EMITTER International Journal of Engineering Technology

An Exploring the Power of Feature Representations: An Empirical Study on Product Reviews for Sentiment Analysis

2025-07-14T14:08:14+00:00

With the rise of e-commerce and online shopping, customer reviews have become a crucial factor in determining the quality and reputation of a product. Online shoppers rely heavily on customer reviews to make informed purchasing decisions, as they don't have the opportunity to physically examine the product before buying. As a result, companies are also investing in sentiment analysis to understand and respond to customer feedback, as well as to enhance the quality of their products and services. Using natural language processing (NLP) and machine learning techniques, sentiment analysis classifies the tone of a customer review as positive, negative, or neutral. It involves analysing text data to determine the overall tone, emotion, and opinion expressed in a review. In this work, we study sentiment analysis of client reviews using machine learning algorithms with different vectorization techniques. The strategy outlined here consists of three distinct phases. The initial step involves some pre-processing to get rid of irrelevant information and find the useful terms. Then, feature extraction was accomplished utilizing numerous vectorization strategies as Bag-Of-Words (BoW), Term Frequency Inverse Document Frequency (TF-IDF), and N-grams. After extracting the features from text data, the final stage is classification and predictions based on machine learning approaches. We evaluated the proposed models on Yelp reviews dataset. The experimental results are evaluated using metrics such as precision, recall, and f1-score, and K-fold cross-validation.

Improving 3D Human Pose Orientation Recognition Through Weight-Voxel Features And 3D CNNs

2025-07-14T14:08:14+00:00

Preprocessing is a widely used process in deep learning applications, and it has been applied in both 2D and 3D computer vision applications. In this research, we propose a preprocessing technique involving weighting to enhance classification performance, incorporated with a 3D CNN architecture. Unlike regular voxel preprocessing, which uses a zero-one (binary) approach, adding weighting incorporates stronger structural information into the voxels. This method is tested with 3D data represented in the form of voxels, followed by weighting preprocessing before entering the core 3D CNN architecture. We evaluate our approach using both public datasets, such as the KITTI dataset, and self-collected 3D human orientation data with four classes. Subsequently, we tested it with five 3D CNN architectures, including VGG16, ResNet50, ResNet50v2, DenseNet121, and VoxNet. Based on experiments conducted with this data, preprocessing with the 3D VGG16 architecture, among the five architectures tested, demonstrates an improvement in accuracy and a reduction in errors in 3D human orientation classification compared to using no preprocessing or other preprocessing methods on the 3D voxel data. The results show that the accuracy and loss in 3D object classification exhibit superior performance compared to specific preprocessing methods, such as binary processing within each voxel.

The Next Generation Wireless Network Deployment Using Machine Learning Based Multi-Objective Genetic Algorithm

2025-07-14T14:08:13+00:00

6G networks provides ubiquitous connectivity, reduced delay and high-speed gigabit connection. The Introduction of AI to the planning process of 5G beyond networks is crucial to ensure the efficient deployment of cells and the minimization of SINR (signal to interference plus noise ratio). The Multi-Objective Genetic Algorithm (MOGA) to take care of the planning issue in 5G and beyond network organizations. This is accomplished by expanding the already existing 4G and 5G infrastructure. The MOGA endeavors to limit the deployment cost, the interference between the cells and maximize the percentage of the clients being served. This work is the solution for deployment problem in next generation networks. The randomly deployment of the cells decreases the network performance, increases the interference and not effective in terms of deployment cost and leads to Dense Multi-Objective Deployment problem. An optimised deployment strategy is employed in the proposed work to address this issue. This work based on optimized utilization of the network through planning. This decreases the cost of deployment, interference and redundancy. It enhances the coverage capacity and quality of service. This excellent coverage of users which is close to 85% is obtained over existing 4G and 5G infrastructure, thereby reducing the total cost of deployment. The work is compared with the meta-heuristic algorithms. The comparison results shows that the proposed work achieves higher SINR, improved coverage capacity than the meta-heuristic algorithms.

Synergistic Integration of LQR Control and PSO Optimization for Advanced Active Suspension Systems Utilizing Electro-Hydraulic Actuators and Electro-Servo Valves

2025-07-14T14:08:13+00:00

This paper investigates the design and optimization of Linear Quadratic Regulator (LQR) controllers for vehicle active suspension systems, incorporating an electro-hydraulic actuator with an electro-servo valve. To enhance both vehicle comfort and road-holding stability, we employ Particle Swarm Optimization (PSO) to optimize the LQR controller parameters. The active suspension system model includes the dynamics of the electro-hydraulic actuator and the electro-servo valve, providing a realistic and practical framework for heavy vehicles. By leveraging PSO, the LQR controller parameters are fine-tuned to minimize a cost function that integrates both comfort and stability up to 76.91%. The results demonstrate substantial improvements in ride comfort and road-holding stability compared to traditional passive suspension systems. This research remarks the fundamentals of the experimental validation and further refinement of these control algorithms to adapt to various driving conditions and vehicle models, ultimately aiming to transition these optimized controllers from theoretical frameworks to practical, real-world applications.

