Automatically generating accurate and comprehensive class diagrams from natural language requirements can minimize human errors, improve accuracy, and streamline requirements analysis. OpenAI’s GPT-4 model has made significant strides in this domain. For GPT-4 to gain traction within requirements engineering, the quality of its class diagrams is essential. This study evaluates GPT-4’s class diagrams by comparing them to those created by experts and existing tools, using precision, recall, and F1 measures, which reveal significant variability. GPT-4’s precision ranges from 0.61 to 0.88, reflecting a varied ability to correctly identify instances. Recall spans from 0.63 to 1.00, indicating differences in capturing all relevant instances. The F1 score, which balances precision and recall, ranges from 0.65 to 0.87, indicating a variety of effectiveness in different contexts. In particular, GPT-4 outperforms existing tools in precision, recall, and F1 score, showcasing its strong aptitude to generate accurate class diagrams from natural language. This paper evaluates the GPT-4 diagrams against expert benchmarks, compares them with four tools, and presents insights into GPT-4’s capacity in requirements engineering.

The main result of this paper is building a new parallel algorithm based on Floyd-Warshall algorithm to find the Shortest Path for all-pair. The problem of finding All Pair Shortest Path (APSP). APSP is performed on many structures such as: MPI, OpenMP, Cuda and others. When the input data is large, we need to find ways to improve computing power to reduce calculation time for the APSP algorithm. We present a contribution for the APSP in the MapReduce architectures. The idea of this algorithm is to multi Workers to work in parallel by Floyd-Warshall algorithm. There is one Master assigning Map and Reduce. The mapper simultaneously executes their work and sends their data to the reducer until the job is finished. The MapReduce algorithm has two functions: Map and Reduce. They receive key/value pairs based on an adjacency list. The algorithm performs in MapReduce and our results prove that the proposed approach improved Computing Performance for Algorithm Finding the Shortest Path for all-pair. Some fundamental results are systematized and proved.

Association Rule Mining (ARM) and demand forecasting are vital in business intelligence, especially in commerce. ARM algorithms can be computationally expensive and require data migration to environments like R or Python. Meanwhile, demand forecasting often needs separate models for each product and is sensitive to data quality. To address ARM challenges, we implemented Apriori and Eclat algorithms, well-known itemsets generation algorithms, along with a rules generation algorithm in PostgreSQL. Additionally, we developed multivariate Long Short-Term Memory (LSTM) models with grouped data based on itemsets generation. Evaluation on the Online Retail II dataset (~one million rows) showed that our SQL ARM implementation achieved low overhead time (3.3s) and acceptable processing times (Eclat: 3s, rules generation: 0.1s). Our method is comparable to the state-of-the-art R implementations (overhead: 4.4s, Eclat: 0.66s, rules generation: 0.4s). For demand forecasting, the multivariate LSTM with grouped data reduced training time from 280.1s to 122.1s, improved Mean Squared Error from 1.045 to 0.882, and Mean Absolute Error from 0.471 to 0.433.

Web applications have become immensely popular and widely utilized across various sectors due to their accessibility, scalability, and ability to provide real-time updates, driving the demand for skilled web programmers. Approximately 75.9% of websites that use server-side languages rely on PHP, a widely adopted and flexible scripting language that integrates seamlessly with databases, making it suitable for dynamic website development. As the complexity of web development increases, self-learning systems become essential for fostering independent programming skills through platforms such as learning management systems (LMS) and MOOCs; however, many of these tools still require human intervention for code verification and lack automated feedback mechanisms. This study proposes a self-learning framework that leverages test-driven development (TDD) to enhance PHP programming education. The framework enables students to participate in structured learning modules, receive realtime feedback, and refine their coding abilities. An evaluation with 150 first-year IT students in Indonesia demonstrated a 100% success rate in all modules, where success was defined as the ability of students to pass all test cases in the predefined test suites after multiple attempts. This highlights the framework’s effectiveness while also noting areas for improvement in content depth and instructional clarity.

Throughout the past decade, wireless sensor networks (WSNs) have become a focus of observation in the wireless and mobile computing research community. The WSN has many uses, and applications from inside (starting from measuring temperature, pressure, humidity, and similar application inside the home) to outside (extension on the tactical battleground). Due to the distributed nature and spread in distant region, these networks are prone to many security attacks, which in turn negatively influences performances of these networks. Therefore, securing wireless sensor networks against these threats is becoming more and more important. A malicious node within the network is one of these threats. In this study, a procedure for detecting a malicious node in the network using the NS-2 simulator is presented. The proposed method uses Artificial Neural Network (ANN), built using MATLAB, for the purpose of prediction. Training, validation and testing have been performed on a dataset containing throughput, power consumption and time delay as inputs of the used ANN. Experimental results show the efficiency of the proposed method in security threats detection.

