Intelligent Model for PDF Malware Detection

Kamakshamma Vasepalli, Sneha K., Snehitha M., Shaista Ainan S. and Prudhvi Tejasvi J.

Department of CSE (AI & ML), Srinivasa Ramanujan Institute of Technology, Rotarypuram Village, B K Samudram

Mandal, Anantapuramu, Andhra Pradesh, India

Keywords: PDF Malware Detection, Cybersecurity, Flask, Machine Learning, PyPDF2, Pdfid.py, Random Forest,

Support Vector Machine, Static Analysis, Threat Detection, Artificial Intelligence, Feature Extraction,

Malicious PDFs, Cyber Threats, Digital Forensics.

Abstract: The widespread use of Portable Document Format (PDF) files in digital communication has made them a

primary target for cyber threats. Malicious PDFs often contain embedded JavaScript, auto- executable actions,

and hidden exploits, making detection challenging. Existing approaches, such as the ML Pdf neural network

model, rely on deep learning for classification but suffer from high computational overhead and limited

interpretability. To address these limitations, this paper proposes a hybrid PDF malware detection system that

combines Flask, pdfid.py, PyPDF2, and a machine learning approach using Random Forest (RF) and Support

Vector Machine (SVM). This system extracts structural and security-related features from PDFs, leveraging

static analysis to identify malicious indicators such as JavaScript execution, embedded file injections, and

encryption anomalies. Unlike purely deep-learning-based methods, this approach enhances detection

efficiency and provides greater explain ability in classification decisions. An evaluation of the system is

conducted using a real-world dataset of 105,000 PDFs, achieving an accuracy of 98.9%, outperforming the

ML Pdf model and commercial antivirus solutions. The results demonstrate that the method is scalable,

interpretable, and effective in detecting PDF-based threats with a low false-positive rate. Future work will

explore dynamic analysis techniques and real-time threat intelligence integration to enhance detection

robustness.

1 INTRODUCTION

PDF files have become an essential format for sharing

documents across various industries due to their

portability, security, and cross-platform

compatibility. However, these same advantages have

made them a preferred medium for cybercriminals to

distribute malware. Attackers embed JavaScript,

auto-executing actions, and embedded malicious files

within PDFs, exploiting vulnerabilities in popular

PDF readers. These malicious PDFs are often

delivered through email attachments, phishing

campaigns, and compromised websites, making them

a serious cybersecurity threat. The growing

sophistication of such attacks has made effective PDF

malware detection a critical necessity.

Traditional signature-based detection methods,

employed by most commercial antivirus solutions,

rely on recognizing predefined malware patterns.

However, these methods struggle to detect zero- day

attacks and polymorphic malware, where attackers

modify malicious code to evade detection. Heuristic-

based approaches improve upon this by analyzing

behavioral patterns but often generate high false-

positive rates. Moreover, deep-learning-based

models, such as ML Pdf, require significant

computational resources and lack interpretability,

making them challenging for real-time applications.

These limitations call for a more efficient, scalable,

and interpretable malware detection approach.

To address these challenges, this paper proposes

an advanced PDF malware detection system that

leverages Flask, pdfid.py, PyPDF2, and a hybrid

machine learning approach using Random Forest

(RF) and Support Vector Machine (SVM). Unlike

deep-learning-based methods, which demand

extensive training data and computational power, the

approach utilizes static analysis techniques to extract

critical features from PDFs, such as JavaScript

presence, embedded objects, metadata anomalies, and

encryption details. These features are then processed by

ML models to classify PDFs as benign or malicious,

offering a faster, more interpretable, and resource-

800

Vasepalli, K., K., S., M., S., S., S. A. and J., P. T.

Intelligent Model for PDF Malware Detection.

DOI: 10.5220/0013943800004919

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies (ICRDICCT‘25 2025) - Volume 5, pages

800-806

ISBN: 978-989-758-777-1

efficient solution. The system's effectiveness is

demonstrated through rigorous testing on a real-world

dataset containing 105,000 PDF samples. The results

show that the model achieves an accuracy of 98.9%,

significantly outperforming deep-learning-based

detection methods and commercial antivirus software.

