Asian Journal of Research in Computer Science
Not a member yet
792 research outputs found
Sort by
Cybersecurity Gaps in Digital Epidemiology: Safeguarding Medical Surveillance in the Age of AI and Global Pandemics
No full textDigital epidemiology leverages real-time data and artificial intelligence (AI) to monitor and predict disease trends. However, the growing integration of public health surveillance and digital technology introduces significant cybersecurity vulnerabilities. This review critically examines the current gaps in cybersecurity within digital epidemiology, emphasizing threats from AI-driven analytics, regulatory fragmentation, and challenges disproportionately affecting low- and middle-income countries (LMICs). To guide mitigation strategies, we propose a layered socio-technical framework comprising three interconnected domains: (1) technological safeguards (e.g., secure AI architectures and data encryption), (2) ethical and governance mechanisms (e.g., consent, transparency, surveillance accountability), and (3) legal and institutional coordination (e.g., harmonized international regulations and LMIC capacity building). By applying this framework, we evaluate current practices and outline integrative recommendations to enhance resilience, equity, and trust in digital disease surveillance systems
Hybrid Models for Retail Demand Forecasting: Integrating Classical Time-series and Machine Learning Approaches
No full textThis paper presents an idea and implementation of using hybrid models in forecasting retail demand by integrating conventional approaches with machine learning. The objectives are to develop methodologies for building composite structures, sequential residual models, group stacking/blending, and orderly mixtures, exemplified by combinations like ARIMA–LSTM and Prophet–LightGBM, and to explore their primary pros and cons in meeting the changing demands of retail forecasting. The present work gains critical importance against a backdrop of markedly increased unpredictability in human behaviour—manifested by rapid sales fluctuations and heightened pressures throughout global supply chains. In such a context, enhancements in forecasting accuracy become directly linked to both revenue optimisation and the reduction of safety-stock expenditure. A critical appraisal of over twenty scholarly sources underpins the novelty of this research, which proposes a comprehensive methodological framework for organising the MLOps lifecycle in hybrid system architectures.The study justifies the separation of linear and nonlinear forecast components, which enables a reduction of MAPE by double-digit percentages and a decrease in error variance for cold SKUs while preserving the interpretability of the statistical component. The main conclusions are as follows: hybrid models demonstrate double-digit reductions in MAPE, RMSE, and WRMSSE compared to standalone algorithms. For practitioners, the resulting modular architecture simplifies maintenance and offers a scalable solution, while integrated CI/CD and CT loops ensure reliability and rapid response to data drift. The modularity of the architecture simplifies maintenance and scaling.
Additionally, CI/CD and CT loops ensure reliability and rapid response to data drift. The key factor for success is the balance between explainability, response time, and the incorporation of spatial correlations. This paper will be useful for researchers and practitioners in Data Science, MLOps engineers, analysts, and operational managers of retail chains
Explainable AI for Breast Cancer Diagnosis: Comparative Analysis of ML Models Using Random Forest Feature Selection and SHAP Interpretability
No full textBreast cancer diagnosis is critical for improving patient outcomes, yet traditional methods face limitations such as invasiveness and human error. This study presents an explainable AI framework for breast cancer classification using six ML models: LR, NB, KNN, RF, SVC, and DT. SMOTE addresses class imbalance, while RF feature selection reduces dimensionality from 30 to 19 features. SHAP interpretability is integrated to provide clinical insights into feature contributions, enhancing trust in model predictions. The SVC model with RF-selected features achieves superior performance, with an accuracy of 0.9930 and recall of 1.0000, highlighting the importance of features such as smoothness mean. This framework balances accuracy, efficiency, and transparency, offering a foundation for clinical deployment and guiding future work on external validation and broader adoption of explainable ML in breast cancer care
AI-Driven Financial Fraud Detection System in Savings Account Using Rule-Based Logic and Random Forest Algorithm
