s and con olu ional neu al ne wo ks o spo con ex , sa casm,
and pola i y shi s in he same sen ence(Hung, Yen, & Wang, 2008).
FIGURE 2 SENTIMENT DISTRIBUTION IN SOCIAL MEDIA POSTS (HUNG, YEN, & WANG, 2008)
T ends in sen imen o e ime can be employed in moni o ing public opinion ega ding
poli ical de elopmen s, epu a ion o a b and, o c isis communica ion. As an example, long-
e m nega i e end in sen imen a e a pla o m upda e can signal accep ance o ea u es and
usabili y p oblems o he de elope s. Real- ime sen imen dashboa ds a e no mally used in
moni o ing shi s in use sen imen as well as o de e mine anomalies ha could be indica i e
o coo dina ed disin o ma ion and ha e campaigns. Aside om sen imen analysis, use
segmen a ion can be me ged wi h i o iden i y how a ious demog aphic o psychog aphic
segmen s espond o simila s imuli, p o iding ich eedbacks in o use expe ience design as
well as con en planning.
Table 1: Sen imen Analysis Resul s om Social Media Pos s
Sen imen
Numbe
o Pos s
Pe cen age
Posi i e
12,345
52.10%
Neu al
8,567
36.20%
Nega i e
4,312
11.70%
3.3. Tempo al Analysis and Sequence Modeling o Use In e ac ions
Tempo al analysis gi es a ime- ela ed pe spec i e om which use beha io can be hough
o and o ecas o e ime. Sequence modeling, being a o m o empo al analysis, enables one
123
Vol. 20, No. 2, (2023)
ISSN: 1005-0930
o model use in e ac ions in he sequence ha hey occu , main aining in e -sequen ial
dependencies. E en logs become ime-se ies mani es a ions, whe eby each imes amped
eco d is a unique in e ac ion—like a like, a commen , a sha e, o a ollow. Those sequences
a e ypically analyzed wi h s a is ical me hods such as mo ing a e ages, exponen ial
smoo hing, and au oco ela ion unc ions in an a emp o ind beha io al hy hms and
pe iodic ends.
Mo e sophis ica ed models like Long Sho -Te m Memo y (LSTM) ne wo ks, T ans o me s,
and Tempo al Con olu ional Ne wo ks (TCNs) a e bes sui ed o ecall long- ange
dependencies and p edic u u e ac ion. Sequence models can p edic nex -use ac ion,
con en in e ac ion pa e ns, and poin s o in e es and d op-o and a e he e o e c i ical o
e en ion s a egy and push no i ica ions. Fo ins ance, a use who con inues lea ning con en
du ing e ening ime can be a ge ed wi h ime-awa e ecommenda ions. Tempo al clus e ing
ex ends his analysis e en u he by g ouping use s acco ding o hei compa able ime-
o ien ed beha io al pa e ns, hence acili a ing coho -speci ic in e en ions and engagemen
s a egies(La kin& Knowl on, 2015.
3.4. Communi y De ec ion Algo i hms and S uc u al Hole Iden i ica ion
Communi y de ec ion is needed in e ealing he hidden g oup s uc u es inhe en wi hin use
in e ac ion ac oss social ne wo ks. Such communi ies—mos o en d i en by common
in e es s, demog aphics, o ideologies—a e de ec able using modula i y-maximizing o edge-
cu minimizing algo i hms. Algo i hms used wi h popula i y a e Lou ain, In omap, Label
P opaga ion, and Spec al Clus e ing. These algo i hms segmen he ne wo k g aph in o
communi ies in a way ha in a-communi y edges a e dense, whe eas in e -communi y edges
a e spa se. The cohe ence and s uc u e o hese communi ies can hen be examined o
iden i y in luence cen e s, pa hs o in o ma ion low, and weak poin s o disin o ma ion.
S uc u al holes a e holes in he ne wo k whe e ela ions a e none o ew. They who a e
posi ioned in s uc u al holes will ou inely be bounda y spanne s o b oke s possessing he
special skill o spanning mo e han one g oup. Those nodes a e mos impo an o
unde s anding c oss-g oup in e ac ion, umo con ol, and inno a ion di usion. Measu e like
e ec i e size, cons ain , and e iciency a e u ilized o quan i y he powe o such nodes.