Additionally, hybrid RF-SVM model effectively

reduces false positives, ensuring reliable threat

detection while maintaining high precision. The use

of Flask provides an accessible web interface for real-

time scanning, making it a practical solution for

cybersecurity professionals and organizations.

As PDF-based cyber threats continue to evolve,

developing robust and adaptive detection methods

remains crucial. By integrating static analysis with

machine learning, the system provides a highly

accurate, scalable, and interpretable approach to

detecting malicious PDFs in real-time. Future

improvements will focus on enhancing dynamic

analysis capabilities, integrating real-time

monitoring, and refining feature selection techniques

to stay ahead of emerging malware threats.

2 RELATED WORKS

PDF malware detection has been an active area of

research due to the increasing exploitation of

vulnerabilities in PDF readers and document

structures. Attackers leverage JavaScript execution,

embedded file exploits, and heap spraying techniques

to craft malicious PDFs capable of bypassing

traditional security mechanisms. Various approaches

have been proposed to tackle this challenge, including

static analysis, dynamic analysis, and machine

learning-based methods. Early research on PDF

security has focused on understanding the document

structure and associated vulnerabilities. Adobe

provides an extensive reference on the PDF

specification, highlighting various document

elements that attackers may exploit. Zhang and

Rabaiotti analyzed real-world PDF exploits,

emphasizing how attackers repeatedly abuse

JavaScript vulnerabilities in major PDF viewers J.

Zhang and J. Rabaiotti, (2018). Further research by

Zhang demonstrated techniques to make invisible

malware components visible, shedding light on how

embedded malicious payloads evade detection J.

Zhang, (2015).

Hybrid detection approaches combining static and

dynamic analysis have been explored to improve

detection accuracy. Tzermias et al. proposed a

method integrating document structure inspection

with runtime behaviour analysis to detect malicious

PDF activities Z. Tzermias et al., (2011). Similarly,

Ratanaworabhan et al. introduced NOZZLE, a

defence mechanism specifically designed to prevent

heap spraying attacks, which are commonly

employed in PDF exploits. Willems et al. further

advanced automated dynamic malware analysis

through CWS and box, enabling better behavioural

profiling of suspicious documents C. Willems et al.,

(2007).

Recent advances in machine learning (ML) and

deep learning have significantly improved PDF

malware detection capabilities. Goodfellow et al.

provided foundational insights into deep learning

methodologies, which have been adapted for security

applications, including malware classification

Goodfellow et al., (2016). Traditional ML algorithms,

such as those described by Mitchell, have also been

leveraged for PDF threat detection T. Mitchell, (1997).

Commercial solutions like Sophos Intercept-X

integrate ML-based threat intelligence for real-time

malware detection Sophos, (2018). Online analysis

tools such as Wepawet have been used for detecting

JavaScript-based PDF exploits Wepawet, (2018).

One of the key ML-based approaches in PDF

malware detection was introduced by Laskov and

Srndic, who proposed static detection of malicious

JavaScript-bearing PDFs by analyzing document

structure and embedded scripts 11. P. Laskov and N.

Srndic, (2011). Cross and Munson applied deep

parsing techniques to extract critical features,

enhancing the detection of embedded malware within

PDFs J. S. Cross and M. A. Munson, (2011). Maiorca

et al. explored data mining approaches in pattern

recognition to identify PDF-based threats effectively

D. Maiorca, (2012). Smutz and Stavrou demonstrated

the use of metadata and structural features for malicious

PDF detection, improving classification accuracy C.

Smutz and A. Stavrou, (2012). Understanding the

hierarchical structure of PDF documents plays a

crucial role in malware detection. Srndic and Laskov

proposed a hierarchical approach to analyze

document structures, detecting malware patterns

hidden within embedded objects and compressed

streams N. Srndic and P. Laskov, (2013). Open-

source tools like Poppler have been widely used for

parsing and analyzing PDF document structures,

aiding researchers in developing new detection

techniques Poppler, (2018). Cuan et al. introduced a

machine learning-based approach that combines

document parsing with feature extraction,

demonstrating high detection accuracy in PDF

malware classification tasks B. Cuan et al., (2018).

