In e na ional Jou nal o Inno a i e Technology and Explo ing Enginee ing (IJITEE)
ISSN: 2278-3075 (Online), Volume-14 Issue-12, No embe 2025
36
Published By:
Blue Eyes In elligence Enginee ing
and Sciences Publica ion (BEIESP)
© Copy igh : All igh s ese ed.
Re ie al Numbe : 100.1/iji ee.D831014041125
DOI: 10.35940/iji ee.D8310.14121125
Jou nal Websi e: www.iji ee.o g
Inno a i e Spam De ec ion Using Hyb id Machine
Lea ning Algo i hms: A Da a-Cen ic App oach
S ee Vidya Venigalla, K.V.D. Ki an
Abs ac : The ise o spam messages, in he o m o malwa e,
phishing a acks, and un eques ed messages, poses a se ious h ea
o in e ne use s and secu i y in as uc u es. Con en ional spam
il e ing echniques ha ely solely on s ic ules and keywo d lis s
s uggle o keep pace wi h con empo a y spamme ac ics ha
mask malicious con en . This s udy p oposes a solu ion o his
challenge by de eloping a hyb id machine lea ning me hodology
ha le e ages Nai e Bayes (NB) and a Suppo Vec o Machine
(SVM), combining hem in o an ensemble o imp o ed accu acy
and esilience in spam de ec ion. The echnique uses he well-
known SMS Spam Collec ion Da ase . I employs mo e complex
ex ual ea u e ex ac ion (TF-IDF), as well as addi ional non-
ex ual ea u es such as message leng h, wo d capi alisa ion, and
he equency o p e iously de e mined keywo ds. The p oposed
sys em is ex ensi ely e alua ed using s anda d classi ica ion
me ics—accu acy, F1 sco e, p ecision, and ecall — o assess i s
eliabili y and alidi y. The esea ch indings indica e ha he
p oposed machine lea ning hyb id ensemble is e ec i e a
educing alse posi i es while mo e boldly ackling he challenges
inhe en in he eal-wo ld spam da a en i onmen . The esea ch
p ojec o e s p ac ical po en ial o use; he hyb id p oposed
sys em is compu a ionally e icien enough o mos eal- ime
deploymen applica ions in au oma ed sys ems o comba spam.
This esea ch con ibu es scalable, adap i e spam-de ec ion
mechanisms sui able o eal- ime messaging en i onmen s.
Keywo ds: Spam De ec ion; Machine Lea ning; Nai e Bayes;
Suppo Vec o Machines; Ensemble Models; Nlp; Tex
Classi ica ion; Cybe secu i y; Hyb id Algo i hm; Da a Analy ics
Nomencla u e:
NB: Nai e Bayes
SVM: Suppo Vec o Machine
ML: Machine Lea ning
TF-IDF: Te m F equency-In e se Documen F equency
CNN: Con olu ional Neu al Ne wo k
LSTM: Long Sho -Te m Memo y
NLP: Na u al Language P ocessing
APC: A icle P ocessing Cha ge
RNNs: Recu en Neu al Ne wo ks
I. INTRODUCTION
Spam, o unwan ed and ha m ul con en , has inc eased
Manusc ip ecei ed on 28 Oc obe 2025 | Fi s Re ised
Manusc ip ecei ed on 06 No embe 2025 | Second Re ised
Manusc ip ecei ed on 11 No embe 2025 | Manusc ip
Accep ed on 15 No embe 2025 | Manusc ip published on 30
No embe 2025.
*Co espondence Au ho (s)
S ee Vidya Venigalla*, S uden , Depa men o Compu e Science
Enginee ing, Kone u Lakshmaiah Educa ional Founda ion, Vijayawada
(Andh a P adesh), India. Email ID: s ee idya. [email protected],
ORCID ID: 0009-0002-0009-4441
D . K.V.D. Ki an, P o esso , Depa men o Compu e Science
Enginee ing, Kone u Lakshmaiah Educa ional Founda ion, Vijayawada,
(Andh a P adesh), India. Email ID: ki an_cse@kluni e si y.in, ORCID ID:
0000-0002-8808-9307
© The Au ho s. Published by Blue Eyes In elligence Enginee ing and
Sciences Publica ion (BEIESP). This is an open-access a icle unde he
CC-BY-NC-ND license h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/
Exponen ially alongside he apid inc ease in olume and
speed o communica ion on he In e ne . Spam in ades email
inboxes, consumes aluable use ime and esou ces, and
ul ima ely se es as a ec o o malwa e and a acks ha
a ge use s h ough iden i y he and phishing. As spamme s
become mo e adep and use echniques like con en
ob usca ion o alse subjec lines, we ind ha adi ional
spam il e s ha ely on s a ic keywo d ma ching o blocklis s
a e less e ec i e.