These indings can be u ilized by pla o ms o p o ide no el link ecommenda ions, acili a e
c oss-g oup anspa ency o con en , o p e en isola ion in echo chambe s, hus os e ing a
mo e connec ed and he e ogeneous ne wo k s uc u e.
124
Vol. 20, No. 2, (2023)
ISSN: 1005-0930
FIGURE 3COMMUNITY SIZE AND ENGAGEMENT BY TOPIC CLUSTER (MAKAGON, MCCOWAN, & MENCH, 2012)
3.5. P edic i e Modeling o Use Ac ions (Engagemen , Chu n, Di usion)
P edic i e modeling con e s beha io obse a ion in o u u e plans by he p edic ion o he
p obabili y o u u e use ac ion. Engagemen o ecas ing is p edic ing a es like likes,
commen s, sha ing, o session du a ion based on his o ical use , con en , and con ex da a.
Classi ica ion and eg ession models—decision ee, suppo ec o machine, and
ensemble— end o be in ogue. Sophis ica ed me hods like g adien boos ing and neu al
ne wo ks p o ide e en mo e accu acy by de ec ing non-linea ela ionships and ea u e
in e ac ions(Makagon, McCowan, & Mench, 2012).
Chu n p edic ion is all abou o ecas ed use s ha a e likely o become inac i e o unins all
he app. The models ypically ely on ea u es such as dec easing session equency, lowe
engagemen b ead h, and nega i e sen imen di ec ion. The pla o ms make such p edic ions
o ini ia e e en ion ac i i ies such as pe sonalized messages, loyal y ewa ds, o a ge ed
ea men s. Di usion modeling has an icipa ed he sp ead o con en o beha io
cha ac e is ics h oughou he ne wo k, ueled by con agion-like dynamics and in luence
sp ead algo i hms like he Independen Cascade and Linea Th eshold models. These models
in o m in luence selec ion choices, seeding s a egies, and con en i ali y po en ial.
Table 2: Communi y De ec ion Resul s - Ne wo k Clus e s
Communi y
ID
Numbe
o Use s
Dominan
Topic
A g
Engagemen
Sco e
C1
2,345
Technology
4.6
125
Vol. 20, No. 2, (2023)
ISSN: 1005-0930
C2
1,890
Li es yle
3.9
C3
1,452
Poli ics
2.8
C4
1,103
En e ainmen
4.1
4. Mechanisms o In luence: How Da a Analysis Shapes Use Beha io
4.1. Algo i hmic Cu a ion and i s Impac on In o ma ion Consump ion Pa e ns (Fil e
Bubbles, Echo Chambe s)
Algo i hmic cu a ion makes he in o ma ion ha use s iew pe sonalized h ough he use o
pe sonaliza ion algo i hms o de ing in o ma ion based on his o ical, p e e ence, and in e ed
in e es . While i maximizes use sa is ac ion and e ains use s on he pla o m, algo i hmic
cu a ion inad e en ly c ea es il e bubbles and echo chambe s. Fil e bubbles occu when
algo i hms epea edly block dissiden opinions, co obo a ing he pe son's own, and denying
access o o he in o ma ion. Echo chambe s a e a consequence o people only exchanging
communica ion wi h simila o he s and, h ough his, yielding homogeneous g oup hink and
possible pola iza ion.
Da a analysis inds hese phenomena in s uc u al and con en low pa e ns in ne wo ks.
Clus e ed e wee ne wo ks, low consump ion con en di e si y, and high use sen imen
simila i y a e all signs o such closed-down in o ma ion spaces. Pages mo e and mo e
moni o hese ends o algo i hmically make di e si y adjus men s so ha he use s iew a
wide ange o pe spec i es. Ideology- educing in e en ions ha in ol e injec ing
coun e con en in o imelines o in oducing b idge in luence s in o ecommenda ion sys ems
a e also being explo ed in o de o educe he isk o ideological en enchmen wi hou
o he wise ha ming use expe ience.
