← Back to Archives
VOLUME 15, ISSUE 9, SEPTEMBER 2026
Publication in progress..
A Scalable AI Recruitment Framework Using RAG and Explainable Multi-Agent Decision-Making
Kudupudi Rajesh
Computing Disciplines Competency Framework for the Digital Age
Bostley Muyembe Asenahabi
Lightweight and Scalable Real-Time Phishing URL Detection Models: A Comparative Review and Proposed Framework
Abhishek Kumar, Anand Kumar
Fusion of Fractal Color Watermarking with RSA
Maninder Kaur, Manish Sachdeva
Assessing AI Awareness, Utilization, and Perceived Benefits Among College Students at Wesleyan University Philippines
Kim Colene S. Dela Cruz and Jovith G. Nabua
Next Generation Authentication Using a Blockchain-Distributed Secure Ledger
Mrs. Manjula K, Niveditha G M
A Unified Analysis of Digital Forensics and Machine Learning in Cybercrime Investigation
Muzamil Amin, Yatham Thirumalanayudu
DESIGN AND DEVELOPMENT OF A FULL-STACK LEARNING MANAGEMENT SYSTEM WITH SECURE AUTHENTICATION, ONLINE PAYMENT, AND COURSE PROGRESS TRACKING
Akshata Mahishawadagi
A Survey on Handwriting Disorder Detection in Individuals with Disabilities
Ramya Deshinamoorthy, Dr.F.Kurus Malai Selvi
Lightweight and Explainable Real-Time Detection of Social Media-Borne Phishing URLS: A Critical Review for Indian Context
Pushpam Kumari
Adaptive Ensemble Machine Learning Framework for Robust Prediction, Classification, and Intelligent Decision Support Across Heterogeneous Data Domains
M. Sasikumar, Addepalli Keerthika, N Saranya
Next-Generation Deep Learning: A Comprehensive Survey on Explainable, Efficient, Privacy-Preserving and Multimodal Artificial Intelligence
K.Ramalakshmi, S.Prema, M.Ajithaveni, M.Ramya
AI-BASED STUDENT PREFORMANCE PREDICTION
Dr.B. Rajesh Kumar, S. Shailesh, Mohammed Aadhil.I
AI-Based Chatbot for Education: Design, Architecture, and Empirical Evaluation
Dr. B. Rajesh Kumar, Abishek.V, Sujan.R, Abishek N
YojanaMitra: An AI-Assisted Framework for Government Scheme Eligibility Matching and Guidance
Chinta Lavanya
Noble Poly Shapes: An AI-Assisted Interactive Interface for the Enumeration of Noble Polyhedra
Akshat Wahal, Anupam Andrew Maxton
Smart Classroom Management Software for Enhanced Learning Environments.
Ms. Sakshi Umakant Chaudhari, Asst. Prof. Sapana A. Fegade
ARMOR: An Extensible Multi-Provider Secret Validation Framework for Automated Credential Verification in DevSecOps Pipelines
Siddharth Gupta, Poonam Bhartiya & Shailendra Shriwastava
A. ISSN (O) 2278-1021, ISSN (P) 2319-5940IJARCCE XplainGuard: Dynamic Policy Enforcement in Cloud Infrastructure as Code: Integrating Generative AI for Auto-Remediation and Explainability
Aadrika Khare, Poonam Bhartiya & Shailendra Shriwastava
Enhancing the ELLANGA Traditional Cascade Tank-Village System in Sri Lanka: Proposal for a Smart IT Based Solution to Enhance Predictability and Sustainability
Rathnayake R.M.N.B Mahesh Hapugoda
Abstract
A Scalable AI Recruitment Framework Using RAG and Explainable Multi-Agent Decision-Making
Kudupudi Rajesh
Abstract: Recruitment processes increasingly rely on Artificial Intelligence (AI) to automate resume screening, candidate-job matching, skill-gap identification, and candidate evaluation. However, conventional AI and Large Language Model (LLM)-based recruitment systems may generate responses that are insufficiently grounded in the actual information contained in resumes and job descriptions, limiting their transparency, reliability, and explainability. This research presents a scalable Retrieval-Augmented Generation (RAG) framework for explainable AI-based recruitment that integrates semantic retrieval with LLM-based response generation. Candidate resumes and job descriptions are processed into overlapping text chunks and converted into semantic embeddings using all-MiniLM-L6-v2, which are indexed in ChromaDB for efficient similarity-based retrieval. For a given recruitment query, the system retrieves the most relevant evidence using different Top-K configurations (K ∈ {1, 3, 5, 10}) and provides the retrieved context to Qwen3-14B Instruct for evidence-grounded response generation. The framework supports resume-based question answering, job-requirement analysis, candidate-job matching, skill-gap identification, and evidence-based explanations. A recruitment benchmark of 500 queries, distributed across five categories (Resume Information, Job Requirements, Candidate-Job Matching, Skill-Gap Identification, and Explainability) is used to evaluate the framework. Retrieval performance is assessed using Precision@K, Recall@K, Mean Reciprocal Rank (MRR), and NDCG, while generated responses are evaluated using Context Precision, Context Recall, and Answer Relevance. The experimental analysis demonstrates that retrieval depth substantially influences response quality, with lower Top-K configurations providing stronger contextual precision and more consistent answer relevance. The proposed framework therefore provides a systematic approach for integrating retrieval, generation, and evidence tracing to support more context-aware, transparent, and explainable AI-assisted recruitment.