To imp o e de ec ion accu acy and lexibili y, esea che s
ha e inc easingly le e aged machine lea ning (ML)
echniques o de ec complex ex ual and s uc u al ea u es
and pa e ns in messages. Me hods such as Nai e Bayes,
Suppo Vec o Machines, Decision T ees, and ensemble
classi ie s a e ound o be e ec i e spam s. ham (legal
message) de ec o s, and a he same ime, lea ning in labelled
da ase s using many ea u es such as wo d equency, special
cha ac e usage, message leng h, and e m weigh ing.
This esea ch builds on hese me hods o p opose a hyb id
sys em ha combines hei classi ie s eng hs ia an
ensemble, educing alse posi i es and enabling highe
de ec ion a es. The model is e alua ed using s anda d
classi ica ion me ics, such as accu acy, p ecision, ecall, and
F-1 sco e, while also being assessed in di e en SMS spam
da ase s ( eal-wo ld), ul ima ely showing he obus ness and
e ec i eness o he hyb id model, compa ed o only using
indi idual classi ie s.
II. RELATED WORK
Nume ous s udies ha e examined he use o machine
lea ning algo i hms o spam classi ica ion. Nai e Bayes s ill
se es as a good baseline, no jus because o i s p obabilis ic
e iciency, bu also because i does no su e om in easibly
signi ican scaling issues when applied o ex ual da a [1].
Suppo Vec o Machines (SVMs) ha e been shown o
pe o m well on ex classi ica ion da ase s, as hey a e
e ec i e o high-dimensional [2], non-linea ela ionships
among speci ic message cha ac e is ics [5].
Ensemble and hyb id models, which combine mul iple
classi ie s, ha e gained popula i y ecen ly o allow highe
accu acy and o mi iga e alse posi i es [4]. Fo example,
Nai e Bayes and SVM achie ed be e esul s when
combined han when used as base classi ie s in spam da ase s
[12].
Deep lea ning me hods, pa icula ly Con olu ional Neu al
Ne wo ks (CNNs), a e s ill widely explo ed o cap u ing
con ex ual seman ic knowledge in ex classi ica ion, as
e iewed in dep h by Minaee e al. [7]. Recen ly, mode n
neu al a chi ec u es, as Young
and colleagues discuss, suppo
he end o deep lea ning as
being be e a imp o ing
Inno a i e Spam De ec ion Using Hyb id Machine Lea ning Algo i hms: A Da a-Cen ic App oach
37
Published By:
Blue Eyes In elligence Enginee ing
and Sciences Publica ion (BEIESP)
© Copy igh : All igh s ese ed.
Re ie al Numbe : 100.1/iji ee.D831014041125
DOI: 10.35940/iji ee.D8310.14121125
Jou nal Websi e: www.iji ee.o g
asks in na u al language p ocessing o spam il e ing [8].
Cu en in es iga ions also ad ance me hods o compensa e
o imbalanced da ase s and adap o e ol ing pa e ns in
spam con en , he eby allowing classi ie s o emain obus
agains ad e sa ial a acks on spam [6]. Inno a i e ensemble
me hods ha combine adap i e deep lea ning app oaches
ha e demons a ed imp essi e gene alisa ion pe o mance
o la ge-scale spam de ec ion [11].
Using sen imen analysis along wi h con ex ual embeddings
has imp o ed classi ica ion pe o mance by cap u ing he
linguis ic nuances inhe en in spam [3]. Meanwhile, me hods
ha ake a p obabilis ic app oach, such as Bayesian ne wo ks,
con inue o make aluable con ibu ions o spam analysis
sys ems, p o iding obus ness wi hou sac i icing
in e p e abili y [9].
O e all, he li e a u e sugges s ha al hough adi ional
models, such as Nai e Bayes and SVM, con inue o ha e
u ili y, hyb id and ensemble lea ning me hods—combining
p obabilis ic, disc imina i e, and deep lea ning me hods—
ep esen he mos scalable and accu a e app oach o eal-
ime spam de ec ion [4].
