Human resource management is a crucial component for the smooth operation of business organizations. However, changing perspectives and values among different age groups present significant challenges for organizations. Resignation and quick job changes can lead to huge budgetary and manpower issues. This research aims to develop and compare machine learning algorithms for predicting employees' turnover decisions using a public dataset on Kaggle.com The dataset includes performance data for 15,000 employees. The eight features and five algorithms, including Random Forest, Support vector machine, Logistic Regression, K-Nearest Neighbor, and Gaussian Naïve Bayes classifier, were employed to develop the model. The results showed that the Random Forest Classifier was the most efficient model with Geometric Mean, F1-score and AUC which are 0.995, 0.995, and 0.999, respectively when evaluated with stratified K-fold cross-validation. Moreover, using the SHapley Additive exPlanations method, five important factors for predicting employee turnover were identified. These factors include satisfaction level, average working hours in a month, number of responsibility projects, time spent with the company and salary level. These features enable the Human Resources department to gain valuable insights, facilitating strategic planning and proactive measures to prevent employee resignations. By doing so, businesses can achieve smoother and more efficient operations.
A user story is commonly applied in requirement elicitation, particularly in agile software development. User story is typically composed in semi-formal natural language, and often follow a predefined template. The user story is used to elicit requirements from the users' perspective, emphasizing who requires the system, what they expect from it, and why it is important. This study aims to acquire a comprehensive understanding of user stories in requirement elicitation. To achieve this aim, this systematic review merged an electronic search of four databases related to computer science. 40 papers were chosen and examined. The majority of selected papers were published through conference channels which comprising 75% of total publications. This study identified 24 problems in user stories related to requirements elicitation, with ambiguity or vagueness being the most frequently occurring problem reported 18 times, followed by incompleteness reported 11 times. Finally, the model approach was the most popular approach reported in the research paper, accounting for 30% of the total approaches reported.
Improper disposal of e-waste poses global environmental and health risks, raising serious concerns. The accurate classification of e-waste images is critical for efficient management and recycling. In this paper, we have presented a comprehensive dataset comprised of eight different classes of images of electronic devices named the E-Waste Vision Dataset. We have also presented EWasteNet, a novel two-stream approach for precise e-waste image classification based on a data-efficient image transformer (DeiT). The first stream of EWasteNet passes through a sobel operator that detects the edges while the second stream is directed through an Atrous Spatial Pyramid Pooling and attention block where multi-scale contextual information is captured. We train both of the streams simultaneously and their features are merged at the decision level. The DeiT is used as the backbone of both streams. Extensive analysis of the e-waste dataset indicates the usefulness of our method, providing 96% accuracy in e-waste classification. The proposed approach demonstrates significant usefulness in addressing the global concern of e-waste management. It facilitates efficient waste management and recycling by accurately classifying e-waste images, reducing health and safety hazards associated with improper disposal.
The Emergency Intercom System (EIS) is a communication and video conferencing system designed to enhance emergency response in areas, such as malfunctioning lifts. This paper presents a feasibility analysis of EIS as internet of things device, focusing on the utilization of WebRTC over WebSocket, Socket.io, and LAN device hostname assignment. The study evaluates performance metrics including latency, throughput, dropped calls, lost packets, audio and video quality, resource utilization, and network bandwidth. The results indicate that EIS offers acceptable latency, sufficient throughput for simultaneous calls, minimal dropped calls and lost packets, and satisfactory audio and video quality. The system demonstrates effective communication and video conferencing capabilities, making it a promising solution for improving emergency response in areas. The findings highlight the potential for further research to optimize performance, address security considerations, and extend the system's applicability in diverse emergency scenarios.
Sarcasm detection is an imperative undertaking within the realm of natural language processing, albeit one that poses considerable challenges when confronted with mash-up languages, characterized by the amalgamation of multiple distinct languages. In response to the intricacies of sarcasm detection in mash-up languages, with a specific focus on the Indonesian-English language mash-up, this study introduces the Hybrid Pretrained Word Embedding approach as a means to enhance sarcasm detection. The primary objective of this research is to augment the precision of sarcasm detection in mash-up languages by amalgamating suitable word embeddings tailored to the employed terms. The present study combines two prevalent pretrained word embeddings, i.e Glove and Fasttext, wherein Glove is utilized to extract semantic context vectors for English words, while Fasttext is employed to extract semantic context vectors for Indonesian words. The classification process in this research leverages the deep learning methodology known as Bidirectional Gated Recurrent Unit (BiGRU). To assess the efficacy of the proposed approach, an extensive dataset comprising sarcastic and non-sarcastic tweets, written in a hybrid language of Indonesian and English, is acquired from the Twitter platform. The results unequivocally demonstrate that the Hybrid Pretrained Word Embedding approach significantly enhances sarcasm detection in mash-up languages, attaining a commendable classification accuracy of 93.57% and an F-measure of 97.94%. By offering an effective methodology to identify sarcasm in mash-up languages, this study yields a substantive contribution to the field of natural language processing.