No full textThe threats of fraud in savings accounts are increasingly escalating in recent years due to the rise of digital banking, prompting the need for advanced security measures. This research addresses this challenge by developing a hybrid fraud detection system which combines rule-based logic with machine learning. This system aims to detect unauthorized transactions in real-time while minimizing re-occurrences and ensuring rapid response to emerging threats with key features suh as transaction pattern analysis and anomaly. The methodology used leverages Python for backend development, Scikit-learn for Machine Learning (ML) models, PHP (Hypertext-Preprocessor) for both frontend and also server-side scripting to create an interactive dashboard, and MySQL for database management to store and retrieve transaction data efficiently. Over 40,000 transactions was processed with 5% labeled as fraudulent and test results metrics with 0.8819 accuracy, 0.8805 precision, 0.9633 recall, 0.8810 ROC-AUC, 0.8518 PR-AUC and 0.9203 F1-Score. The high accuracy and recall suggest the hybrid approach effectively detects fraud in savings accounts. This work contributes to the field of financial cybersecurity by bridging the gap between static rule-based systems and adaptive machine learning approaches, offering a robust framework for safeguarding savings accounts against evolving fraudulent activities
Strengthening Cyber Resilience: Policy and Governance Responses to Global Cyber Threats
No full textThis study investigates policy responses to strengthening cyber resilience in the face of expanding global threats, using U.S.–China cyber dynamics as a reference case. Four open datasets were employed: Verizon Data Breach Investigations Report (DBIR), CISA Known Exploited Vulnerabilities (KEV) Catalogue, ENISA Threat Landscape Reports, and the (ISC)² Cybersecurity Workforce Study to generate quantitative insights that connect technical evidence to policy outcomes. The analysis applied ARIMA modelling for forecasting incident trends, Pareto concentration analysis for identifying critical vulnerability clusters, and a Difference-in-Differences regression to evaluate the effectiveness of the European Union’s NIS2 directive. Findings reveal that state-sponsored cyber incidents have risen sharply from 130 in 2013 to a projected 535 by 2027, with 20% of technologies accounting for over 70% of exploited vulnerabilities. The NIS2 directive demonstrated an 18% reduction in post-policy incidents across the EU compared to continued increases in the United States, underscoring the value of structured regulatory intervention. Additionally, the study emphasises Zero Trust architecture as a cornerstone of modern resilience, highlighting its potential to contain adversarial movement and reduce systemic exposure. The results provide actionable insights for policymakers, regulators, and security leaders seeking evidence-based approaches to improve governance, reporting frameworks, and international coordination. Ultimately, this research advances the understanding of how data-driven policy design anchored in Zero Trust principles and cross-jurisdictional governance, such as NIS2, can reinforce global cyber stability and foster a more resilient digital ecosystem
Optimizing Online Educational Experiences through Semantic Recommender Systems: A Review
No full textSemantic Web technologies in providing personalized recommendations to learners in digital educational environments. It involves applying advanced methods like Resource Description Framework (RDF), Web Ontology Language (OWL), and SPARQL query language, and machine learning to understand the meaning and context of learning materials and to align recommendations with individual learner goals, preferences, and knowledge gaps. The rapid expansion of online education platforms has introduced new challenges in delivering personalized and effective learning experiences to diverse learners. Traditional recommender systems, lack the sophistication needed to tailor educational content to the unique goals, cognitive abilities, and knowledge gaps of individual learners. This paper explores the potential of semantic recommender systems in enhancing online educational experiences by employing semantic web technologies such as ontologies, knowledge graphs, and machine learning. We review recent studies and implementations of semantic technologies within Learning Management Systems (LMS), showcasing how they contribute to more adaptive and engaging learning environments. Future directions for research emphasize the need for scalable, adaptive ontologies, and cross-platform integration to further enhance the personalization and effectiveness of online learning environments
Comparative Evaluation of Cloud-Native and VM-Based CI/CD Pipelines for Automated DevOps Deployments