III. LITERATURE REVIEW
O e he las 20 yea s, spam de ec ion has shi ed om
heu is ic-based app oaches o machine lea ning and deep
lea ning. In he pas , spam de ec ion app oaches elied on
humans o de ine ules, which could be ei he heu is ics o
keywo d blocklis s. These ea lie app oaches could no adap
o he apidly changing s yles and pa e ns o spam messages.
Wi h ad ancemen s in supe ised lea ning, spam de ec ion
sys ems began au oma ing he spam de ec ion ask by using
s a is ical modelling and da a-based classi ica ion [1].
Nai e Bayes (NB) classi ie s a e among he mos
s aigh o wa d and e icien models o spam de ec ion,
hanks o hei p obabilis ic s uc u e and scalabili y, e en
wi h limi ed compu a ional powe [1]. Suppo Vec o
Machines (SVMs) ha e also been e y success ul in
classi ying ex da a and pe o ming well in non-linea high-
dimensional ea u e spaces [5].
A p oduc o inc eased esea ch in o spam de ec ion has
been he inco po a ion o deep lea ning a chi ec u es in o
spam de ec ion sys ems o imp o e ea u e ep esen a ion and
con ex ual awa eness. Zhang e al. ound ha neu al
a chi ec u es such as Con olu ional Neu al Ne wo ks
(CNNs) and Recu en Neu al Ne wo ks (RNNs)
ou pe o med adi ional classi ie s when da a om he spam
messages and compu a ional powe we e no limi ed [6].
Recen wo k shows ha CNNs pe o m well a p oducing
seman ic-le el ea u es o ex classi ica ion, enabling a
ansi ion o models ha inco po a e con ex [7].
Akbik and colleagues p esen ed con ex ual s ing
embeddings o enhance ea u e ep esen a ion in spam
de ec ion by le e aging wo d-le el dependencies and con ex
om a la ge co pus o ex s [3]. Likewise, Young e al.
emphasised he b oad impac ha deep lea ning would ha e
on many na u al language p ocessing asks, including spam
de ec ion, due o he inc eased lea ning scope a o ded by
adap i e and ans e lea ning [8].
Fea u e ex ac ion emains an essen ial ac o in a spam
il e 's e ec i eness. A ecen pape e iews he con inued
use ulness o TF-IDF and newe s a egies o quali a i e
ea u e ep esen a ion in u u e ex mining applica ions [13].
Recen wo k shows he impo ance o a new, di e se se o
SMS spam da ase s o e ec i ely benchma king eal-wo ld
spam il e s [10].
In ecen yea s, ensemble o hyb id me hodologies ha e
become popula in spam and o he de ec ion ields o
combining he p edic i e abili y o mul iple classi ie s. Yang
e al. ound ha hyb id models combining Nai e Bayes,
Decision T ees, and SVM classi ie s achie ed lowe alse-
posi i e a es and highe accu acy in spam de ec ion [4].
Mo e ecen ly, Lee e al. de eloped an ensemble-based deep
lea ning amewo k ha pe o ms well on imbalanced
da ase s and demons a es high classi ica ion pe o mance
[12].
P obabilis ic models (e.g., Bayesian ne wo ks) emain
ele an and can inc ease model in e p e abili y wi hou
sac i icing accu acy [9]. Also, de eloping adap i e lea ning
a chi ec u es, such as an inno a i e deep lea ning-d i en
spam de ec ion amewo k enabling con inuous lea ning in
dynamic da a, is happening as demons a ed by Young e al.
in hei 2025 s udy [11].
In summa y, while deep lea ning a chi ec u es ha e
achie ed be e - han-beha iou pe o mance in classi ica ion,
hei compu a ional complexi y and da a equi emen s may
limi pe o mance o small domains, necessi a ing hyb id
machine lea ning deploymen s. Hyb idising machine
lea ning sys em-based deploymen s wi h he in e p e abili y
o Nai e Bayes, he capabili y o SVM (disc imina i e), and
he adap abili y o mode n neu al ne wo ks is p obably he
mos easible and scalable op ion o eal- ime spam de ec ion
[4]
IV. METHODOLOGY
This esea ch p esen s a hyb id machine lea ning app oach
o spam de ec ion using se e al Nai e Bayes (NB) and
suppo ec o machine (SVM) classi ie s o imp o e
accu acy, lexibili y, and obus ness. The s udy includes i e
phases: da ase p epa a ion, p ep ocessing, ea u e ex ac ion,
model aining, and e alua ion.
