AlKadhim Journal for Computer Science

The Connected Classroom: Leveraging EdTech to Enhance Student Engagement

2025-11-02T22:40:55+03:00

This mixed-methods study examined university students' perceptions of the role of technology in promoting engagement, collaboration, and autonomy. Data were collected from 120 students (65 females and 55 miles) from three universities (Al-Mustansiriya University, University of Diyala, and University of Wasit) through participation in questionnaires, classroom observations, and open-ended feedback.

Results showed that students had highly positive viewpoints, with engagement (M=4.12) being the most strongly observed advantage, followed by collaboration (M=3.98) and autonomy (M=3.85). Inferential analysis revealed significant differences in perceptions of collaboration: female students reported higher scores than male students (t (118) =2.04, p=.04), and students from Al-Mustansiriyah University reported significantly higher collaboration than those from another (F (2,117) =3.67, p=.028). No significant differences were discovered for engagement or autonomy. Observational data revealed high levels of peer collaboration, while qualitative feedback highlighted key opportunities (e.g., learning flexibility) and challenges (e.g., communication issues, digital literacy gaps).

The results indicate that while technology is an effective tool for student engagement, its ability to enhance collaboration depends on demographic and institutional factors. The study underscores the importance of thoughtful pedagogical design and strong institutional support to leverage the potential of collaborative technology efficiently and equitably.

Analyzing Big Data to Mitigate Cyber Attacks Using Machine Learning Classifications: A Comparative Study of an Efficient Classifier Set

2025-08-17T20:05:17+03:00

In this research, we address a set of methodological approaches and approaches in scientific research in the field of cybersecurity and big data analysis using machine learning techniques, which have made the data analysis process easier to understand and mitigate cyberattacks. This method involves data processing and analysis in a series of stages, or in the form of analytical protection layers, including analysis and organization of large databases,. Training and qualifying classifiers is a method for reducing the dimensionality and skewness of feature vectors according to basic procedure analysis.. Using the principal components in the analysis, we employ various binary classifiers for vector K-nearest neighbors and many other algorithms, such as Bayesian algorithms and others, analyze vector machines, as well as artificial intelligence, which work to increase the efficiency and accuracy of attack detection devices. Artificial neural networks are also highly efficient in detecting cyberattacks and analyzing all network disturbances. Through our research, the basic idea was to classify and segment data and deal with each type of data separately to facilitate processing. Another approach is to combine data classifiers. This idea relies on two options: soft voting and majority voting. Each approach has its own method and method for specific use. methods for intrusion detection and penetration of large databases. In the first approach, we analyze and process data in parallel by classifying and dividing it into several subsets, and assigning each subset a separate path. In this method, we partition the problem, making the solution simpler and the data less complex. In the other approach, we also use sensors to collect client data within search servers. The sensor contains a set of parallel paths, each path analyzes the data and information for the client, and is conducted through a parallel network that detects anomalies through two different data sets. One group contains computer network traffic, which includes examining data and hosts from distributed denial of service (DDOS) attacks. The other group contains the Internet of Things and its data.

Explainable Artificial Intelligence Integrated Ensemble Learning Framework for Diabetes Prediction

2025-12-06T00:38:04+03:00

Accurate prediction of diabetes based on clinical and demographic indicators is essential, as early prediction of this chronic metabolic disorder plays a critical role in preventing long-term organ complications. However, existing research continues to face significant challenges, including pronounced class imbalance, the scarcity of large and diverse datasets, and limited integration of explainable artificial intelligence. This research compares several ensemble learning methods (Decision Tree, Random Forest, AdaBoost, Gradient Boosting, Histogram based Gradient Boosting, extremely randomized Trees, and XGBoost) on a large imbalanced dataset (87,664 negative vs. 8,482 positive samples). To mitigate imbalance, we evaluate six resampling approaches including (Random Over Sampling, Random Under Sampling, Synthetic Minority Over-sampling, Adaptive Synthetic Sampling, Tomek Links Removal Sampling and (Synthetic Minority Over-sampling with Edited Nearest Neighbors). We assess models using metrics robust to class imbalance (precision, recall, F1, AUC-ROC, and AUC-PR) and calibration measures. The Extra Trees classifier achieved the highest measured accuracy (0.994); with Random Over Sampling for balancing dataset. also, these results were compared with several previous works and number of machine learning algorithms, and the results showed superiority. Explainability is performed at both global and local levels: permutation and SHAP for global feature importance, and (Local Interpretable Model-Agnostic Explanations) force plots for instance-level reasoning. however, we analyzed this result using sensitivity, specificity, PR-AUC and calibration, we report detailed experiments showing how resampling method, hyperparameter tuning, and stratified validation influence performance. Finally, we provide clinical-relevant insights from SHAP analyses and discuss limitations and future directions for deploying interpretable models in screening workflows.