Reliability improvement of distribution networks: A case study of Duhok distribution network

2025-07-14T14:08:12+00:00

Power system is considered one of the most complicated infrastructures. The main components of the system are generation, transmission and distribution. The main function of the system is to supply consumers with electricity as economically and reliably as possible. In order to provide uninterrupted power supply to the consumers, the reliability of distribution system needs to be improved. Several strategies are in place in order to enhance the reliability of the distribution networks. The distribution system could encounter the challenges of aging infrastructure, environmental factors, and the rising in demand power which can cause frequent power interruptions. This paper aims to enhance the reliability of distribution networks by utilizing network reconfiguration techniques to improve voltage profiles, reduce power losses, and restore power to interrupt sections as quickly as possible in the event of a failure. Additionally, the study incorporates the use of fault passage indicator devices installed along the lines. These devices are intended to swiftly identify fault locations, thereby minimizing outage durations and further improving network reliability. An investment in these measures, can obtain significant reliability improvements in the network which at the end lead to consumer satisfaction and huge economic advantages for the system operator.

Factors impacting adoption of electronic HRM in public sector organizations: Case study of Hudury mobile attendance application in Ministry of Education in the Saudi Arabia

2025-07-14T14:10:11+00:00

This study investigates the factors influencing the adoption of the Hudury electronic attendance system among employees of the Ministry of Education (MOE) in Saudi Arabia. Using the Technology Acceptance Model (TAM) and the Theory of Planned Behavior (TPB), this research examines the impact of perceived ease of use (PEOU), perceived usefulness (PU), trust, security, attitude, and behavioral intentions on actual system usage. A non-probability sampling technique was employed to collect 225 responses from employees across three MOE departments through an online survey. Statistical analysis revealed that PEOU, PU, security, and attitude significantly and positively influence the adoption of Hudury. However, while trust and behavioral intention also have a positive impact, their effects on system adoption were found to be statistically insignificant. These findings highlight the importance of addressing trust deficits by conducting training sessions on Hudury’s efficacy to enhance employees' behavioral intentions toward its use. The study is limited by its non-probability sampling method, which may affect the generalizability of the findings to the broader MOE workforce.

Optimization of Gray Level Co-occurrence Matrix (GLCM) Texture Feature Parameters in Determining Rice Seed Quality

2025-07-18T14:16:48+00:00

Rice seed quality assessment is a critical measure in promoting agricultural productivity, as high-quality seeds directly influence crop yield and resilience. One of method for evaluating seed quality is texture analysis, which leverages the Gray Level Co-occurrence Matrix (GLCM) to extract meaningful features from seed images, providing insights into their condition and potential performance. This research aims to determine the optimal performance of GLCM parameters in identifying the texture characteristics of rice seed quality. The experiments were conducted using four angles (0°, 45°, 90°, and 135°) and three-pixel distances (1, 2, and 3), evaluating features such as homogeneity, contrast, dissimilarity, and energy. The results indicate that certain parameter configurations significantly affect the discriminative power of the extracted features, with the Support Vector Machine (SVM) classifier achieving the highest performance at a pixel distance of 1, with an accuracy of 0.73, precision of 0.79, recall of 0.73, and F1-score of 0.72. These findings demonstrate that optimizing GLCM parameter settings directly contributes to improved classification performance, highlighting the method's potential for enhancing rice seed quality assessment.

Utilizing Evolutionary Mating Algorithm Optimized Deep Learning to Assess Cardiovascular Diseases Risk

2025-07-18T14:15:02+00:00

Cardiovascular Diseases (CVD) continue to be a primary cause of death worldwide, underscoring the critical importance of early and accurate risk prediction. However, traditional predictive models struggle with the complexity and interdependencies in medical data. This study addresses this gap by proposing a deep learning-based risk assessment model optimized with the Evolutionary Mating Algorithm (EMA) to enhance prediction accuracy and efficiency. Our contributions include developing a dedicated risk variable for machine learning applications and benchmarking the EMA-optimized model against ADAM and Particle Swarm Optimization (PSO). The proposed method was evaluated using Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Coefficient of Determination (R²), and Standard Deviation (STD). Experimental results demonstrate that the EMA-optimized model outperforms traditional optimization methods, achieving an MAE of 0.037, RMSE of 0.0464, and an R² of approximately 0.91. These results highlight the effectiveness of EMA in enhancing cardiovascular risk assessment models, providing a more reliable tool for early diagnosis and clinical decision-making.

A Detailed Set of Ideas for Designing a Quantum Computing Framework Based on Smart Contracts, Configured Using Foundry and Qiskit

2025-07-18T14:13:16+00:00

The purpose of this paper is to describe a new system design for integrating quantum computing algorithms (and their results) into a blockchain network. In this selected context, we can use, create and upload smart contracts (SCs) that allow users to perform various quantum computations, by using the corresponding circuits. We are therefore proposing a system that uses gas fees in the blockchain context, in order to offer access to certain circuits and their simulation results; the system also allows for the previously analyzed circuits to become publicly available, through SCs – this can act like a quantum circuit encyclopedia. Most users in the first generation will have to pay, in addition to the normal transaction fees (gas) required to call the SC methods, a small development fee for the contract creation for most of the tasks; after a certain number of SCs, enough configurations and results will become accessible to everyone, and only custom, unprocessed circuits will require the development fee. Optionally, a dedicated blockchain network (similar to one of the existing test ones) can also be designed, with contracts that have access to real quantum hardware; its owners can decide (if necessary) the value of the virtual coin in connection to a real-world currency. For our experiments, we selected the Solidity language for the development of SCs, and Python for the development and simulation of quantum circuits, with the help of the Qiskit framework, an open-source library for quantum processing developed by IBM.