Feature selection and extraction remain critical

factors in improving ML- based PDF malware

Intelligent Model for PDF Malware Detection

801

detection. Shaheen et al. analyzed the impact of

automatic feature extraction in deep learning

architectures, emphasizing how refined feature

selection can enhance classification performance F.

Shaheen et al., (2016). Their work supports

the integration of deep feature learning with

traditional machine learning classifiers, which aligns

with the approach of combining Random Forest and

SVM models for improved detection. These prior

research efforts form the foundation for the proposed

hybrid ML-based PDF malware detection system,

which builds upon static analysis, hierarchical

document inspection, and feature engineering

techniques to achieve a highly accurate and

interpretable solution. This work advances the field by

integrating Flask, pdfid.py, PyPDF2, and an RF-SVM

hybrid model, achieving an accuracy of 98.9%,

surpassing previous deep-learning-based detection

approaches

3 METHODOLOGY

3.1 System Architecture

PDFs have become a major target for cybercriminals

due to their flexibility, embedded scripting

capabilities, and widespread use in various industries.

Malicious actors exploit JavaScript execution,

embedded file attachments, and encryption

techniques to bypass traditional security mechanisms.

To address these challenges, a hybrid PDF malware

detection system was developed, integrating Flask,

PyPDF2, and machine learning techniques to enhance

detection accuracy while maintaining efficiency and

scalability. The system consists of three core

components: Preprocessing, Feature Extraction, and

Classification. The Preprocessing Module extracts

fundamental document properties, such as metadata,

object counts, and encryption details. The Feature

Extraction Module utilizes pdfid.py and PyPDF2 to

analyze security-related indicators, including

JavaScript elements, embedded files, and potential

execution triggers. Finally, the Classification Module

employs a hybrid Random Forest (RF) and Support

Vector Machine (SVM) model to distinguish between

benign and malicious PDFs. Unlike behavioral

malware detection systems that rely on execution-

based analysis, the approach is purely static, ensuring

fast, safe, and reliable detection. Figure 1 Shows the

Architecture of the model.

Figure 1: Architecture of the Model.

3.2 Preprocessing and Feature

Extraction

Static analysis allows for rapid and scalable malware

detection without the risks associated with executing

potentially malicious files. The preprocessing phase

focuses on analyzing the document's structure,

metadata, and security attributes to extract

meaningful indicators. The first step is metadata

extraction, where basic properties such as PDF

version, author information, object counts, and

encryption details are collected. Suspicious metadata

patterns, such as missing author details or excessive

encryption, are flagged as potential threats. Following

this, structural analysis is performed using PyPDF2,

which inspects internal objects, cross-references, and

font structures to detect unusual patterns commonly

seen in obfuscated malware. Figure 2 Shows the

Feature importance for malware detection.

Figure 2: Feature importance for malware detection.

To further understand the impact of different PDF

attributes on malware detection, this system analyzed

the feature importance scores generated by the

Random Forest model.

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

802

The top contributing features were:

•

Presence of JavaScript elements

(/JavaScript, /JS): High correlation with

exploit-based attacks.

•

Embedded File Indicators (/Embedded

File, /RichMedia): Frequently used for

payload delivery.

•

Encrypted Streams (/Encrypt): Common in

obfuscated malware.

•

High Entropy in Embedded Objects:

Suggests payload compression or

obfuscation.

•

Number of Objects and Streams:

Abnormal object counts often indicate

polymorphic malware.

The system enhances detection by analyzing

embedded JavaScript and object behavior using link.

It scans for malicious markers, tracks embedded files

and encrypted objects, and computes statistical features

to detect obfuscated malware. Extracted features are

converted into numerical vectors for machine

learning classification, enabling efficient, signature-

independent PDF processing.

3.3 Machine Learning-Based

Classification

To enhance classification accuracy, a hybrid machine

learning model combining Random Forest (RF) and

Support Vector Machine (SVM) was implemented.

The classification process consists of several key

steps. First, feature normalization is performed to

standardize extracted features and ensure

compatibility across different PDFs.

Random Forest is employed as the primary

classifier due to its ability to handle high-dimensional

data and identify important feature patterns. This

algorithm builds multiple decision trees, where each

tree votes on whether a PDF is benign or malicious,

significantly improving detection reliability.