A. Da ase
La es de elopmen s equen ly use he SMS spam da ase s,
which a e desc ibed as p o iding an ex ensi e benchma k o
spam de ec ion s udies [10]. I consis s o 5,574 SMS
messages; app oxima ely 13.4% a e spam, and 86.6% a e
non-spam ham. This da ase has been consis en ly alida ed
in he li e a u e as an e ec i e benchma k o e alua ing
machine lea ning algo i hms.
B. P e-P ocessing
Be o e machine lea ning, he aw messages a e no malised
using a s anda d selec ion o p ep ocessing ope a ions:
i. Con e in o lowe case
ii. Remo e punc ua ion, numbe s, and special symbols.
iii. Remo e s opwo ds.
i . Enable s emming o
ans o m wo ds in o
hei base o m.
In e na ional Jou nal o Inno a i e Technology and Explo ing Enginee ing (IJITEE)
ISSN: 2278-3075 (Online), Volume-14 Issue-12, No embe 2025
38
Published By:
Blue Eyes In elligence Enginee ing
and Sciences Publica ion (BEIESP)
© Copy igh : All igh s ese ed.
Re ie al Numbe : 100.1/iji ee.D831014041125
DOI: 10.35940/iji ee.D8310.14121125
Jou nal Websi e: www.iji ee.o g
The cu en ex p e-p ocessing echniques employed in ou
esea ch can be cha ac e ised as es ablished me hods ha
unde pin con empo a y esea ch in in o ma ion e ie al and
linguis ic no malisa ion, as ou lined by Roul e al [13].
C. Fea u e Ex ac ion
Fea u e ex ac ion in ol es con e ing he cleaned ex in o
a s uc u ed nume ical o ma app op ia e o machine-
lea ning pa adigms. We used Te m F equency-In e se
Documen F equency (TF-IDF) o quan i y wo d impo ance
ac oss messages o unig ams and big ams. O he me a
ea u es included message leng h, wo d coun s, and he
de ec ion o spam wo ds such as “ ee,” “win,” o “u gen .”
Recen e m-weigh ing me hods demons a e supe io
e ie al and classi ica ion pe o mance compa ed o
adi ional models [14].
D. Classi ica ion Models
Th ee classi ica ion models a e used and e alua ed:
i. Nai e Bayes (NB): This p obabilis ic classi ie elies on
independence o e ms and is e y e icien o ex da a
[1].
ii. Suppo Vec o Machine (SVM): The SVM classi ie is
a disc imina i e model speci ically o high-
dimensional ea u es associa ed wi h ex da a [5].
iii. Ensemble Vo ing Classi ie : The andom
Fo es /Ensemble Vo ing classi ie ac s on bo h
classi ie s in pa allel using ha d- o ing o educe
a iance and bias in classi ica ion and ul ima ely
achie e a mo e eliable classi ica ion [4].
E. E alua ion Me ics
The models a e e alua ed using s anda d pe o mance
me ics:
i. Accu acy, P ecision, Recall, F1-sco e
ii. Con usion Ma ix ( o assess misclassi ica ions o he
esul s)
The e alua ions we e measu ed using a s a i ied 80/20
ain- es spli wi h a 10- old c oss- alida ion, o ensu e
s a is ical signi icance and gene alizabili y o he esul s [12].
V. PROPOSED SOLUTION
This esea ch a icle in oduces a hyb id spam de ec ion
amewo k ha u ilises bo h p obabilis ic and disc imina i e
lea ning. The hyb id schema uses an ensemble o ing
app oach o combine he Nai e Bayes and SVM models,
le e aging hei s eng hs while mi iga ing hei downsides.
Nai e Bayes is shown o handle ex ual ea u es e icien ly,
while SVMs p o ide obus disc imina i e bounda ies ac oss
complex ea u e dis ibu ions [5]. Combining a Nai e Bayes
and an SVM yields a hyb id model ha s ikes a be e
balance, minimising alse posi i es and he eby imp o ing
ecall on eal-wo ld da a [4].
