Database Systems Journal

ISSN 2069 - 3230

The journal is published under the sponsorship of
The Bucharest University of Economic Studies
and it is produced by the university's own publishing division,
The Bucharest University of Economic Studies Publishing House


Database Systems Journal, Vol. XVI, 2025


Open PDF Journal


CONTENTS


1. The Use of Communication Platforms in Military Operations: Enhancing Strategic and Tactical Effectiveness (p. 1-10)
Rexhep MUSTAFOVSKI, Ss. Cyril and Methodius University, Skopje, Republic of North Macedonia
The way communication platforms are used in military operations has changed a lot over the years. They're now essential for mission success, quick decision-making, and maintaining strategic advantages. This paper dives into the modern communication tools that are making waves in military settings, with a spotlight on networked communication, signal support, cybersecurity strategies, and how different forces work together in joint and multinational missions. It also tackles some of the major hurdles, like congestion in the electromagnetic spectrum, cyber threats, and the need for secure data transmission in challenging environments. By pulling insights from FM 6-02: Signal Support to Operations and the latest scientific research, this paper highlights recent advancements in military communication tech and how they boost operational effectiveness. Additionally, the research looks ahead at future possibilities, such as AI-driven communication platforms, quantum encryption, and cutting-edge satellite networks for defense purposes. This paper's primary contribution is the development of a structured, AI-enabled communication workflow that integrates quantum-safe encryption, blockchain authentication, and satellite-based coordination to improve decision-making and resilience in multi-domain operations.
Keywords: Military communication, secure networks, battlefield connectivity, radio systems, cybersecurity, signal support, tactical communications, AI in defense
2. Query Completion for Small-Scale Distributed Databases in PostgreSQL and MongoDB (p. 11-27)
Marin FOTACHE, Alexandru Ioan Cuza University of Iasi, Romania
Catalina BADEA, Alexandru Ioan Cuza University of Iasi, Romania
Marius-Iulian CLUCI, Alexandru Ioan Cuza University of Iasi, Romania
Codrin-Stefan ESANU, Alexandru Ioan Cuza University of Iasi, Romania
Relational/SQL and document/JSON data stores are competing but also complementary technologies in the OLAP (On-Line Analytical Processing) systems. Whereas the traditional approaches for performance comparison use the duration of queries performing similar tasks, in this paper we compare the performance of two distributed setups deployed on PostgreSQL/Citus and MongoDB by focusing only on the query's successful completion within a 10-minute timeout. The TPC-H benchmark database was converted into a denormalized JSON schema in MongoDB. An initial set of 296 SQL queries was devised for execution in PostgreSQL/Citus and then mapped for execution in MongoDB using Aggregation Framework (AF). Query execution success within a 10-minute timeout was collected for both PostgreSQL and MongoDB in six scenarios defined by two small-scale data factors (0.01 and 0.1 GB) and three different node counts (3, 6, and 9) for data distribution and processing. The relationships between the query completion and the query parameters were assessed with statistical tests and a series of machine learning techniques.
Keywords: PostgreSQL, Citus, MongoDB, SQL, Aggregation Framework, OLAP performance comparison
3. An Overview of Big Data and NoSQL in the Video Game Industry (p. 28-36)
Cristiana COSTAN, The Bucharest University of Economic Studies, Romania
The connection between relational and non-relational databases and game development is covered in this article. The concept of big data and its implications in the gaming industry is also stated, the capability to store user data becoming essential for marketing and game improvements. The ideas in this article are reinforced by creating a web application to see how NoSQL databases can be implemented in game development and how they can help us extract relevant data about users. Overall, the findings of this paper are intended to guide people who want to be part of or are part of the gaming industry in the choices related to which database to utilize in future games and to provide insights on how they might obtain information about potential consumers to improve their products or services.
Keywords: Big Data, NoSQL, Video Games, Gaming Industry
4. Evaluating Deep Learning and Machine Learning Models in Federated Learning for Credit Card Fraud Detection. A comparative study (p. 37-44)
Sener ALI, The Bucharest University of Economic Studies, Romania
This study aims to evaluate the effectiveness of both machine learning and deep learning algorithms for credit card fraud detection in the context of a federated learning framework. The fast evolution of digital banking created better experiences for customers and facilitated their access to financial services, but it has simultaneously opened new pathways for cybercriminals, making real-time fraud detection essential. Both models used in this study, XGBoost and a neural network, were trained on a publicly available dataset containing highly imbalanced data, reflecting realistic fraud scenarios. Results demonstrate that both models achieved high accuracy, yet the neural network consistently outperformed XGBoost across critical metrics such as precision, recall, and F1 score. This indicates a superior ability of deep learning models to detect fraudulent transactions in federated learning environments, highlighting their potential to improve financial security through collaborative yet privacy-preserving approaches.
Keywords: Federated Learning, Credit Card Fraud Detection, Privacy Protection, Machine Learning, Deep Learning
5. The Advantage of NoSQL Databases over SQL Databases (p. 45-54)
Mario-Tudor CHIRIAC, The Bucharest University of Economic Studies, Romania
The aim of this article is to show the advantages of No-SQL databases compared to SQL (relational) databases. Traditionally, the concept of database is implicitly associated with relational databases, most often ignoring the huge potential that non-relational databases have in the fields such as BigData, Data Analysis, Artificial Intelligence. Besides, non-relational databases are also remarkable for their structure flexibility, their speed and for the distribution of their data on multiple servers simultaneously, for their fast writing and reading speed, which is crucial when analysing very large volumes of data.
Keywords: Relational databases, non-relational databases, No-SQL databases, SQL databases, SQL, No-SQL, redundancy, efficiency, scalability, portability, large volume of data, big data, management of large volumes of data, big data flow
6. House Market Prediction Using Machine Learning (p. 55-64)
Nicusor-Andrei ANDREI, The Bucharest University of Economic Studies, Romania
This study explores and compares the performance of tree-based machine learning algorithms' predictions for Bucharest real estate market prices. The dataset was obtained from a local platform in March 2025 and contains residential apartments for sale in Bucharest. The comprehensive data preprocessing step, including imputation of missing values, encoding of categorical variables, and the engineering of new key features such as distance to public transport, played a key role in the models' performance. The models were optimized using a grid search algorithm with 5-fold cross-validation and evaluated with key performance indicators including root mean squared error, mean absolute error, and coefficient of determination. The results indicate that XGBoost outperforms both Random Forest and a single Decision Tree, reducing all the key performance indicators used in analysis.
Keywords: House market, Machine learning, Price prediction, Tree-based machine learning, XGBoost
7. Detection of Fake News Using Deep Learning and Machine Learning (p. 65-81)
Gabriela CHIRIAC, The Bucharest University of Economic Studies, Romania
Ada Maria CATINA, The Bucharest University of Economic Studies, Romania
Automatically identifying fake news is a complex challenge, involving detailed knowledge of how fake news is propagated and advanced data processing technologies. The use of Machine Learning and Deep Learning algorithms for misinformation detection involves a continuous learning process as manipulation methods are constantly evolving. Effective detection of fake news requires constant adaptation of algorithms to keep up with new disinformation methods. While these technologies offer promising solutions, the challenge is to calibrate them properly so that they work optimally in different contexts.
This paper explores automated methods for detecting fake news, analyzing the effectiveness of the various techniques and how they can be improved to face the challenges of data quality, domain variability and the continuous evolution of disinformation strategies.
Keywords: misinformation, Machine Learning, Deep Learning, fake news