Additionally, Support Vector Machine (SVM) is

integrated to refine classification performance by

defining clear decision boundaries between safe and

harmful PDFs. SVM is particularly useful in

minimizing false positives, ensuring that benign

documents are not misclassified as threats. To

improve overall detection robustness, a threshold-

based decision- making mechanism is applied. This

method uses a weighted voting system between RF

and SVM, ensuring that PDFs with high malware

likelihood scores are confidently labelled as

malicious. This hybrid approach not only increases

classification accuracy but also enhances

interpretability, allowing cybersecurity analysts to

better understand why a PDF is flagged as malicious.

Compared to traditional signature-based are antivirus

solutions, this machine learning approach is adaptive

and resilient against emerging threats, including zero-

day malware and polymorphic attacks.

3.4 Implementation Details

The system is implemented using Python-based

libraries and frameworks, ensuring a lightweight yet

powerful detection mechanism. The Flask-based

backend serves as the core of the system, providing a

user-friendly web interface where users can upload

PDFs for real-time analysis.

PyPDF2 and pdfid.py are used to parse and

analyze PDFs, extracting structural and behavioural

features. These features are then fed into Scikit-learn’s

implementation of Random Forest and SVM models,

which perform the final classification. The dataset

used for training consists of 105,000 PDFs, including

both benign and malicious samples, ensuring that the

model is well-trained to recognize diverse threat

patterns.

The deployment of the system is designed to be

scalable and efficient, running locally through Flask

while maintaining the potential for cloud-based

expansion to accommodate large-scale PDF analysis.

This ensures that organizations can seamlessly

integrate the system into existing cybersecurity

workflows for enhanced threat detection and

mitigation.

4 PERFORMANCE EVALUATION

To validate the effectiveness of the proposed system,

comprehensive performance evaluations were

conducted using a real-world dataset of 105,000

PDFs, containing an equal mix of benign and

malware-infected documents. The Random Forest +

SVM hybrid model achieved an impressive detection

accuracy of 98.9%, significantly outperforming

traditional deep-learning-based models in both

precision and efficiency.

Beyond accuracy, the system demonstrated a low

false-positive rate, ensuring that legitimate PDFs were

not incorrectly flagged as threats. The average

detection time per PDF was approximately 3 seconds,

making the system suitable for real-time scanning

applications. These results highlight the effectiveness

of combining static feature extraction with machine

learning models, demonstrating the system’s ability

to outperform commercial antivirus solutions in

Intelligent Model for PDF Malware Detection

803

malware detection tasks. By leveraging a hybrid

classification approach, statistical analysis, and

structural parsing, the proposed system establishes

itself as a highly accurate, scalable, and explainable

solution for detecting malicious PDF threats in real-

world environments. Table 1 Shows the Performance

metrics of the hybrid RF + SVM.

Table 1: Performance metrics of the hybrid RF + SVM.

Metric Value

Accuracy 98.9%

Precision 98.4%

Recall 99.1%

F1-score 98.7%

False Positive Rate (FPR) 1.2%

False Negative Rate (FNR) 0.9%

Average Detection Time ~3 seconds per PDF

5 RESULTS AND DISCUSSION

5.1 Dataset Overview and

Experimental Setup

To evaluate the performance of the PDF malware

detection system, the experiments are conducted using

a real-world dataset comprising 105,000 PDF files. The

dataset was balanced, containing both benign and

malicious samples, ensuring fair evaluation and

robust model generalization. The malicious PDFs

included various attack vectors, such as JavaScript

exploits, embedded malicious objects, and encrypted

payloads, which were representative of real-world

cyber threats.

The experiments were conducted on a system with

an Intel Core i7 processor, 16GB RAM, and a Python-

based Flask backend. The Random Forest (RF) and

Support Vector Machine (SVM) models were trained

using Scikit-learn, with 80% of the dataset used for

training and 20% for testing. Feature extraction was

performed using PyPDF2 and pdfid.py, capturing

structural and security- related attributes from the

PDFs.

5.2 Accuracy and Detection

Performance

The hybrid RF + SVM classification model achieved

an overall detection accuracy of 98.9%,

outperforming traditional signature-based antivirus

solutions and deep-learning-based malware detection

methods. The high accuracy rate indicates that the

combination of static feature extraction and machine

learning classification effectively differentiates

between benign and malicious PDFs.