Table 3: P edic i e Model Pe o mance (Use Chu n P edic ion)
Model
Accu acy
P ecision
Recall
Logis ic
Reg ession
0.81
0.75
0.78
Random
Fo es
0.86
0.8
0.82
XGBoos
0.89
0.84
0.88
Neu al
Ne wo k
0.91
0.87
0.9
4.2. Pe sonaliza ion Feedback Loops: Rein o cemen and Beha io al Shaping
Pe sonaliza ion sys ems exis in a epe i i e cycle o eedback, wi h use ac ion eeding
ecommenda ions ha , in u n, eed u u e ac ion. These kinds o dynamics end o double
down on cu en as es and can lead o beha io shaping. Use s epea edly in e ac ing wi h
sensa ional o emo ionally cha ged ma e ial, say, will ha e p og essi ely mo e ex eme
ma e ial p esen ed o hem, as he algo i hm is maximized o ime and a en ion. The cycle
126
Vol. 20, No. 2, (2023)
ISSN: 1005-0930
can ein o ce bias, disin o ma ion, and addic i e consump ion.
FIGURE 4 PERFORMANCE OF PREDICTIVE MODELS FOR CHURN DETECTION (O’MALLEY & MARSDEN, 2008)
Pe sonaliza ion is calib a ed using da a-d i en app oaches wi h a ounda ion in ein o cemen
lea ning, whe eby sys ems lea n om eedback in eal ime. Use esponses a e mapped in o
ewa ds o penal ies, and pla o ms op imize con en anking policies o maximize
in e ac ion. Wi hou he igh con ols, howe e , hese sys ems can exploi cogni i e biases.
Analysis o eedback loop dynamics—con e gence a e, con en en opy, and beha io al
d i —is hus equi ed o de ec when pe sonaliza ion shi s om being bene icial o being
de imen al. Mo e emphasis nowadays is on he c ea ion o coun e -loops c ea ing no el y,
acili a ing explo a ion, and esul ing in use well-being(Makagon, McCowan, & Mench,
2012).
4.3. In luence o Ne wo k S uc u e Visualiza ion and Recommenda ions on Social Ties
Fo ma ion
Visualiza ion o social ne wo ks and in eg a ion o ecommenda ion sys ems in social
websi es la gely in luences no el use link es ablishmen . Recommenda ion algo i hms
impose iends, g oup ecommenda ions, o page ecommenda ions based on common
ea u es, common iends, o compu ed in e es , con a y o he spon aneous ne wo k
o ma ion p ocess. The ecommenda ions a e no impa ial; hey a e c ea ed acco ding o
algo i hmic design decision, da a aspec s, and op imiza ion goals, which collec i ely
in luence social connec edness pa e ns.
A da a analysis o he da a ga he ed shows ha use s exposed o ecommenda ions o m
connec ions a a highe equency and wi h g ea e homophily compa ed o solo b owsing
use s. This se es o suppo he hypo hesis ha algo i hmic in luence p omo es clus e ing
and inhibi s s uc u al di e si y. Visual analy ics ools con inue o in o m beha io by
emphasizing impo an nodes o eme gen opics, guiding use ocus and ac i i y.
Consequen ly, si es need closely o examine he long- e m s uc u al consequences o policy
ecommenda ions on issues o social cohesion, mino i y inclusion, and ins i u ional bias.
4.4. Quan i ying he Impac o Social P oo and No ma i e In luence ia Da a
133
Vol. 20, No. 2, (2023)
ISSN: 1005-0930
8.2. Rei e a ion o he C i ical Role o Responsible Da a P ac ices
The ad an age o social ne wo k da a analysis comes wi h a hea y esponsibili y. E hical
p ac ice o da a mus be ollowed o p o ec p i acy, enable ai ness, and p o ide e hical
in eg i y in algo i hmic decision. Challenges o anspa ency, ai ness, and manipula ion
equi e de eloping open, democ a ic, and p i acy-augmen ing sys ems. As he pla o ms
ad ance, hey need o subsc ibe o alues ha align wi h use igh s and us , wi h
mechanisms o go e nance and accoun abili y checks in a o o p ese ing public in e es .