Keywords: Retrieval-Augmented Generation, Explainable AI, AI-Based Recruitment, Large Language Models, Semantic Retrieval.
Keywords: Retrieval-Augmented Generation, Explainable AI, AI-Based Recruitment, Large Language Models, Semantic Retrieval.
Abstract
Scalable and Quantum-Resistant Authentication for the Blockchain of Medical Things
Meruga Vijaya Sri
Abstract: The rapid proliferation of the Internet of Medical Things (IoMT) has revolutionized remote patient monitoring but introduces severe data privacy and security challenges. Integrating Blockchain technology (BIoMT) provides immutable auditing, yet classical cryptographic signatures (e.g., RSA, ECC) are highly vulnerable to the looming threat of quantum computing via Shor’s algorithm. Furthermore, storing individual digital signatures on a distributed ledger incurs massive storage overhead, limiting scalability. This paper proposes a Post-Quantum Aggregate Blind Signature (PQ-ABS) framework tailored for BIoMT. Our architecture leverages Module Learning With Errors (Module-LWE) to guarantee quantum-resistant security. To ensure patient privacy, we employ blind signatures that allow medical authorities to authenticate health data without exposing the plaintext payload. Signatures from multiple sources are mathematically aggregated into a single, constant-size proof. We validate the architecture by deploying the verification logic within a Hyperledger Fabric smart contract. Furthermore, we extend base security models by formalizing PQ-ABS in the Random Oracle Model and analyzing its resistance against Grover’s quantum search algorithm.
Keywords: Post-Quantum Cryptography, Internet of Medical Things, Blockchain, Hyperledger Fabric, Blind Signatures, Aggregate Signatures, Module-LWE
Keywords: Post-Quantum Cryptography, Internet of Medical Things, Blockchain, Hyperledger Fabric, Blind Signatures, Aggregate Signatures, Module-LWE
Abstract
Computing Disciplines Competency Framework for the Digital Age
Bostley Muyembe Asenahabi
DOI: 10.17148/IJARCCE.2026.15901
Abstract: The field of computing has rapidly evolved from the 1960s into distinct disciplines: Computer Science, Information and Communication Technology. While this diversity enriches the profession, it has also created a problem. The overlapping competencies and blurred boundaries make it difficult for learners to choose appropriate pathways, trainers to impart discipline-specific skills, and employers to interpret graduate profiles. This leads to mismatched expectations, skill gaps, and inefficiencies in the workforce, thus weakening the ability of computing education to meet organizational and societal demands. This study clarifies the uniqueness of each computing discipline, highlighting their philosophies, competencies, and graduate profiles. The study employed a desktop research method, drawing on secondary data from academic publications, professional guidelines, and industry reports. Sources were selected based on credibility, relevance, and recognition by established bodies. The analysis involved systematically reviewing literature to identify recurring themes, disciplinary distinctions, and competency frameworks. Sources were cross-checked for were synthesized by comparing perspectives across disciplines, highlighting overlaps, and distinguishing unique competencies. This triangulated approach ensured that conclusions were both academically robust and practically relevant. By distinguishing these pathways, the study ensures that learners acquire targeted skills aligned with career aspirations, trainers deliver curricula that reflect discipline-specific competencies, and employers can interpret graduate profiles accurately to reduce skill mismatches. Ultimately, society benefits from graduates whose competencies are tailored to solving real-world problems through computing technologies. This study contributes to bridging the gap between education and practice by offering clarity on computing disciplines. It empowers learners to make informed choices, trainers to design effective programs, employers to align talent with organizational needs, and policymakers to strengthen curricula that support national and global development. In doing so, the paper positions computing education as a driver of innovation, efficiency, and societal progress.
Keywords: computing disciplines, computing curriculum pathways, unique competencies, graduate profiles, digital transformation, ICT
Keywords: computing disciplines, computing curriculum pathways, unique competencies, graduate profiles, digital transformation, ICT
Abstract
Lightweight and Scalable Real-Time Phishing URL Detection Models: A Comparative Review and Proposed Framework
Abhishek Kumar, Anand Kumar
DOI: 10.17148/IJARCCE.2026.15902
Abstract: Phishing URL detection has advanced considerably through machine learning and deep learning, with recent published systems reporting accuracy figures that frequently exceed 95%. However, a close reading of this literature shows that high accuracy is often achieved using large feature sets, deep sequence architectures, or ensembles comprising a dozen or more classifiers — approaches that are difficult to justify on resource-constrained platforms such as browser extensions, mobile applications, or edge gateways, where real-time response and a small memory footprint matter as much as raw detection accuracy. This paper reviews six recent studies on machine-learning-based phishing URL detection, comparing their algorithms, feature counts, and reported accuracy, and identifies a consistent gap: none of the reviewed works measure deployment cost — inference latency, model size, or throughput under concurrent load — alongside their accuracy claims. Building on this analysis, the paper proposes a tentative methodology for a lightweight, scalable, real-time phishing-detection framework that restricts features to those computable from the URL string alone, benchmarks compact classical classifiers against the heavier baselines identified in the review, and treats latency, model size, and throughput as first-class evaluation metrics rather than afterthoughts. The proposed direction is intended to narrow the gap between laboratory-grade accuracy and field-deployable, real-time phishing protection.