SQL injections are a significant and ever-present threat to web applications and database security. During these attacks, malicious SQL statements are injected into input fields of data-driven systems, leading to unauthorized access and data breaches. Consequently, a need is generated to understand the nature of the attacks, detection, and effective prevention techniques. This research paper focuses on providing a taxonomy and comprehensive survey of SQL injection attacks, detection, and prevention, including their various types and techniques. Additionally, it explores the current state-of-the-art and evaluation for attacks, detection, and prevention techniques. This research paper also discusses and provides a taxonomy of current machine learning (ML) trends (Taxonomy) and their open challenges for detection purposes. Finally, this paper ends with a discussion aiming to equip system administrators, researchers, scientists and practitioners with the knowledge and strategies to mitigate the risks associated with SQL injection attacks effectively. Eventually, this will help to enhance the security and resilience of web applications and databases in the face of this significant threat.
Rail transport has been one of the main intercity passenger train operators in Malaysia. The Electrification Department is the most important department to make sure the trains have enough power to operate and is responsible for the maintenance of the power supply infrastructure and SCADA system to ensure the Electrification System is accurately monitored and controlled and assure the Reliability, Availability, Maintainability and Safety of the system. The key issue faced by this department is related to reporting on maintenance. The report was very time-consuming to prepare because it was on a paper-based method, and information was presented badly due to the lack of a visualization tool and being vulnerable to data quality problems. To address this issue, the current study aims to propose a tool that provides screening and detecting of device fault status using a dashboard developed using Microsoft Power BI by extracting the raw data log from the Power SCADA server. We demonstrate the process of adopting the cloud-based dashboard using Power BI at the strategic management level. The development of the dashboard system is adopted from the previous research framework, consisting of four main stages which are designing, data analyzing process, visualizing, and validating. One of the findings is that the data cleansing process is the most important stage to produce the right information. This interactive tool made instant visibility of device status and assisted the maintenance team to capture the status trends. The study benefits to the organization are to reduce 86% of the time and cost to prepare the maintenance reporting and planning.
Recognizing Textual Entailment (RTE) is one of the important tasks in Natural Language Processing (NLP). Various approaches have been taken, starting from a simple statistical framework to the neural network (NN) that is currently the mainstay, including RTE in Bahasa Indonesia. Currently, RTE in Bahasa Indonesia has started using the neural network approach, but the value of the resulting accuracy is still less than 77%. This is because the new NN architecture has just accommodated the lexical elements and not yet the syntactical elements of sentences. Syntactical elements in sentences are important components in obtaining the local information contained therein. In RTE, local information is useful for determining how closely related text fragments are. This study proposes a new approach to NN-based Bahasa Indonesia RTE using the Biplet (head-dependency) individual comparison technique. Biplet is generated from the process of word pair dependency. The concept of word pair dependency is used to improve alignment and inference assessment that is optimized by adjusting the weight of the phrase using an attention mechanism. From experiments conducted using the SNLI dataset that has been translated into Bahasa Indonesia (SNLI Indo), it was obtained that the highest training accuracy value is 83.56% with the validation accuracy value is 64.61% for the number of pairs of sentences of 100k.
The Internet has emerged as an indispensable tool in both our personal and professional life in our modern day. As a direct consequence of this, the number of customers who make their purchases over the Internet is quickly increasing. Internet users may be vulnerable to a wide variety of web threats because of this fact. These threats may result in monetary loss, fraudulent use of credit cards, loss of personal data, potential damage to a brand's reputation, and customer mistrust in e-commerce and online banking. Phishing is a sort of cyber threat that may be defined as the practice of imitating a genuine website for the purpose of stealing sensitive information such as usernames, passwords, and credit card numbers. This research focuses on strategies for detecting phishing attacks. This study apply a machine learning approach to detect a phishing attack. As a result, this study able to detect phishing with accuracy 94%.