Practical Implementation and Analysis of Software Metrics Impact on Maintainability in Open-Source Systems

2025-12-06T00:49:56+03:00

Common software metrics and maintainability measures in open-source Java are examined in this research.The authors test the major object-oriented metrics—Coupling Between Objects (CBO), Lines of Code (LOC), Weighted Methods per Class (WMC), Lack of Cohesion of Methods (LCOM), Depth of Inheritance Tree (DIT), and Cyclomatic Complexity—against real-world maintainability indicators like bug counts, code modifications, and developer turnover. The authors use Python data analysis and visualization to find statistically significant patterns in Spearman correlation analysis.

The findings indicate that metrics such as CBO and cyclomatic complexity are very predictive in terms of maintenance effort, whereas others, such as DIT, provide little insight. In addition to theoretical justification, this work provides a practical, reproducible workflow to be implemented by software engineers to give priority to code quality and make more intelligent maintenance decisions. Finally, this study ties the strengths of academic measures and the real-world aspects of daily software engineering.

Optimized Hybrid CNN-LSTM Framework with Multi-Feature Analysis and SMOTE for Intrusion Detection in SDN

2025-11-28T19:59:08+03:00

There is a growing trend toward the use of software-defined networks (SDN), which presents new security challenges requiring advanced intrusion detection systems (IDS). This paper proposes a deep learning-based hybrid system combining convolutional neural networks (CNNs) and long-term short-term memory networks (LSTMs) that can be used for effective intrusion detection in SDN environments. The model uses CNNs to acquire the spatial features of network traffic data and LSTMs to learn temporal patterns, enabling the identification of complex attack patterns. We evaluate our model using an InSDN dataset and test its performance using various feature sets, ranging from 6 to 83 features in our model. Experimental results indicate that our model has a high multi-class classification accuracy of 99.63% when using all 83 features in Group1. Furthermore, we use a Synthetic Minority Over-sampling Technique (SMOTE) to address the issue of class imbalance which considerably enhances detection accuracy of minority attack classes which is reaching 99.76%. It is established that the presented hybrid CNN-LSTM model is powerful and successful solution to SDN security improvement.

Deep Learning-Based Blood Cell Classification with Enhanced Data Preprocessing and Augmentation

2025-10-12T13:06:45+03:00

Classification of blood cells accurately is an extremely important task in the field of hematology, for the diagnosis of blood disorders and for guiding decision making in the clinical context. In this paper, we use a high-quality dataset of 17,092 microscopic peripheral blood cell images from the Hospital Clinic of Barcelona, encompassing eight different cell types, all of which have been annotated by expert pathologists. To improve model performance and to tackle the class imbalance in the dataset, we developed a strong data preprocessing and data augmentation pipeline which includes contrast enhancement, normalization, geometric and photometric transformations, injection of noise, and mixup style synthetic data. We develop two state-of-the-art deep learning models (EfficientNet-B0, ResNet50) to enable benchmarking of the proposed pipeline. In our experimental results, EfficientNet-B0 achieved overall accuracy of approximately 98.3% and ResNet50 achieved accuracy of 98.6%, with very good precision, recall, and F1-scores for all classes. These preliminary results demonstrate the effectiveness of the designed data preprocessing and data augmentation strategies, as well as provide a benchmark for managing blood cell images in hematology for future research.

Federated Learning for Early Detection of Advanced Persistent Threats in IoT Networks

2025-12-07T16:20:25+03:00

In the era of connected IoT devices, ensuring cybersecurity while preserving data privacy is increasingly critical. Federated learning offers a promising approach by enabling collaborative training of detection systems without sharing raw data. This paper presents a novel federated Intrusion Detection System (IDS) based on XGBoost algorithm, and for the first time designed to detect initial compromise (I.C.) phase of Advanced Persistent Threats (APTs) in distributed Internet of Things (IoT) environments. By leveraging the federated framework, the IDS achieves robust detection across multiple devices while maintaining privacy and minimizing computational overhead. Extensive simulation results indicated that our proposed method achieved a precision of 97%, recall of 100%, and F1-score of 98%, providing a practical and efficient solution for real-world IoT security challenges.

Computationally Efficient Hybrid Framework for MNIST Handwritten Digit Recognition

2025-11-26T13:18:18+03:00

This research work suggests a combination classification technique for handwritten digit recognition based on MNIST dataset. The design embraces Principal Component Analysis (PCA) to reduce the level of dimensionality, KMeans clustering to get the pattern of the framework, and Random Forest to make the final prediction. By compressing the feature space into a 50 dimensional space and adding in cluster information, we are able to attain an accuracy of about 93.1 percent. The significant contributions of this paper are to prove the PCA-KMeans integration useful to improve classification, offering a computationally efficient solution as alternative to deep learning model, and to prove that feature augmentation using clustering outcomes leads to better discrimination of visually similar digits. The data suggest that the referred to hybrid model is stable and feasible approach to image identification and pattern recognition, which could be applied to more complicated dataset.