The rapid generation of data from various sources by the public sector, private corporations, business associations, and local communities is referred to as big data. This large and complex dataset is often regarded as the ‘new oil’ by public administrations (PAs), and data-driven approaches are employed to transform it into valuable insights that can improve governance, transparency, digital services, and public engagement. The government’s big-data ecosystem (GBDE) is a result of this initiative. Effective data management is the first step towards large-scale data analysis, which yields insights that benefit your work and your customers. However, managing big data throughout its life cycle is a daunting challenge for public agencies. Despite its widespread use, big data management is still a significant obstacle. To address this issue, this study proposes a hybrid approach to secure the data management life cycle for GBDE. Specifically, we use a combination of the ECC algorithm with AES 128 BITS encryption to ensure that the data remain confidential and secure. We identified and analyzed various data life cycle models through a systematic literature review to create a data management life cycle for data-driven governments. This approach enhances the security and privacy of data management and addresses the challenges faced by public agencies.

This study addresses the challenge of accurate object detection in foggy environments, a critical issue in computer vision. We propose a novel approach using a real dataset collected from diverse foggy weather conditions, focusing on varying fog densities. By annotating the dataset from Real-Time Traffic Surveillance (RTTS) and using the YOLOv8x architecture, we systematically analyze the impact of fog density on detection performance. Our experiments demonstrate that the YOLOv8x model achieves a mean average precision (mAP) of 78.6% across varying fog densities, outperforming state-of-the-art methods by 4.2% on the augmented dataset. Additionally, we show that increased dataset diversity significantly enhances the robustness of the model in detecting objects under challenging foggy conditions. Our research contributes to advancing object detection systems tailored for foggy environments, with implications for safety and efficiency in domains like autonomous driving and surveillance.

K-Nearest Neighbor search is used extensively in fields like computer vision, DNA specification, object recognition and many more. Online map applications are also used tremendously due to the advances in Geographic Information System (GIS) data. These geospatial objects can be retrieved by using spatial range query. In our proposed work we find the nearest neighbors across any query point q, using an over estimated value of k. The Range Query calculated area is used to estimate the number of k neighbors. If the initial value of k doesn’t reveal all the nearest neighbors inside R, the value of k is multiplied with a constant factor epsilon. In this way all the nearest neighbors inside R are retrieved with in 1 or 2 iterations. The computational complexity of the proposed algorithm turns out to be O(n), compared to the complexity of Density Based Range Query algorithm i.e., O log(n2 + n). This makes our proposed algorithm a more optimized solution.

In the digital education system, the increasing demand for measuring lexical and semantic text similarity has led to the need for developing scalable and intelligent automated grading systems. Traditional assessment methods used for short-answer grading are labor-intensive, time-consuming and prone to human-evaluator bias. Furthermore, computer science curricula have different evaluation challenges as programming-centric courses require rigorous syntactical and structural validation, whereas theoretical courses require deep conceptual and semantic comprehension. To address these dual challenges, we introduce the Virtual University Automatic Short-Answer Grading (VUASAG) framework, a model addressing both statistical-based lexical analysis and Transformer-based semantic architectures. Utilizing an expanded Mohler dataset spanning multiple computer science domains (including Data Structures, Introduction to Programming, Object-Oriented Programming, and Software Engineering), we evaluated the efficiency of traditional string-matching metrics alongside state-of-the-art Transformer-based models (T5, BERT, XLNet, and SBERT). The empirical results demonstrate that token-based statistical models perform robustly on keyword-dependent syntax validation while Transformer-based models excel by capturing deep conceptual context. Specifically, Sentence-BERT (SBERT) achieved superior predictive accuracy across courses, yielding a minimum Root Mean Square Error (RMSE) of 0.9511 for the Introduction to Programming course, paired with moderate positive Pearson and Spearman correlation coefficients. Confidence intervals are estimated using Fisher’s z-transformation with a 95 percentile confidence level. Question-level five-fold and two-fold cross-validation is conducted for all four courses, and SBERT is found to perform better. Finally, to demonstrate the practical utility, we deployed a web-based portal for students’ evaluation leveraging a LLaMA-based pre-trained model to deliver real-time, personalized pedagogical feedback measuring code functionality, quality, documentation, and error handling. This reduces the educator workload while preserving grading integrity.