Abstract
Fusion of Fractal Color Watermarking with RSA
Maninder Kaur, Manish Sachdeva
DOI: 10.17148/IJARCCE.2026.15903
Abstract: Now day’s security is an important aspect of life. We require security in every part of communication to transmit our information safely. Here our main aim is to secure the important information hidden in the images, while communicating we are hiding watermark inside the image by using RSA algorithm. To obtain very high image quality we use Fractal color image watermarking. Firstly we encrypt the watermark using RSA and then this encrypted watermark is used to hide inside the image using Fractal color image watermarking. To achieve watermarking we convert RGB into space. Also we are trying to achieve better parameters like PSNR (power signal to noise ratio), BER (Bit error rate) and MSE (Mean square Error).
Keywords: DCT, DWT, HVS, CIE, SCDFT etc
Keywords: DCT, DWT, HVS, CIE, SCDFT etc
Abstract
Assessing AI Awareness, Utilization, and Perceived Benefits Among College Students at Wesleyan University Philippines
Kim Colene S. Dela Cruz and Jovith G. Nabua
DOI: 10.17148/IJARCCE.2026.15904
Abstract: Artificial intelligence (AI) has become increasingly integrated into higher education, influencing how students access information, complete academic tasks, and support their learning. This study assessed the awareness, utilization,
respondents’ demographic profiles, identified AI tools used for studying and their purposes, and determined students’ levels of AI awareness, utilization, and perceived benefits. A quantitative research design was employed using a researcher-made, expert-validated, and pilot-tested survey questionnaire. Purposive sampling was used to select 129 undergraduate students from different departments who were familiar with AI tools. Data were analyzed using frequency, percentage, and weighted mean.Findings revealed that most respondents were female and fourth-year students. ChatGPT was the most commonly used AI tool, while writing assistance and research and information gathering were the leading purposes for AI use. Respondents demonstrated a high level of AI awareness, with an overall weighted mean of 3.57, indicating that they were fully aware of AI concepts, applications, limitations, and related concerns. AI utilization obtained an overall weighted mean of 3.19, indicating that students sometimes used AI, particularly for accessing AI tools and searching complex topics. Perceived benefits obtained an overall weighted mean of 3.23, interpreted as moderately agree, with reduced educational expenses, time savings, and improved productivity among the notable benefits. The study concludes that AI serves as a valuable academic support tool, although responsible utilization, critical evaluation of AI-generated information, and strengthened AI literacy remain important for students in higher education.
Keywords: Artificial Intelligence, AI Awareness, AI Utilization, Perceived Benefits, Higher Education, College Students, Academic Support
respondents’ demographic profiles, identified AI tools used for studying and their purposes, and determined students’ levels of AI awareness, utilization, and perceived benefits. A quantitative research design was employed using a researcher-made, expert-validated, and pilot-tested survey questionnaire. Purposive sampling was used to select 129 undergraduate students from different departments who were familiar with AI tools. Data were analyzed using frequency, percentage, and weighted mean.Findings revealed that most respondents were female and fourth-year students. ChatGPT was the most commonly used AI tool, while writing assistance and research and information gathering were the leading purposes for AI use. Respondents demonstrated a high level of AI awareness, with an overall weighted mean of 3.57, indicating that they were fully aware of AI concepts, applications, limitations, and related concerns. AI utilization obtained an overall weighted mean of 3.19, indicating that students sometimes used AI, particularly for accessing AI tools and searching complex topics. Perceived benefits obtained an overall weighted mean of 3.23, interpreted as moderately agree, with reduced educational expenses, time savings, and improved productivity among the notable benefits. The study concludes that AI serves as a valuable academic support tool, although responsible utilization, critical evaluation of AI-generated information, and strengthened AI literacy remain important for students in higher education.
Keywords: Artificial Intelligence, AI Awareness, AI Utilization, Perceived Benefits, Higher Education, College Students, Academic Support
Abstract
Next Generation Authentication Using a Blockchain-Distributed Secure Ledger
Mrs. Manjula K, Niveditha G M
DOI: 10.17148/IJARCCE.2026.15905
Abstract: The increasing demand for digital services has created an insatiable demand for authentication systems that are trustworthy, safe, and tamper-proof. Traditional centralized authentication systems could potentially expose the identity of users to the risks of data breaches, unauthorized modification, and insider attacks. In contrast, blockchain technology offers distributed data integrity and tamper-proof data. However, traditional blockchain-based authentication solutions could also add new difficulties in infrastructure, computation, and deployment. In this paper, the author presents a lightweight distributed authentication model that stems from the idea of blockchain and makes use of three synchronized MySQL databases. In the proposed model, AES-256 encryption is used to protect user information, while BCrypt is used to hash passwords, SHA256 is used for integrity verification, and majority voting is applied to distinguish rightful and faulty replicas. The proposed model is made even stronger owing to the use of JWT-based authentication and RBAC. In addition to that, the results of experiments that involved 1000 simulated users and 200 provoked tampering events are shown, as it refers to 100% tampering detection accuracy and consistency of the replication of data during the experiments. In addition to that, an average recovery time for resolving corrupted records proved to
Keywords: authentication based on blockchain technology; distributed ledger; AES-256 encryption; BCrypt hashing; SHA-256 integrity check; majority voting; tampering definition; backup recovery; JWT authentication; access control based on the role.