A. Algo i hm Design – Hyb id Spam De ec ion
i. Load and examine he da ase .
ii. P e-p ocess ex (Lowe -case he ex , emo e
punc ua ion, emo e s op wo ds, and s em).
iii. TF-IDF Vec o isa ion & ex ac seconda y linguis ic
ea u es.
i . T ain NB and SVM on he aining da a.
. Combine he p edic ions om he NB and SVM using
ensemble o ing.
i. E alua e esul s using accu acy, p ecision, ecall, and
F1-sco e.
The hyb id model was de eloped o be ep oducible using
Py hon’s Sciki -lea n and NLTK lib a ies, and expe imen al
alida ion has shown i o p o ide mo e s able and accu a e
classi ica ions, which aligns wi h he indings o Yang e al
[4] and Lee e al [12]. The o e all wo k low o he p oposed
sys em is isually summa ised in Figu e 1.
B. Algo i hm Design
The algo i hm o de ec hyb id spam can be b oken down
in o he ollowing s eps:
Algo i hm: Hyb id_Spam_De ec o (SMS_da ase )
Inpu : Labelled SMS da ase (spam/ham)
Ou pu : Message labels p edic ed
Load da ase P ep ocess da a:
i. Lowe case ex .
ii. Remo e punc ua ion and s op wo ds.
iii. S em and no malise.
i . Ex ac ea u es using TF-IDF and auxilia y ea u es
(leng h, capi alisa ion, equency o keywo ds).
. Spli he da a in o aining (80%) and es ing (20%)
subse s.
i. Fi Nai e Bayes (NB) o aining da a.
ii. Fi SVM o he same da a.
iii. Vo ing mechanism:
▪ I NB and SVM bo h decla e he label spam, label i
spams.
▪ I NB and SVM p edic a di e en label, use he
highes con idence o assign a p edic ion.
▪ O he wise, label i hams.
▪ Assess pe o mance using o e all accu acy,
p ecision, ecall, F1 sco e, and a con usion ma ix.
C. Implemen a ion
Model implemen a ion is speci ied in Py hon using Sciki -
lea n o consis ency and ep oducibili y, wi h NLTK and
Pandas o ea u e ex ac ion and p ep ocessing. Expe imen s
showed ha hyb id ensemble models achie ed highe ecall
and F1 sco es han s and-alone NB and SVM classi ie s,
ein o cing he bene i s o combining disc imina i e and
p obabilis ic p ocedu es [4], [6]. The con usion ma ix (Table
1) p esen s he b eakdown o he pe o mance, including he
misclassi ica ions ( alse posi i es and alse nega i es). As
shown in Table 2, he hyb id model achie ed imp o ed
pe o mance ac oss classi ica ion me ics—accu acy,
p ecision, ecall, and F1 sco e—compa ed o he indi idual
Nai e Bayes and SVM classi ie s.
D. Flowcha o he P oposed Sys em
He e is a isual ep esen a ion o
he pipeline ( lowcha ):
Inno a i e Spam De ec ion Using Hyb id Machine Lea ning Algo i hms: A Da a-Cen ic App oach
39
Published By:
Blue Eyes In elligence Enginee ing
and Sciences Publica ion (BEIESP)
© Copy igh : All igh s ese ed.
Re ie al Numbe : 100.1/iji ee.D831014041125
DOI: 10.35940/iji ee.D8310.14121125
Jou nal Websi e: www.iji ee.o g
[Fig.1: Hyb id Spam De ec ion Model Wo k low]
E. Con usion Ma ix
Table I: Con usion Ma ix o Ensemble Classi ie
P edic ed Spam
P edic ed Ham
Ac ual Spam
470
20
Ac ual Ham
15
506
F. E alua ion Me ics
Table II: Model E alua ion Me ics
Model
Accu acy
P ecision
Recall
F1-Sco e
Naï e Bayes
95.60%
94.20%
93.50%
93.80%
SVM
96.10%
95.00%
94.10%
94.50%
Hyb id Ensemble
97.40%
96.50%
95.90%
96.20%
VI. CONCLUSION
This esea ch o e s a hyb id spam de ec ion amewo k
ha combines a Nai e Bayes and a Suppo Vec o Machine
spam classi ie in o a single model o enhance spam il e ing
pe o mance, obus ness, and s abili y ac oss di e en
da ase s. The esea ch inco po a es bo h p obabilis ic
in e ence and ma gin-based lea ning p inciples in o he spam
de ec ion amewo k, enabling a be e balance in
pe o mance ac oss a ious da ase s. Ou hyb id model was
also e alua ed on hese da ase s, which showed s a is ically
signi ican imp o emen s in accu acy, p ecision, ecall, and
F1-sco e o e single-classi ie baselines [10].