Knowledge representation and reasoning require knowledge graph embedding as it is crucial in the area. It involves mapping entities and relationships from a knowledge graph into vectors of lower dimensions that are continuous in nature. This encoding enables machine learning algorithms to effectively reason and make predictions on graph-structured data. This review article offers an overview and critical analysis specifically about the methods of knowledge graph embedding which are TransE, TransH, and TransR. The key concepts, methodologies, strengths, and limitations of these methods, along with examining their applications and experiments conducted by existing researchers have been studied. The motivation to conduct this study is to review the well-known and most applied knowledge embedding methods and compare the features of those methods so that a comprehensive resource for researchers and practitioners interested in delving into knowledge graph embedding techniques is delivered.
Microservices-based software architecture promotes scalability and flexibility by breaking down a software application into smaller modules and making it more independent and loosely coupled services compared to monolith systems. However, securing microservices in a distributed nature has become one of the challenges. Authentication is one of the most critical components that should be focused in the microservices security measures. It helps to identify that only authenticated personnel and services can access sensitive information and secure the trust between microservices. This discussion paper aims to provide an overview analysis and extensive understanding on the authentication mechanism in microservices-based software architecture. In this study, we explore different authentication mechanisms including Mutual Transport Layer Security (mTLS), Token based authentication and API Gateway authentication. This study examines the strengths and limitations of different authentication mechanisms in microservices-based software architecture. It also emphasizes the importance of authentication and the need for having a well-designed authentication mechanism to ensure the integrity and security of microservices-based software architecture is crucial.
The Internet of Things (IoT) is expanding exponentially, increasing network traffic flow. This trend causes network security vulnerabilities and draws the attention of cybercriminals. Consequently, an intrusion detection system is designed to identify various network attacks and provide network resource protection. On the other hand, building a steadfast intrusion detection system is difficult since there are numerous flaws to address, such as the presence of supernumerary and irrelevant features in the dataset, leading to low detection accuracy and a high false alarm rate. To address these flaws, researchers are attempting to research on applying supervised machine learning techniques in intrusion detection systems for IoT. Therefore, this paper explores the prevailing machine learning techniques utilized in the intrusion detection system research area to provide better insight in this field.
This study explores the relationship between respondent demographic profiling and the factors affecting teachers' readiness in adopting augmented reality (AR) for teaching and learning. While AR holds promise for transforming educational practices, the readiness of teachers to adopt this technology remains a critical factor for successful implementation. However, there is a lack of research examining the specific relationship between respondent demographics and the factors influencing teachers' readiness in AR adoption. This study aims to fill this gap by employing a quantitative research approach and survey methodology. By analyzing the data collected from primary school teachers, including demographic information and factors affecting readiness, the study seeks to identify any significant correlations or patterns. The findings will contribute to a better understanding of how respondent demographics influence teachers' readiness in AR integration, addressing a crucial problem in the field of educational technology. The insights gained from this study will inform policymakers and educational institutions in developing targeted strategies to enhance teachers' readiness and support the successful integration of AR in primary school classrooms.
Ammunition plays a crucial role in military and defense operations, requiring significant investments to arm military forces adequately. However, ammunition is susceptible to environmental factors that can degrade its quality, leading to defects or even accidental explosions. To ensure constant combat readiness, it is vital to maintain secure storage facilities with sufficient ammunition supplies. This project aims to enhance ammunition inventory and safety management procedures by leveraging loT technologies. This project proposed the implementation of an loT-powered web application dashboard that utilizes weight measurements to provide real-time inventory tracking and monitors environmental conditions such as temperature and humidity for quality control. Additionally, the system can predict ammunition condition outcomes. By adopting this loT-based solution, ammunition management processes will be streamlined, resulting in improved efficiency and effectiveness.
Ever more IoT devices and services find their way into private homes and industry, coming along with a plethora of risks to users' privacy. The General Data Protection Regulation (GDPR) became effective in 2018 and protects rights of IoT (and other) users in the European Union (EU). Manufacturers can address these rights, for example, with firmware updates. In this paper, we conduct a large-scale analysis that identifies changes in the age of the installed firmware and general device age after the GDPR went into effect. We utilize a set of 400 terabytes of real-world IoT data from Censys.io dating from 2015 until the end of 2021. Based on grouped mean age values, we conduct difference-in-differences analyses for devices deployed in the EU, compared to Malaysia (MY), Indonesia (ID), Singapore (SG) and USA. The results show unexpected insights. For a majority of EU member states, the GDPR leads to an increase of the devices' mean age by 101 days compared to the other countries in our data set. Compared to ID it increases by 201 days, SG by 11 days, MY+ID+SG by 89 days and USA by 194 days. Results for MY are not significant, however. This work offers first insights into effects of the GDPR in the IoT ecosystem and highlights the need for more research for sense-making.