Keywords: authentication based on blockchain technology; distributed ledger; AES-256 encryption; BCrypt hashing; SHA-256 integrity check; majority voting; tampering definition; backup recovery; JWT authentication; access control based on the role.
Abstract
A Unified Analysis of Digital Forensics and Machine Learning in Cybercrime Investigation
Muzamil Amin, Yatham Thirumalanayudu
DOI: 10.17148/IJARCCE.2026.15906
Abstract: The rapid increase in cybercrime and the growing volume of digital evidence have made digital forensics an essential component of modern cyber investigations. Digital forensic tools play a crucial role in identifying, acquiring, preserving, analyzing, and presenting digital evidence from computers, mobile devices, cloud platforms, networks, and Internet of Things (IoT) environments. Over the years, numerous commercial and open-source forensic tools have been developed to address a wide range of investigative requirements. Meanwhile, machine learning (ML) techniques have emerged as effective approaches for improving the efficiency of forensic analysis through automated evidence classification, malware detection, anomaly detection, user behavior analysis, and event reconstruction. This paper presents an analysis of digital forensic tools, and their functions, and the application of machine learning techniques to digital evidence analysis throughout the investigation process. The study also examines the application of different ML algorithms across forensic domains based on the characteristics of the available evidence and extracted features. In addition, the study experimentally evaluates selected forensic tools and proposes a generalized architecture for integrating digital forensic tools with machine learning. The comparative analysis highlights how ML can assist in addressing challenges related to data volume, complexity, timeline analysis, evidence correlation, and analysis efficiency. The findings demonstrate that the integration of digital forensic tools with machine learning can enhance the efficiency and scalability of forensic investigations while maintaining the need for evidence validation and expert interpretation.
Keywords: Digital Forensics, Cybercrime Investigation, Digital Forensic Tools, Machine Learning, Digital Evidence, Forensic Analysis, Evidence Classification.
Keywords: Digital Forensics, Cybercrime Investigation, Digital Forensic Tools, Machine Learning, Digital Evidence, Forensic Analysis, Evidence Classification.
Abstract
Enhanced Intellects-Masses Optimizer Routing Strategy to improve the performance of MANET
Hamela K
DOI: 10.17148/IJARCCE.2026.15907
Abstract: Mobile Ad Hoc Networks (MANETs) consist of infrastructure-less, dynamically configured mobile nodes that communicate through wireless links. The frequent changes in network topology, limited energy resources, and vulnerability to misbehaving nodes pose significant challenges to reliable and secure routing. Optimized Link State Routing (OLSR) is a proactive routing protocol that employs Multipoint Relays (MPRs) to reduce routing overhead and improve communication efficiency. However, the selection of MPR nodes may result in excessive energy consumption, while misbehaving MPR nodes can drop data packets to conserve their energy, thereby degrading network performance and security. To address these challenges, this research proposes an Energy and Security-aware Optimized Link State Routing (ESOLSR) model based on an Enhanced Intellects-Masses Optimizer (EIMO). The proposed approach optimizes the MPR selection and route discovery process by considering available bandwidth, queue occupancy, and node lifetime as willingness parameters. In addition, misbehaving probability, power factor, and forwarding behavior are incorporated into a Composite Eligibility Index (CEI) to identify reliable and energy-efficient nodes for routing. The proposed EIMO- based routing mechanism aims to minimize energy consumption while improving network security and communication reliability. Simulation results demonstrate that the proposed ESOLSR approach outperforms existing energy-efficient routing schemes in terms of energy consumption, remaining network lifetime, average network lifetime, and variance of energy consumption. The results indicate that integrating security-aware node selection with optimization-based routing can significantly enhance the overall performance and reliability of MANETs.
Keywords: Mobile Ad Hoc Network, Optimized Link State Routing, Multipoint Relay, MPR, Energy Efficiency, Network Security, Enhanced Intellects-Masses Optimizer.
Keywords: Mobile Ad Hoc Network, Optimized Link State Routing, Multipoint Relay, MPR, Energy Efficiency, Network Security, Enhanced Intellects-Masses Optimizer.
Abstract
DESIGN AND DEVELOPMENT OF A FULL-STACK LEARNING MANAGEMENT SYSTEM WITH SECURE AUTHENTICATION, ONLINE PAYMENT, AND COURSE PROGRESS TRACKING
Akshata Mahishawadagi
DOI: 10.17148/IJARCCE.2026.15908
Abstract: This paper presents the design and development of a full-stack Learning Management System (LMS) intended to provide an integrated platform for course management, digital learning, authentication, online enrollment, payment processing, and learner progress tracking. The system is implemented using the MERN technology stack, with React.js and Vite on the client side, Node.js and Express.js on the server side, and MongoDB Atlas for persistent data storage. Clerk is used for user authentication, while Stripe supports online payments and webhook-based payment confirmation. Cloudinary is used for course thumbnail media storage. The platform supports educator-oriented course creation with chapters and lectures and student-oriented course discovery, purchase, enrollment, video learning, and progress management. RESTful APIs connect the client and server, while database models maintain course, purchase, enrollment, and progress information. Functional testing of the implemented workflow demonstrates that the major learning cycle— from authentication and course publication through payment, enrollment, lecture access, and progress tracking—can be handled within a unified web application. The work demonstrates how contemporary full-stack web technologies and third-party services can be combined to construct a practical and deployable LMS.