No full textThe proliferation of DevOps methodologies has fundamentally reshaped the software development lifecycle, establishing Continuous Integration and Continuous Delivery/Deployment (CI/CD) as the central mechanism for achieving velocity and quality. This paradigm shift has given rise to two dominant architectural approaches for pipeline implementation: traditional, on-premise or IaaS-based Virtual Machine (VM) environments, and modern, fully managed cloud-native platforms. Each approach presents a distinct set of architectural, performance, economic, and security profiles that profoundly impact an organization\u27s ability to deliver software effectively. This paper presents a comprehensive comparative analysis of these two paradigms. It introduces a quantitative framework for evaluation based on the industry-standard DevOps Research and Assessment (DORA) metrics and a detailed Total Cost of Ownership (TCO) model. Furthermore, it proposes a novel, weighted decision-making algorithm designed to guide practitioners and organizational leaders in selecting the optimal environment tailored to their specific strategic context, operational capabilities, and financial constraints. The analysis reveals a fundamental and recurring trade-off between the granular control, full-stack responsibility, and capital-intensive nature of VM-based environments versus the dynamic scalability, operational expenditure-driven model, and shared-responsibility security posture of the cloud. The findings indicate that while VM-based pipelines offer unparalleled customization, cloud-native architectures are more aligned with the core DevOps principles of speed, resilience, and elasticity, thereby enabling organizations to achieve elite levels of performance. This research provides a structured foundation for strategic decision-making in the critical domain of DevOps infrastructure
Patterns of Mobile Awareness and Security Practices: A Clustering Analysis on College Faculty and Students
No full textCyber malware attacks, coupled with a lack of digital learning skills, can create significant issues in educational spheres where digitalization is in its infancy. This research integrates Educational Data Mining (EDM) methodologies to address the lack of actionable, predictive analytics in these environments. This research must (1) determine the malware awareness level and the demographic characteristics of the students and staff, (2) determine the malware awareness and the corresponding defensive strategies, (3) utilize cluster analysis in the detection of trends relative to protection and awareness, and (4) identify and describe the significant deficiencies in malware awareness and in the respective countermeasures. Using Google Colab and Python, survey data were cleaned and preprocessed, and a K-Means clustering analysis was performed, with 180 faculty and 188 students as respondents in the study. The analysis initially showed three different profiles in cybersecurity awareness: one cluster showed high awareness of traditional malware, while social engineering was less known, a second cluster had high knowledge of phishing, smishing, and vishing, accompanied by negative security; and a third cluster showed equilibrium in awareness but a lack of enacted security. The ANOVA test showed significant differences across all variables of malware awareness and security practices (p < .05), especially for social engineering threats, which were the highest. There were also significant gaps in the use of basic security practices, such as antivirus software, safe browsing, and keeping systems updated, indicating a lack of alignment between knowledge and practice. The results show significant gaps in defending against cyberattacks, as well as strong practices within the educational industry. This research elaborates on the need for the design of specific, targeted, revised training programs that address the specific goals of the SDGs for Quality Education, Industry Innovation, and Infrastructure. The silos of personalized adaptive responses and the flows of information regarding emerging threats must tap into educators\u27 and learners\u27 potential to amplify Digital Resilience
Modern Claims Management: A Comparative Analysis of Guidewire ClaimCenter and Duck Creek Claims
No full textAims: This study provides a detailed comparison of two leading claims management software in the insurance industry, Guidewire ClaimCenter and Duck Creek Claims in terms of architecture, key features, scalability, deployment options, and industry adoption by evaluating real-world implementations and customer case studies. The study aims to identify key differentiators, business impacts, and technological advantages of both platforms in the insurance sector. By examining their architecture, scalability, functionality and real-world applications, this analysis provides insights to help insurers select the platform that aligns with their operational needs and digital transformation goals.
Study Design: Comparative, multi-case study analysis using qualitative and quantitative data extracted from product brochures, industry reports, and client case studies from official sources.
Place and Duration of Study: Analysis conducted between June 2024 and November 2024, based on data from North America, Europe, and Asia-Pacific insurance markets as documented in vendor case studies, customer testimonials, and industry reports.