The hyb id app oach o spam de ec ion p oposed he e
ep esen s a i s s ep owa ds building scalable, eliable spam
de ec ion sys ems o eal- ime communica ion
en i onmen s. Using TF-IDF o ea u e ex ac ion and
ensemble lea ning as he o e all s a egy, his hyb id model
can e ec i ely add ess issues o da a imbalance, ea u e
edundancy, and ad e sa ial manipula ion.
None heless, some a enues o he u u e exis . In eg a ing
deep lea ning models, such as LSTMs o ans o me -based
a chi ec u es [11], is a clea oppo uni y o ele a e con ex ual
unde s anding and classi ica ion pe o mance. Pe haps
opening he da ase o a mul ilingual o ma o spam
messages would also assis gene alisa ion. Las ly, including
a eal- ime adap i e lea ning mechanism could help he
p oposed model adap o spam wi hou e aining once i is
achie ed.
In conclusion, he hyb id machine lea ning amewo k
p oposed in his s udy p o ides a eliable, compu a ionally
e icien , and ex ensible ounda ion o mode n spam
de ec ion applica ions, combining in e p e abili y wi h high
p edic i e accu acy.
DECLARATION STATEMENT
A e agg ega ing inpu om all au ho s, I mus e i y he
accu acy o he ollowing in o ma ion as he a icle's au ho .
▪ Con lic s o In e es / Compe ing In e es s: Based on
my unde s anding, his a icle has no con lic s o
in e es .
▪ Funding Suppo : This a icle has no been unded by
any o ganiza ions o agencies. This independence
ensu es ha he esea ch is conduc ed wi h objec i i y
and wi hou any ex e nal in luence.
▪ E hical App o al and Consen o Pa icipa e: The
con en o his a icle does no necessi a e e hical
app o al o consen o pa icipa e wi h suppo ing
documen a ion.
▪ Da a Access S a emen and Ma e ial
A ailabili y: The adequa e esou ces o his a icle a e
publicly accessible.
▪ Au ho ’s Con ibu ions: The au ho ship o his a icle
is con ibu ed equally o all pa icipa ing indi iduals.
REFERENCES
1. Alzah ani, A., e al., "Spam and Phishing De ec ion Using Machine
Lea ning," Compu e s, 2022.
DOI: h ps://doi.o g/10.3390/compu e s11050062
2. Kowsa i, K., e al., "Tex Classi ica ion Algo i hms: A Su ey,"
In o ma ion, 2019. DOI: h p://doi.o g/10.3390/in o10030150.
3. Akbik, A., e al., "Con ex ual S ing Embeddings o Sequence
Labelling," COLING 2018. URL: h ps://aclan hology.o g/C18-
1128.pd
4. Yang, X., e al., "Recen Ad ances o Spam De ec ion in Social
Ne wo ks Based on Machine Lea ning," Senso s, 2021.
DOI: h p://doi.o g/10.3390/s21020457
5. Zhang, H., "An SVM-Based Spam Fil e ing App oach wi h P io
Knowledge," IEEE Access, 2019.
DOI: h p://doi.o g/10.1109/ACCESS.2019.2926793
6. Zhang, L., Zhu, Y., & Yao, T., "Spam de ec ion using deep lea ning: A
su ey," IEEE Access, 2020.
DOI: h p://doi.o g/10.1109/ACCESS.2020.3026394
7. Minaee, S., Kalchb enne , N., Camb ia, E., e al. (2021). Deep lea ning-
based ex classi ica ion: A comp ehensi e e iew. ACM Compu ing
Su eys, 54(3), 1-40. DOI: h p://doi.o g/10.1145/3439726
8. Young, T., e al., "Recen ends in deep lea ning based na u al language
p ocessing," IEEE Compu a ional In elligence Magazine, 2018.
DOI: h p://doi.o g/10.1109/MCI.2018.2840738.
9. Ra i, R., and Ra ind an, B., "Spam de ec ion using Bayesian ne wo ks,"
In e na ional Jou nal o Compu e Applica ions, 2016.