Keywords: Learning Management System, MERN Stack, React.js, MongoDB Atlas, Clerk Authentication, Stripe Payment, Webhook, Course Progress Tracking
Keywords: Learning Management System, MERN Stack, React.js, MongoDB Atlas, Clerk Authentication, Stripe Payment, Webhook, Course Progress Tracking
Abstract
A Survey on Handwriting Disorder Detection in Individuals with Disabilities
Ramya Deshinamoorthy, Dr.F.Kurus Malai Selvi
DOI: 10.17148/IJARCCE.2026.15909
Abstract: Handwritten writing is a complex biometric marker, which reflects both neurometer skill and cognitive processing abilities. The growing occurrence of writing and learning difficulties such as dysgraphia and dyslexia along with neurological conditions such as Parkinson's disease (PD), Multiple Sclerosis (MS), essential tremor (ET), and Alzheimer's disease (AD). There has been significant interest in automated handwritten based diagnostic and assessment systems. This research conducted a detailed systematic assessment of AI and Machine Learning approaches applied to handwritten analysis for prediction severity classification and monitoring of these issues. The comprehensive studies involve a broad range of approaches including CNN, transfer learning, ensemble methods, multimodal fusion frameworks, and data augmentation techniques. All analyses consistently compare the main Key performance measures including accuracy, precision, recall F1-score area under the curve of ROC sensitivity and specificity. The H2FCD model, CNN and SVM hybrid achieved highest accuracy of 99.26% and 99.33% as a result. The research gap is identified highlighting the lack of Multilanguage and multiclass severity datasets, absence of real world clinical deployment pipelines and the underrepresentation of diverse populations in training data. This review supports AI- based handwriting analysis for disability evaluation.
Keywords: Dysgraphia, Handwriting Analysis, Machine Learning, CNN, Feature Extraction
Keywords: Dysgraphia, Handwriting Analysis, Machine Learning, CNN, Feature Extraction
Abstract
Lightweight and Explainable Real-Time Detection of Social Media-Borne Phishing URLS: A Critical Review for Indian Context
Pushpam Kumari
DOI: 10.17148/IJARCCE.2026.15910
Abstract: Cybercrime in India has undergone a fundamental transformation in the last five years. Phishing, which was once limited to fraudulent emails, has now shifted aggressively to social media platforms such as WhatsApp, Telegram, Instagram and Facebook. Cybercriminals exploit trust by circulating deceptive links that promise free mobile recharge, government jobs, scholarship schemes, KYC updates and lottery prizes. Academic researchers have responded with a large number of machine learning based detection systems that report very high accuracy, often exceeding 98 percent. However, these systems are developed in isolation from real-world deployment constraints. Critical practical metrics such as inference latency, model file size, throughput and explainability for non-technical users are completely ignored. Most models depend on external services such as WHOIS lookup, DNS query and PageRank fetching, which require network calls and introduce significant delay. This paper presents a comprehensive critical theoretical review of six prominent recent studies published between 2025 and 2026.
Keywords: Phishing Detection, Social Media Security, Lightweight ML, Explainable AI, Real-Time Detection
Keywords: Phishing Detection, Social Media Security, Lightweight ML, Explainable AI, Real-Time Detection
Abstract
Adaptive Ensemble Machine Learning Framework for Robust Prediction, Classification, and Intelligent Decision Support Across Heterogeneous Data Domains
M. Sasikumar, Addepalli Keerthika, N Saranya
DOI: 10.17148/IJARCCE.2026.15911
Abstract: The increasing diversity and complexity of data across different application domains create significant challenges for conventional machine learning models in achieving robust prediction, classification, and intelligent decision support. This paper proposes an Adaptive Heterogeneous Dynamic Ensemble for Intelligent Decision Support (AHDE-DS) framework that integrates multiple heterogeneous machine learning and deep learning models within a unified adaptive architecture. The framework performs data preprocessing and feature representation, followed by parallel learning using complementary classifiers. A dynamic competence evaluation mechanism assesses each learner according to predictive performance, local competence, confidence, and uncertainty. Based on these measures, adaptive weights are assigned to the most competent models, and their outputs are fused through a stacking-based meta-learning mechanism. The final prediction is further processed by a confidence-aware decision-support layer. Experimental evaluation demonstrates that AHDE-DS achieves 98.73% accuracy, 98.61% precision, 98.54% recall, and 98.57% F1- score, outperforming the considered existing ensemble approaches. The results demonstrate the framework's effectiveness, adaptability, and robustness for heterogeneous data-driven intelligent applications.