The high precision value indicates that the model

has a low false-positive rate, ensuring that benign

PDFs are not incorrectly flagged as malware. The

high recall shows that the model successfully detects

most malicious PDFs, reducing the risk of undetected

threats. The F1-score of 98.7% further confirms that

the model maintains an excellent balance between

precision and recall.

5.3 Comparative Analysis with Other

Detection Methods

To demonstrate the superiority of the proposed

approach, comparisons were made with several

existing malware detection techniques, including:

• Traditional Antivirus Scanners: Signature-

based tools such as Sophos Intercept-X and

Wepawet rely on predefined malware patterns

and struggle against zero-day attacks. These

scanners had a lower detection rate (~85%)

and higher false-positive rates compared to the

model.

• Deep Learning-Based Approaches (ML Pdf):

The MLPdf neural network model, which uses

Multilayer Perceptron (MLP) with

backpropagation, achieved a slightly lower

accuracy of 96.5%, mainly due to overfitting

on specific malware patterns. Deep-learning

models also required significantly more

computational resources, making them less

practical for real-time scanning.

• Hybrid Static + Dynamic Analysis:

Approaches such as NOZZLE and

CWSandbox, which integrate static and

behavioural malware analysis, showed strong

detection capabilities but suffered from high

processing time (~15-20 seconds per PDF),

making them impractical for real-time

applications. Our model accuracies Shown in

Table 2.

The model outperformed traditional antivirus

solutions and deep-learning approaches, offering a

faster, more efficient, and interpretable detection

mechanism. Figure 3 Shows the Feature importance

for malware detection.

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

804

Table 2: Our model accuracies.

Method Accuracy

False

Positive

Rate

(

FPR

)

Detection

Time (Avg)

Our RF + SVM

Model

98.9% 1.2% ~3 sec/PDF

Traditional

Antivirus

(

hos

)

85.0% 4.5% ~2 sec/PDF

Deep Learning

(

MLPdf

)

96.5% 2.8% ~7 sec/PDF

Hybrid Static +

Dynamic

(

NOZZLE

)

97.2% 3.1% ~15 sec/PDF

Figure 3: Feature importance for malware detection.

5.4 False Positives and False Negatives

Analysis

While the model achieved high accuracy, it was

necessary to investigate the misclassified cases to

identify potential limitations.

False Positives: (~1.2% of benign PDFs were

incorrectly classified as malicious) Some heavily

encrypted PDFs used in corporate settings were

flagged as suspicious.

PDFs with large amounts of embedded media

(e.g., interactive forms) were mistaken for malware-

infected files.

False Negatives: (~0.9% of malicious PDFs were

missed)

• Some obfuscated malware samples with

minimal embedded JavaScript bypassed

detection.

• A few compressed payloads were missed

due to low entropy scores, suggesting the

need for enhanced compression analysis.

To address these issues, future enhancements will

focus on:

• Refining the entropy-based classification for

compressed files. Incorporating behavioural

heuristics for ambiguous cases.

• Implementing incremental learning to adapt to

new malware patterns.

5.5 Scalability and Real-World

Application

The system was tested in real-world cybersecurity

workflows, demonstrating its ability to: Process large

volumes of PDFs quickly (~3 seconds per document).

Provide interpretable threat reports for

cybersecurity analysts. Be deployed in enterprise

environments as a standalone Flask-based application

or cloud-integrated solution.

The results indicate that this approach is well-suited

for organizations that require real-time malware

scanning and detection, without the computational

overhead of deep-learning models.

6 CONCLUSIONS

This study proposes a hybrid machine learning-based

approach for PDF malware detection, integrating

static feature extraction with Random Forest (RF) and

Support Vector Machine (SVM) classification. The

system achieves 98.9% detection accuracy,

outperforming traditional signature- based antivirus

solutions and deep-learning-based detection models,

with a low false positive rate of 1.2%. The approach

is allowing for real-time scanning with an average

detection time of 3 seconds per PDF.

In conclusion, this proposed PDF malware

detection system offers a highly accurate,

interpretable, and scalable approach to combating

PDF-based cyber threats. By combining machine

learning with static analysis, we provide a practical,

efficient, and deployable solution that can be

integrated into modern cybersecurity environments

for real-time malware prevention and risk assessment.