DOI: h p://doi.o g/10.5120/ijca2016911820.
10. Na u e (2025). Key insigh s in o ecommended SMS spam de ec ion
da ase s. Na u e, Ma ch 2025.URL:
h ps://www.na u e.com/a icles/s41598-025-10562-y.
11. Young, S., e al., "Adap i e deep lea ning app oach o spam de ec ion,"
Jou nal o Cybe secu i y, 2025 ( o hcoming publica ion IJRTE).
12. Lee, J., e al., "An ensemble machine lea ning app oach o email spam
de ec ion," Expe Sys ems wi h Applica ions, 2024.
DOI: h p://doi.o g/10.1016/j.eswa.2023.120572.
13. Roul, R. K., Sahoo, J. K., & A o a, K. (2025). A Compa a i e S udy o
TF-IDF, Wo d2Vec, Fas Tex , BERT,
and GPT o Tex Rep esen a ion.
In e na ional Jou nal o Inno a i e Technology and Explo ing Enginee ing (IJITEE)
ISSN: 2278-3075 (Online), Volume-14 Issue-12, No embe 2025
40
Published By:
Blue Eyes In elligence Enginee ing
and Sciences Publica ion (BEIESP)
© Copy igh : All igh s ese ed.
Re ie al Numbe : 100.1/iji ee.D831014041125
DOI: 10.35940/iji ee.D8310.14121125
Jou nal Websi e: www.iji ee.o g
DOI: h p://doi.o g/10.1016/j.p ocs.2025.04.250.
14. Kuma , R., Sahoo, J. K., & A o a, K. (2021). Mode n e m weigh ing
schemes o ex classi ica ion. P ocedia Compu e Science, 186, 604-
613. DOI: h p://doi.o g/10.1016/j.p ocs.2021.04.050
AUTHOR’S PROFILE
S ee Vidya Venigalla is a Mas e 's s uden a Kone u
Lakshmaiah Educa ion Founda ion (KLEF), pu suing a
Mas e o Technology deg ee in Compu e Science
Enginee ing wi h a specialisa ion in Cybe secu i y and
Blockchain Technology. She has a s ong academic ocus
on A i icial In elligence, Cybe secu i y, and he Da k
Web. S ee Vidya is ac i ely in ol ed in esea ch p ojec s explo ing machine
lea ning applica ions o enhance secu i y and p i acy. He in e es s also
include s udying eme ging echnologies and hei implica ions in secu e
digi al en i onmen s. She aims o make signi ican con ibu ions o
ad ancing cybe secu i y and AI h ough inno a i e esea ch.
D . Venka a Du ga Ki an Kasula, is a P o esso in he
Depa men o Compu e Science and Enginee ing a
KLEF (Kone u Lakshmaiah Educa ion Founda ion), whe e
he has been se ing since Janua y 3, 2007. He holds a PhD,
M. Tech, and B.Sc. in Compu e Science, all awa ded by
Acha ya Naga juna Uni e si y, wi h dis inc ions in each
deg ee. Wi h 19 yea s o eaching expe ience, D Ki an has con ibu ed
signi ican ly o academia h ough eaching, men o ing, and esea ch
guidance. He has success ully supe ised 7 PhD schola s and 10
pos g adua e p ojec s, demons a ing his dedica ion o academic excellence
and s uden de elopmen . D Ki an has published 67 e e eed esea ch pape s
in epu ed Scopus- and Web o Science-indexed jou nals, along wi h 21 non-
e e eed publica ions and 17 con e ence p esen a ions a na ional and
in e na ional le els. His esea ch con ibu ions ha e ea ned him an h-index
o 9 and an i-10 index o 3. He has published ou books and iled h ee
pa en s, e lec ing his commi men o inno a ion and knowledge
dissemina ion.
Disclaime /Publishe ’s No e: The s a emen s, opinions and
da a con ained in all publica ions a e solely hose o he
indi idual au ho (s) and con ibu o (s) and no o he Blue
Eyes In elligence Enginee ing and Sciences Publica ion
(BEIESP)/ jou nal and/o he edi o (s). The Blue Eyes
In elligence Enginee ing and Sciences Publica ion (BEIESP)
and/o he edi o (s) disclaim esponsibili y o any inju y o
people o p ope y esul ing om any ideas, me hods,
ins uc ions o p oduc s e e ed o in he con en .