Keywords: Adaptive Ensemble Learning, Heterogeneous Data, Dynamic Ensemble Selection, Machine Learning, Deep Learning, Stacking, Adaptive Weighting, Robust Prediction, Classification, Intelligent Decision Support
Keywords: Adaptive Ensemble Learning, Heterogeneous Data, Dynamic Ensemble Selection, Machine Learning, Deep Learning, Stacking, Adaptive Weighting, Robust Prediction, Classification, Intelligent Decision Support
Abstract
Next-Generation Deep Learning: A Comprehensive Survey on Explainable, Efficient, Privacy-Preserving and Multimodal Artificial Intelligence
K.Ramalakshmi, S.Prema, M.Ajithaveni, M.Ramya
DOI: 10.17148/IJARCCE.2026.15912
Abstract: The continuous evolution of deep learning has significantly expanded the capabilities of artificial intelligence, enabling intelligent systems to solve increasingly complex problems across healthcare, computer vision, natural language processing, cybersecurity, finance, autonomous systems, and smart environments. Beginning with artificial neural networks and progressing through convolutional and recurrent networks to Transformers, Graph Neural Networks, Vision Transformers, and Large Language Models, deep learning has achieved remarkable improvements in feature representation, prediction accuracy, and knowledge transfer. Nevertheless, the growing complexity of these models introduces several challenges, including limited model transparency, high computational and energy requirements, data privacy risks, and the effective utilization of heterogeneous multimodal information. Unlike conventional surveys that primarily classify deep learning according to network architectures or application domains, this work presents a capability-oriented taxonomy that organizes recent developments into five major research directions: Explainable Deep Learning, Efficient Deep Learning, Privacy-Preserving Deep Learning, Green Deep Learning and Multimodal Deep Learning. Based on this framework, representative architectures are critically examined with respect to their operating principles, strengths, limitations, and suitability for diverse real-world applications. The survey also analyses emerging application areas, identifies unresolved challenges related to fairness, robustness, scalability, and trustworthy artificial intelligence, and discusses promising research opportunities involving Federated Large Language Models, Edge AI, Green AI, Self-supervised Learning, and Multimodal Foundation Models. The proposed framework provides a structured understanding of current advances while offering practical insights for the design of transparent, efficient, secure, and sustainable next-generation deep learning systems.
Keywords: Deep Learning, Explainable Artificial Intelligence, Privacy-Preserving Learning, Green Deep Learning, Multimodal Learning, Foundation Models.
Keywords: Deep Learning, Explainable Artificial Intelligence, Privacy-Preserving Learning, Green Deep Learning, Multimodal Learning, Foundation Models.
Abstract
AI-BASED STUDENT PREFORMANCE PREDICTION
Dr.B. Rajesh Kumar, S. Shailesh, Mohammed Aadhil.I
DOI: 10.17148/IJARCCE.2026.15913
Abstract: Education is increasingly generating large amounts of digital data through examinations, assignments, attendance systems, online learning platforms, and internal assessments. This data can provide valuable information about student learning and academic progress.
Student Performance Prediction using Machine Learning is a system that analyzes historical student information and predicts the likely academic performance of students. The system can consider several factors, including attendance, previous examination marks, assignment performance, internal assessment marks, study hours, classroom participation, and other relevant attributes.
The proposed system consists of multiple stages, including data collection, data preprocessing, feature selection, model training, model testing, and performance prediction. Machine learning algorithms such as Decision Tree, Random Forest, Logistic Regression, and Support Vector Machine can be evaluated for the prediction task.
The system can help teachers identify students who may require additional academic support. It can also help educational institutions understand patterns associated with student achievement and improve academic planning.
However, predictions should be treated as supporting information rather than as a final judgment about students. Privacy, fairness, transparency, and human supervision are important when applying machine learning to educational data.
Keywords: Student Performance, Machine Learning, Educational Data Mining, Prediction, Classification, Academic Analytics, Data Science.
Student Performance Prediction using Machine Learning is a system that analyzes historical student information and predicts the likely academic performance of students. The system can consider several factors, including attendance, previous examination marks, assignment performance, internal assessment marks, study hours, classroom participation, and other relevant attributes.
The proposed system consists of multiple stages, including data collection, data preprocessing, feature selection, model training, model testing, and performance prediction. Machine learning algorithms such as Decision Tree, Random Forest, Logistic Regression, and Support Vector Machine can be evaluated for the prediction task.
The system can help teachers identify students who may require additional academic support. It can also help educational institutions understand patterns associated with student achievement and improve academic planning.
However, predictions should be treated as supporting information rather than as a final judgment about students. Privacy, fairness, transparency, and human supervision are important when applying machine learning to educational data.
Keywords: Student Performance, Machine Learning, Educational Data Mining, Prediction, Classification, Academic Analytics, Data Science.
Abstract
AI-Based Chatbot for Education: Design, Architecture, and Empirical Evaluation
Dr. B. Rajesh Kumar, Abishek.V, Sujan.R, Abishek N
DOI: 10.17148/IJARCCE.2026.15914
Abstract: Modern higher education is characterized by an exponential expansion of digital study materials; however, students continue to encounter friction in locating syllabus-aligned, context-aware academic resolutions in real time. Conventional academic support channels, such as scheduled office hours, physical libraries, and generic web search engines, are constrained by temporal availability, commercial clutter, and a lack of persistent pedagogical context. This paper presents the design, architectural formulation, and empirical evaluation of the AI-Based Chatbot for Education, an intelligent, interactive web platform engineered as an on-demand, non-judgmental digital academic companion. Built upon a modular stack incorporating Python 3.10+, Streamlit, and an embedded SQLite relational database, the system bridges conversational Natural Language Processing (NLP) models with structured syllabus navigation. The platform encapsulates four core modules: an AI Chat Module with session-aware state management, a Subjects Module for curriculum-scoped queries, a Study Tools Module for extractive summarization and synthetic quiz generation, and a Chat History Module providing persistent local auditability. Comprehensive verification across Unit, Integration, Functional, System, and User Acceptance levels demonstrated a 100% test pass rate across 45 executed test cases, maintaining sub-two-second query response latency and robust session recovery under adverse network conditions. Beyond the functional evaluation, this extended study reports a detailed breakdown of latency contributors, a structured usability survey administered to 42 undergraduate participants, a discussion of construct, internal, and external validity threats, and a roadmap for retrieval-augmented and multi-tenant extensions of the platform.