Image forgery is the alteration of a digital image to hide some of the important and useful information. Copy-move forgery (CMF) is one of the most difficult to detect because the copied part of the image has the same characteristics as the original image. Most of the existing datasets only highlight additional attacks in the copied part. Since there are no categories of duplication elements in the datasets, this research analyzed three categories of duplication elements in CMF which are animals, food and non-living things using DEFACTO and CoMo3Dataset. The analysis is performed on PatchMatch-based detection method and the results show that the method able to maintain at least 83% for all duplication elements in both DEFACTO and CoMo3Dataset. Furthermore, the method is able to detect a minimum 92% score for the food category in both datasets.
Binocular vision is a type of vision that allows an individual to perceive depth and distance using both eyes to create a single image of their environment. However, there is an illness called strabismus, where it is difficult for some people to focus on seeing things clearly at a time. There are many diagnoses that need to be done for doctors to diagnose whether patients suffer from strabismus or not. Besides, a new practitioner could lead to misdiagnosis due to lack of professional experience and knowledge. To overcome these limitations, a machine learning algorithm, which is a case-based reasoning, is developed to automate the strabismus classification. The results showed that the case-based reasoning algorithm provides 91.8% accuracy, 89.29% precision, 92.59% recall and 90.91% F1-Score. This shows that using the case-based reasoning algorithm can give better performance in classifying the class.
Low-code application development platforms (LCDPs) have been widely used recently to replace typical application development with lower development costs and speed up application delivery. However, LCDP in the market is still evolving, with several limitations. This research aims to demonstrate a new LCDP implementation that can generate an application with improved functionality practically used in real-world business applications. The invention starts from the concept of model-driven engineering (MDE) theory and designs an intelligent code generator engine. This LCDP implementation is a software-as-a-service platform allowing users to design their custom applications with the visualized designer. The output of the design then is to generate an executable source code that enables to build of the software applications with minimum limitation, less code for development, future extensibility, and unleash from vendor locked-in of a proprietary platform. The research has also extended the study by comparing functionality with benchmark commercial applications so that the generated software applications from this LCDP research can perform similarly.
Wireless sensor networks (WSNs) play a crucial role in environmental monitoring and data collection. However, ensuring data security in WSNs poses challenges due to the vulnerabilities of wireless communication channels. In this paper, we address this concern by exploring the application of cryptographic techniques to enhance data security in WSNs. Considering the limited sensor power, computing power, and storage resources, we propose a novel approach that evaluates the suitability of symmetric and asymmetric cryptographic algorithms in WSNs. Through performance comparisons based on computation power and storage capacity requirements, we identify key insights for selecting appropriate encryption algorithms in WSNs. Our findings emphasize the importance of considering the specific requirements and constraints of WSN applications, highlighting the efficiency of symmetric key-based encryption algorithms in resource-constrained environments and the stronger security and key distribution mechanisms provided by ECC-based asymmetric encryption algorithms for secure communication among multiple nodes. This research contributes to the existing knowledge by offering an effective solution to enhance data security in WSNs while considering computational and storage limitations
With the increasing demand for secure, trustworthy, and transparent voting systems, electronic voting (e-voting) has emerged as a promising solution to address the shortcomings of traditional methods. However, traditional e-voting systems face numerous challenges in ensuring integrity, privacy, and auditability such as paper-based ballots and electronic voting machines. These traditional systems are susceptible to issues such as voter fraud, coercion, and a lack of transparency, undermining the democratic process. This paper proposes a novel approach to designing an e-voting framework using blockchain technology. By leveraging the distributed ledger and consensus mechanisms of blockchain, our framework aims to provide a secure, transparent, and tamper-resistant voting system. These features enable the creation of a secure and transparent voting system that ensures the integrity of votes, protects voter privacy, and enables verifiability and auditability of the entire voting process. The paper presents the key components, design considerations, and benefits of the proposed framework, along with an analysis of its potential challenges and future directions for research and development. The system architecture of the proposed framework establishes communication channels, data flows, and interfaces that facilitate voters' attendance, secure interactions and information exchange within the voting system. The use of smart contracts helps enforce the rules and conditions of the voting process on the blockchain, ensuring the accuracy and fairness of the electoral outcome. In conclusion, the proposed e-voting framework using blockchain technology has the potential to revolutionize the electoral process by providing a secure, transparent, and tamper-resistant voting system. By addressing the challenges of traditional e-voting systems and leveraging the inherent features of blockchain technology, we can enhance the integrity, privacy, and trustworthiness of the voting process.