Methodology: Reviewed product documentation, industry reports, and implementation case studies from Guidewire and Duck Creek, focusing on client success stories such as AXA Belgium (Guidewire) and Berkshire Hathaway Specialty Insurance (Duck Creek). Evaluation criteria included platform architecture, functionality, scalability, customization capabilities, and market performance. Data from Celent’s vendor evaluation framework provided third-party validation, while qualitative content analysis was performed using client testimonials and success stories. Case studies and industry reports were analyzed for operational, financial, and technological outcomes. By integrating qualitative and quantitative insights, this methodology offers a holistic evaluation of the platforms.
Results: Guidewire ClaimCenter showcased enterprise scalability, deep configuration, and strong adoption by global insurers such as Zurich Insurance, AXA Belgium, Promutuel Insurance and The General Insurance. With AWS-protected cloud-native architecture, large-scale implementations became possible, thereby reducing claim processing time by 60%. Duck Creek Claims, excelled in regulatory compliance and rapid deployment. This has helped insurers like Berkshire Hathaway Specialty Insurance to build a robust and flexible core insurance IT platform for rapid product launches and Liberty Mutual Insurance to develop a next-generation workers’ compensation claims management system achieving full compliance with workers’ compensation regulations. Both platforms improved claim processing, customer engagement, and operational efficiency by using automation and predictive analytics.
Conclusion: Both Guidewire ClaimCenter and Duck Creek Claims are strong claims management solutions, each with unique advantages. Guidewire has strengths in large-scale, data-driven, and highly customized deployments, making it suitable for complex enterprise insurers. Compliance, speed-to-market product launches, and regulatory adherence are all helped by Duck Creek\u27s architecture: cloud-native, low-code. The choice of platform is thus related to the complexity of operations, growth plans, and IT strategies. Further research validation with current data across different markets would be beneficial in arriving at best practices for deployment
Machine Learning Based Finger Print Analysis for Gender Detection: A Review
No full textGender detection using fingerprint biometrics has emerged as a promising area of research due to its non-intrusive nature and potential applications in biometric identification systems. The procedure can involve multiple steps are the size of finger print and their ridge pattern, minutiae point, machine learning and image processing and accuracy and limitations. This review explores the effectiveness of machine learning techniques for gender classification based on fingerprint patterns, emphasizing the role of advanced classification algorithms and feature extraction methods. Machine learning is crucial for gender detection since it classifies fingerprint patterns and biometric information using models like Convolutional Neural Networks (CNN) and Support Vector Machines (SVM). To identify traits unique to a gender, such as ridge density and minutiae points, these algorithms are trained using labelled datasets. Compared to manual procedures, these models are more effective at handling high-dimensional data and identifying subtle gender-related patterns. Although hybrid models like CNN-DNN and AlexNet further increase classification precision, Convolutional Neural Networks (CNNs) are especially effective due to their automatic feature extraction capabilities. Despite their effectiveness, factors like as picture resolution, demographic balance, and dataset heterogeneity might affect performance, highlighting the need for carefully selected datasets and improved model designs. A structured comparative analysis of multiple studies reveals the impact of various datasets, feature types, and model architectures on classification accuracy and reliability. The findings suggest that deep learning models often outperform traditional classifiers, while dimensionality reduction and hybrid approaches can further enhance performance. However, challenges such as dataset imbalances, limited diversity, and susceptibility to low-quality fingerprint data remain prominent barriers to achieving consistent results. This review also outlines key limitations observed across the studies and provides recommendations for future research, including the need for more diverse datasets and optimized classification frameworks. This study aims to improve fingerprint feature extraction for gender detection, reduce processing costs, fix dataset imbalances, and increase classification accuracy. By stating the objective, the scope and objectives of each investigation are made clear. The generalizability of machine learning models is significantly impacted by the amount, variety, and quality of the dataset. The analysis aims to support the development of more accurate, inclusive, and scalable fingerprint-based gender detection systems