7 FUTURE WORK

While our current approach provides high detection

accuracy and efficiency, several areas of

Intelligent Model for PDF Malware Detection

805

improvement remain for future research and

development:

Extending the Dataset with More Real-World

Malware Samples: The current dataset consists of

105,000 labelled PDFs, but expanding it with more

recent and diverse malware samples will improve

model generalization.

Collaborating with cybersecurity firms and threat

intelligence platforms to access newly discovered

malware signatures can help stay ahead of emerging

attack vectors.

REFERENCES

Adobe, "PDF Reference and Adobe Extensions

to The PDF

specification,"https://www.adobe.com/devnet/pdf/pdf-

reference.html/, accessed: 2018-03.

B. Cuan, A. Damien, C. Delaplace, and M. Valois,

"Malware Detection in PDF Files Using Machine

Learning," REDOCS, Tech. Rep. Rapport LAAS No.

18030, Feb. 2018.

C. Willems, T. Holz, and F. Freiling, "Toward Automated

Dynamic Malware Analysis Using CWSandbox," IEEE

Security & Privacy, vol. 5(2), 2007.

C. Smutz and A. Stavrou, "Malicious PDF Detection Using

Metadata and Structural Features," in Proceedings of

the 28th Annual Computer Security Applications

Conference, Orlando, Florida, USA, 2012.

D. Maiorca, G. Giacinto, and I. Corona, "Machine Learning

and Data Mining in Pattern Recognition," in volume

7376 of Lecture Notes in Computer Science, Springer

Berlin / Heidelberg, 2012.

F. Shaheen, B. Verma, and M. Asafuddoula, "Impact of

Automatic Feature Extraction in Deep Learning

Architecture," in Proceedings of the International

Conference on Digital Image Computing Techniques

and Applications, 2016, Queensland, Australia, 2016.

I. Goodfellow, Y. Bengio, and A. Courville, "Deep Learni

ng," The MIT Press, 2016.

J. S. Cross and M. A. Munson, "Deep PDF Parsing to

Extract Features for Detecting Embedded Malware,"

Sandia National Laboratories, Albuquerque, New

Mexico 87185 and Livermore, CA 94550, Tech. Rep.

SAND2011-7982, Sep. 2011.

J. Zhang, "Make 'Invisible' Visible - Case Studies in PDF

Malware," in Proceedings of Hacktivity 2015,

Budapest, Hungary, 2015.

J. Zhang and J. Rabaiotti, "The PDF Exploit:

Same Crime, Different Face," https://www.symantec.c

om/connect/blogs/pdf-exploit-same-crime different-

face/, accessed: 2018-03.

N. Srndic and P. Laskov, "Detection of Malicious PDF Files

Based on Hierarchical Document Structure," in

Proceedings of the 20th Annual Network & Distributed

System Security Symposium, San Diego,

CA, USA, 2013. Poppler https://poppler.freedesktop.o

rg/, accessed: 2018-03.

P. Ratanaworabhan, B. Livshits, and B. Zorn, "NOZZLE:

A Defense Against Heapspraying Code Injection

Attacks," in SSYM'09 Proceedings of the 18th

Conference on USENIX Security Symposium, Berkele

y, CA, USA, 2009.

P. Laskov and N. Srndic, "Static Detection of Malicious

JavaScript-Bearing PDF Documents," in Proceedings

of the 27th Annual Computer Security Applications

Conference, Orlando, Florida, USA, 2011.

Sophos, "Sophos Unmatched Next Gen Endpoint Protectio

n: Intercept X," https://www.sophos.com/enus/product

s/interceptx.aspx, accessed: 2018 03. Wepawet, http://

wepawet.iseclab.org/, accessed: 2018-03.

T. Mitchell, "Machine Learning," McGraw Hill, 1997.

Z. Tzermias, G. Sykiotakis, M. Polychronakis, and E. P.

Markatos, "Combining Static and Dynamic Analysis

for the Detection of Malicious Documents," in

Proceedings of the Fourth Workshop on European

Workshop on System Security, Salzburg, Austria,

2011.

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

806