Keywords: Artificial Intelligence, Natural Language Processing, Educational Chatbot, Streamlit, SQLite, Pedagogical Assistants, Prompt Engineering, Human-Computer Interaction, Software Verification, Usability Evaluation.
Keywords: Artificial Intelligence, Natural Language Processing, Educational Chatbot, Streamlit, SQLite, Pedagogical Assistants, Prompt Engineering, Human-Computer Interaction, Software Verification, Usability Evaluation.
Abstract
YojanaMitra: An AI-Assisted Framework for Government Scheme Eligibility Matching and Guidance
Chinta Lavanya
DOI: 10.17148/IJARCCE.2026.15916
Abstract: Artificial Intelligence and Machine Learning are being utilised to automate eligibility assessment and provide personalisation in public service provision. They address longstanding issues faced by traditional systems such as fragmented scheme information, multi-condition eligibility criteria and limited citizen awareness. In this paper, we present YojanaMitra, a web application which provides access to 134 central and state government welfare schemes of India via a hierarchical and AI-driven matching pipeline. In the initial step, our matching pipeline uses a deterministic rule engine which supports both AND/OR logic of eligibility criteria to classify a citizen into three categories based on each welfare scheme: Eligible, Near Miss and Not Eligible. Only the schemes classified as Eligible are then ranked via a Random Forest classifier by engineering ten features from citizen-profile; finally, retrieval augmented conversational assistant provides citizen responses based on the matching scheme data alone. The application has been developed using Flask, SQLite database and Groq large language model API. It guarantees citizen security, structured profile-based matching and multilinguality. Automated testing shows accurate eligibility classification for all test cases; meanwhile, the accuracy of Random Forest ranking model is 0.597 with precision of 0.606, recall of 0.598, F1-score of 0.602 and ROC-AUC of 0.627 in the evaluation on held-out test set. We find that vulnerability and income margin are the two most important ranking features.
Keywords: Government Scheme Eligibility, Rule-Based Matching, Machine Learning Recommendation, Retrieval- Augmented Generation, Random Forest, E-Governance, Eligibility Prediction, Progressive Web Application, Flask, Multilingual Interface, User Engagement.
Keywords: Government Scheme Eligibility, Rule-Based Matching, Machine Learning Recommendation, Retrieval- Augmented Generation, Random Forest, E-Governance, Eligibility Prediction, Progressive Web Application, Flask, Multilingual Interface, User Engagement.
Abstract
Noble Poly Shapes: An AI-Assisted Interactive Interface for the Enumeration of Noble Polyhedra
Akshat Wahal, Anupam Andrew Maxton
DOI: 10.17148/IJARCCE.2026.15918
Abstract: We describe Noble Poly Shapes, a browser-based graphical interface for exploring, visualizing, and extracting the complete set of noble polyhedra enumerated by Hill [1]. Noble polyhedra are those polyhedra that are simultaneously vertex-transitive and face-transitive; Hill's computer-assisted proof establishes that, aside from the two known infinite families, exactly 146 such polyhedra exist. The accompanying data library, noble-tools-revised [2], contains the enumeration code and a full set of exported models. This document summarizes the relevant definitions and results of that work and describes the respects in which Noble Poly Shapes differs from the original repository: namely, in providing browser-native visualization, interactive exploration of the catalog, and direct extraction of shape and face data, without requiring the user to run the original enumeration software.
Abstract
Smart Classroom Management Software for Enhanced Learning Environments.
Ms. Sakshi Umakant Chaudhari, Asst. Prof. Sapana A. Fegade
DOI: 10.17148/IJARCCE.2026.15919
Abstract: Smart classroom technology is increasingly used to support teaching, attendance, communication, and classroom resources. However, many institutions still manage these activities through separate tools or manual procedures. This paper proposes Smart Classroom Management Software (SCMS) as a centralized platform for administrators, teachers, and students. The system combines attendance, timetable management, learning resources, announcements, student information, and basic analytics. Optional AI and IoT features can support attendance and classroom monitoring, while teacher verification and privacy controls remain important. Recent 2025–2026 research shows growing use of computer vision, IoT, AI-based behaviour analysis, and smart classroom interoperability. Based on this literature, the proposed system focuses on practical integration rather than a single feature. A design-oriented mixed- method approach is proposed, using questionnaires, interviews, task-based testing, and measures such as attendance accuracy, task completion time, usability, and response time. The paper presents a research-ready model and evaluation plan and does not invent experimental results that have not yet been collected.