The article aims to explore the concept of "accessibility" and provide a general context for the creation of an inclusive society in modern conditions, making a special accent on IT industry. As part of the work, an analysis of statistical data from public sources on people with disabilities, as well as legal norms and laws regarding their support and integration into society was carried out among Ukraine, the European Union, and the United States of America. It was found that people with disabilities make up a significant part of modern society, and the legal basis of the developed countries is systematically updated and expanded in the context of supporting people with special needs, placing formal requirements on suppliers of goods and services. On the other hand, it was also found that people with disabilities form a significant group of consumers, so the support of inclusiveness trends has a positive effect on the income of modern companies, from which a conclusion was made about the relevance of an accessibility processes in the conditions of modern business and the IT sphere in particular. During the study, an analysis of current trends in the software development industry in the context of accessibility was carried out, as well as an analysis of web accessibility standards, mainly the "Web Content Accessibility Guidelines (WCAG)" due to its recognition and prevalence, flexibility and adaptability to new technologies, systematic maintenance, and support, as well as integration into the legal framework of the developed countries. Accessibility of the web application was studied in the context of the development lifecycle, at the next stages: requirements analysis, design creation, writing of software code and testing. A model of continuous accessibility process for projects with flexible iterative processes was proposed, which involves a use of technical means, such as static code validators, browser extensions or comprehensive solutions for automated accessibility testing. It was concluded that the comprehensive and systematic use of both organizational and technical accessibility practices at all stages of iterative development is a key factor in the success of the implementation of this initiative in modern software development processes.

Although the Global Positioning System (GPS) is a cornerstone of modern navigation, its accuracy can diminish in urban areas, indoors, and intentionally jammed locations. This poses significant challenges for Unmanned Aerial Vehicles (UAVs) operating autonomously in these "GPS-denied" environments. In this context, multi-sensor data fusion (MSDF) is suggested as a viable technique, as it integrates inputs from various sensors to create a more robust and reliable navigation solution. In this paper, a system has been developed that enables an AUAV flying along a predetermined route to reliably detect both fixed and moving obstacles in challenging environments where GPS signals are weak or absent, and to perform effective avoidance maneuvers to prevent potential collisions, offering superior situational awareness and operational efficiency. The results obtained demonstrate that the AUAV can navigate safely and accurately in complex and continuously changing environments. The findings reveal that the proposed system has the ability to reliably detect both stationary and moving obstacles in challenging environments where GPS signals are absent or weak, and to perform effective avoidance maneuvers to prevent potential collisions in real time.

The article examines and analyzes the numerous advantages of using information technologies to ensure the information security of organizations in connection with the wide spread of the number of modern methods of cyber attacks. It is established that effective cyber protection requires an information security management system with a set of modern event monitoring tools depending on the specifics of each organization. To select an appropriate system and evaluate the effectiveness of its tools, the method of expert evaluation is used in the work. In order to improve the determination of the weight coefficient of each tool of the system, a composite indicator is proposed, based on the sum of the products of individual indicators of the system tools and their priority coefficients. The features of the modern widely used solutions considered in the study confirmed the feasibility of the proposed methodology for determining effective tools of the information security monitoring system. The resulting data allows us to help organizations make an evidence-based decision about the optimal composition of the information security monitoring system.

Facial expressions serve as a means of conveying human emotions and individual intentions. The ability to perceive and interpret facial emotions is a relatively effortless job for humans, although it poses significant challenges when attempting to replicate this capability using a computer. Facial expressions can be detected from static photos, video, webcam data, or real-time photographs. The primary focus of this study is to create the SMM Facial Expression dataset and to develop a Convolutional Neural Network (CNN) model for accurately recognizing and classifying facial expressions. This model utilizes End-to-end (E2E) networks to analyze individual frames or clusters of frames received from the camera. The analysis is conducted through various layers of convolutional and pooling operations. The proposed model is evaluated on two benchmark datasets, add long form Cohn-Kanade (CK+) and Facial Expression Recognition 2013 Dataset (FER2013) for facial expression recognition (FER). The obtained accuracy rates are 85.27% and 82.18% for CK+ and FER2013, respectively. This study demonstrates that the SMM Facial dataset is comparable in quality to previously benchmarked datasets, and the proposed model holds potential for real-time facial expression recognition.

The architecture of the Industrial Internet of Things (IIoT) system assumes the presence of a router as the main subsystem, which will allow connecting the subsystems into a single network and connecting them to the equipment of the Internet service provider. Failure of the router can lead to downtime of industrial equipment and the whole workshop at the industrial enterprise, and to the failure of the IIoT system as a whole. Control processing units of industrial routers (RCPU) are complex microprocessor systems (MS). Violation of their operability can be caused by both hardware (HW) and software (SW) failures. Therefore, to increase the reliability of RCPU it is reasonable to use such methods of reliability assurance as HW redundancy (duplication of processors with comparison of results), as well as the use of modern means of control and diagnostics. However, the number of HW and SW failures in RCPUs, is still quite high. In this connection, the solution of issues related to the development of methods for ensuring high reliability of RCPU is of great urgency. This paper proposes a method of ensuring the reliability of RCPU using SW diversity on the basis of priority series. The method let to choose the variant of redundancy of HW and the number of versions of SW as diversity in RCPU.