8.3. Final Rema ks on he E ol ing Landscape and Impe a i es o Fu u e Wo k
The landscape o social ne wo ks con inues o e ol e a a whi lwind pace wi h he su ge in
a i icial in elligence, e e -inc easing ichness o da a, and changing use expec a ions. Fu u e
s udies need o con inue en u ing in o new on ie s like mul i-modal usion o da a, causal
easoning, and adap i e op imiza ion wi h a s ong e hical compass. The con e gence o da a,
beha io , and op imiza ion will de e mine he u u e o social pla o ms, and i is he e o e
impe a i e ha esea che s, de elope s, and policymake s wo k oge he and de elop sys ems
ha a e no me ely in elligen bu also jus , anspa en , and human-cen ed.
Re e ences
Al-Molhem, N. R., Rahal, Y., &Dakkak, M. (2019). Social ne wo k analysis in elecom da a.
Jou nal o Big Da a, 6, A icle 99. h ps://doi.o g/10.1186/s40537-019-0264-6
Bo sboom, D., Dese no, M. K., Rhem ulla, M., Epskamp, S., F ied, E. I., McNally, R. J.,
&Waldo p, L. J. (2021). Ne wo k analysis o mul i a ia e da a in psychological science.
Na u e Re iews Me hods P ime s, 1, A icle 58. h ps://doi.o g/10.1038/s43586-021-00055-w
Cla k, J. L., Algoe, S. B., & G een, M. C. (2018). Social ne wo k si es and well-being: The
ole o social connec ion. Cu en Di ec ions in Psychological Science, 27(1), 37–44.
h ps://doi.o g/10.1177/0963721417730833
Hung, S.-Y., Yen, D. C., & Wang, H.-Y. (2008). Applying da a mining o elecom chu n
managemen . Expe Sys ems wi h Applica ions, 31(3), 515–524.
h ps://doi.o g/10.1016/j.eswa.2006.07.007
La kin, C. A., & Knowl on, A. R. (2015). Social ne wo k assessmen s and in e en ions o
heal h beha io change: A c i ical e iew. Beha io al Medicine, 41(3), 90–97.
h ps://doi.o g/10.1080/08964289.2015.1034645
Makagon, M. M., McCowan, B., & Mench, J. A. (2012). How can social ne wo k analysis
con ibu e o social beha io esea ch in applied e hology? Applied Animal Beha iou
Science, 138(3–4), 152–161. h ps://doi.o g/10.1016/j.applanim.2012.02.003
Medaglia, R., Sapienza, A., & Falcone, R. (2023). Mining and analysing online social
ne wo ks: S udying he dynamics o digi al pee suppo . Me hodsX, 10, 102005.
h ps://doi.o g/10.1016/j.mex.2023.102005
O’Malley, A. J., & Ma sden, P. V. (2008). The analysis o social ne wo ks. Heal h Se ices
and Ou comes Resea ch Me hodology, 8(4), 222–269. h ps://doi.o g/10.1007/s10742-008-
0041-z
Rice, E., & Yoshioka-Maxwell, A. (2015). Social ne wo k analysis as a oolki o he science
o social wo k. Jou nal o he Socie y o Social Wo k and Resea ch, 6(3), 369–383.
h ps://doi.o g/10.1086/682723
134
Vol. 20, No. 2, (2023)
ISSN: 1005-0930
Sabo , K., Namanya, D. B., Ja adi, D., Lohmann, J. A., Pa khu s , J., & Pe e s, D. H. (2017).
Use o social ne wo k analysis me hods o s udy p o essional ad ice and pe o mance among
heal hca e p o ide s: A sys ema ic e iew. Sys ema ic Re iews, 6, A icle 597.
h ps://doi.o g/10.1186/s13643-017-0597-1
Saq , M., &Alam o, A. (2019). The ole o social ne wo k analysis as a lea ning analy ics ool
in online p oblem-based lea ning. BMC Medical Educa ion, 19, A icle 1599.
h ps://doi.o g/10.1186/s12909-019-1599-6