Keywords: Smart classroom, classroom management software, artificial intelligence, Internet of Things, attendance, learning analytics, classroom monitoring, privacy, learning environment
Keywords: Smart classroom, classroom management software, artificial intelligence, Internet of Things, attendance, learning analytics, classroom monitoring, privacy, learning environment
Abstract
ARMOR: An Extensible Multi-Provider Secret Validation Framework for Automated Credential Verification in DevSecOps Pipelines
Siddharth Gupta, Poonam Bhartiya & Shailendra Shriwastava
DOI: 10.17148/IJARCCE.2026.15920
Abstract: Credential leakage through public repositories creates an urgent incident-response problem: detection tools identify candidate secrets but do not establish whether a credential remains active. This paper presents Armor, an open- source active credential-validation framework that issues read-only, idempotent probes to official provider endpoints. Armor supports seventeen credential types across cloud, source-code, AI/ML, communication, payment, and DevOps services, and returns a normalised five-state verdict schema: valid, invalid, no permission, failed validation, or nvalidjwt. Its five-layer pipeline combines a CLI, service facade, validator registry, data models, and provider-specific strategies, with an optional Groq LLaMA-3.3-70B triage agent for unstructured text. The local evaluation contains 126 passing unit tests, 100% branch coverage for tested validator modules, and 93.55% total package coverage.
Keywords: active credential validation, DevSecOps, secret scanning, cloud security, API authentication, JWT, Python.
Keywords: active credential validation, DevSecOps, secret scanning, cloud security, API authentication, JWT, Python.
Abstract
A. ISSN (O) 2278-1021, ISSN (P) 2319-5940IJARCCE XplainGuard: Dynamic Policy Enforcement in Cloud Infrastructure as Code: Integrating Generative AI for Auto-Remediation and Explainability
Aadrika Khare, Poonam Bhartiya & Shailendra Shriwastava
DOI: 10.17148/IJARCCE.2026.15921
Abstract: Infrastructure as Code (IaC) concentrates cloud security posture into text files, where one misconfigured attribute propagates to every environment where it is applied. Four gaps persist in prior tooling: static scanners such as Checkov and Trivy detect misconfigurations but generate no corrective code, leaving a detection-to-repair gap; generative repair methods report incorrect-code rates as high as 20.4%, precluding unattended use; automated fixes carry no security rationale, so practitioners cannot judge why a change is needed; and existing tools evaluate misconfigurations in isolation, missing compound attack paths formed by co-located resources. This paper presents XplainGuard, a command-line Terraform auditor that targets all four directly: model output is constrained to a strict JSON schema; every generated patch passes through a self-verification guardrail that executes terraform validate before the developer sees it, with one bounded correction attempt; each finding carries a structured four-element rationale (resource and attribute, attack vector, compound-risk context, compliance control). Across 21 runs on a seven-file fixture, every response parsed as valid JSON, and the guardrail reduced the observed incorrect code rate from 6.7% to 0%. Detection coverage was 100% on isolated vulnerabilities and 91.7% overall, and locally executed scans show Trivy 0.72.0 missing two of the four target categories that Checkov 3.3.8 detects, illustrating the specific gaps this work addresses across prior tooling rather than uniform superiority over any single tool. The evaluation is small; its statistical limits are stated explicitly.
Keywords: Infrastructure as Code, Terraform, dynamic policy enforcement, generative AI, large language models, auto- remediation, explainable AI, cloud security, self-verification.
Keywords: Infrastructure as Code, Terraform, dynamic policy enforcement, generative AI, large language models, auto- remediation, explainable AI, cloud security, self-verification.
Abstract
Enhancing the ELLANGA Traditional Cascade Tank-Village System in Sri Lanka: Proposal for a Smart IT Based Solution to Enhance Predictability and Sustainability
Rathnayake R.M.N.B Mahesh Hapugoda
DOI: 10.17148/IJARCCE.2026.15922
Abstract: The Ellanga Traditional Cascade -Village System (ETCTVS), practiced in Sri Lanka, becomes an important factor in the agriculture of this island nation through providing irrigation water to farmers and sustaining livelihoods of communities residing around these systems. However, its sustainability and stability are threatened as a result of specific factors such as climate change, water scarcity, and changing farming practices. The paper, therefore, discusses an agricultural information system application with regard to enhancing the sustainability and stability of the Ellanga Traditional Cascade -Village System (ETCTVS). In collecting data from farmers and other agents within the ETCTVS, a mixed-methods approach was used: survey, interview, and field observation. The information was later used in the GIS technology and analysis with the aid of a statistical analysis computer software. "This study shows that the implementation of an agricultural information system in the Ellanga Traditional Cascade -Village System (ETCTVS) will be very much efficient in making this system more stable and sustainable in nature. The study explains how, in this system, farmers will be able to have updated information with regard to the current status of climate, soil water content, crop handling practices, and market price of their products. The farmer will be able to access all this information sources which will aid them in taking prudent decisions in order to make maximum usage and achieve sustainability in agriculture," according to this research.
Keywords: Ellanga Cascade Water Tank System, IT Network System, Traditional Knowledge, Efficiency Enhancement
Keywords: Ellanga Cascade Water Tank System, IT Network System, Traditional Knowledge, Efficiency Enhancement