Blockchain technology presents a paradigm shift in fraud detection and prevention within financial institutions. This paper explores the integration of blockchain solutions to address prevalent challenges in identity management, transaction verification, and data security. The proposed system leverages blockchain's decentralized, immutable ledger, coupled with smart contract functionality, to automate verification processes and reduce administrative overhead. Drawing inspiration from the Super Cert model, we discuss the implementation of Ethereum and IPFS for securing educational certificates and adapting these principles to financial systems. We demonstrate the integration of smart contracts to create a decentralized, tamper-proof system that enhances transparency, reduces costs, and improves security. Future implementations include scalability solutions, cross-chain interoperability, and enhanced AI integration.

Concrete paver blocks have emerged as a versatile and sustainable alternative to traditional paving materials, offering numerous functional, environmental, and aesthetic advantages. These precast concrete elements are extensively used in a wide range of applications, including residential driveways, public sidewalks, commercial spaces, and heavy-duty industrial pavements. Their modular design, ease of installation, and ability to be customized in various shapes, colors, and patterns make them highly suitable for modern construction needs. The primary aim of this study is to evaluate the feasibility of incorporating Waste Foundry Sand (WFS) into the production of concrete paver blocks. Foundry sand, a byproduct of the metal casting industry, contains high silica content and favorable physical properties, making it a potential partial substitute for natural sand in concrete. This study investigates the effects of WFS on the mechanical and durability characteristics of M40 grade concrete. By replacing natural sand with varying proportions of WFS, the research aims to determine the optimal replacement level that maintains or improves the performance of concrete paver blocks.

The rapid expansion of dynamic web applications has introduced significant security vulnerabilities, particularly Cross-Site Scripting (XSS), which remains among the top threats to modern websites. XSS attacks allow malicious actors to inject client-side scripts into trusted websites, leading to session hijacking, defacement, data theft, and phishing. This research focuses on identifying and mitigating XSS vulnerabilities in web applications, using the LMS (Learning Management System) platform as a case study. Initial evaluations using commercial scanners revealed only a subset of vulnerabilities, prompting the implementation of a thorough manual testing approach. This led to the discovery of critical flaws associated with sanitized input variables such as $_GET, $_POST, $_REQUEST, and direct output via echo. These vectors were found to be exploitable through stored, reflected, and DOM-based XSS attacks. To address the limitations of existing one-way filters and scanners, this thesis proposes a novel two-way filtering and detection mechanism, termed XSS_OBLITERATOR. Unlike conventional tools, XSS_OBLITERATOR is capable of detecting and neutralizing malicious code on both the client and server sides. It offers language-independent protection and is designed to be scalable and platform-agnostic, ensuring robustness across diverse web environments. The research outlines the design and implementation of this framework and evaluates its effectiveness through pilot experiments using various XSS vectors on the LMS platform. The results demonstrate that the proposed solution significantly improves the detection rate and minimizes false positives compared to existing methods. Furthermore, it addresses challenges in securing legacy codebases already compromised by persistent XSS payloads.

Bayesian networks are powerful tools for decision-making and reasoning under uncertainty. A simplified form of these networks, known as Naïve Bayes, is particularly efficient for inference tasks due to its computational simplicity. However, Naïve Bayes relies on a strong assumption of feature independence, which may not always hold in real-world scenarios. This dissertation presents an experimental study on the application of the Naïve Bayes algorithm for intrusion detection, incorporating Rough Set Theory to enhance performance. Despite its simple structure, Naïve Bayes demonstrates competitive results in terms of classification accuracy and F-measure. Furthermore, the study compares the performance of Bayesian networks with the Classification and Regression Tree (CART) model using the Kyoto dataset, highlighting the superior or comparable effectiveness of Bayesian approaches. The dissertation also introduces the fundamental concepts of Intrusion Detection Systems (IDS) and Rough Set Theory. An IDS is a crucial security mechanism designed to monitor network activities and alert administrators to potential malicious behavior. Given the increasing frequency and sophistication of intrusion attempts aimed at compromising organizational data, network security has become a critical area of research. Consequently, enhancing intrusion detection capabilities remains a significant focus in the field of cyber-security.

A large volume of detailed personal data—such as shopping habits, criminal records, medical histories, and credit information—is routinely collected and shared for various data mining applications. On one hand, this data represents a valuable asset for businesses and government agencies, enabling informed decision-making through comprehensive analysis. On the other hand, privacy regulations and growing concerns over individual data protection can hinder data sharing and limit its use in analytics. To address the challenge of balancing data utility with privacy preservation, this work proposes a vector quantization-based approach for privacy-preserving data clustering. The method involves segmenting each row of the dataset into multiple parts (piecewise segmentation), followed by applying K-Means quantization to each segment individually. The quantized segments are then recombined to form a transformed dataset suitable for clustering while maintaining privacy. Experimental results are presented to evaluate the performance of the proposed method. These experiments aim to identify the optimal segment size and quantization parameter that strike the best balance between clustering accuracy and data privacy. The results demonstrate that appropriate tuning of these parameters can significantly improve the trade-off, enabling effective data analysis without compromising sensitive information.

Arterial blood gas (ABG) analysis is an essential diagnostic tool used in healthcare to evaluate a patient’s respiratory and metabolic condition. This article provides a comprehensive overview of ABG machine functionality, its significance in clinical decision-making, and best practices for obtaining accurate results. It explores two types of ABG analyser’s, highlighting their accuracy and technological advancements that improve efficiency and reliability. The role of nurses in conducting ABG tests is also addressed, including overview of reinforcement training sessions conducted to enhance their competency. Additionally, the article emphasizes the importance of correct sample handling, interpretation of critical or abnormal values, and timely reporting to ensure optimal patient outcomes. Understanding the correct usage and calibration of the ABG machine enables healthcare professionals to manage critically ill patients more effectively, reinforcing the importance of this technology in emergency and intensive care environments.