This paper investigates the practical applicability of Machine Learning (ML) techniques for buffer pool optimization in Database Management Systems (DBMSs), a critical component traditionally tuned manually in mission-critical environments. While prior work proposes ML-based approaches, most evaluations are limited to simulated settings, leaving their feasibility in production systems—particularly on mainframes—largely unexplored. To address this gap, we present an empirical, production-level validation of an automated, data-driven optimization methodology applied to a relational DBMS running on a mainframe. Our approach integrates Bayesian Optimization based on Gaussian Process Regression (BO-GPR) with a three-phase pipeline: Exploratory Factor Analysis with clustering for metric reduction, LASSO regression for parameter selection, and BO-GPR for fine-grained configuration tuning. We validate the methodology using real workloads from a large-scale financial system, achieving substantial performance gains, including up to a 95\% reduction in maximum synchronous I/O wait time. These results provide reproducible evidence that ML-based optimization is both effective and operationally viable in mainframe-based DBMSs.
This article presents a mixed method study conducted with elementary and secondary school teachers involved in a project that promotes interest in computing through computer science, robotics, and programming in Brazilian public schools. The study examines teachers' perceptions about the impact of the project on their professional development, student learning and participation, and surrounding school and community contexts. The results highlight the importance of early exposure to computing in fostering inclusion and social transformation. Teachers report positive outcomes, including the development of technical skills, increased motivation to teach, and significant improvements in the learning and participation of girls. Broader impacts at the school and community levels, such as increased parental participation, further underscore the role of teachers in initiatives aimed at broadening participation in computing.
Diseases transmitted by Aedes aegypti, such as dengue, zika, and chikungunya, are a significant challenge for public health in Brazil. Geolocation and georeferencing are relevant resources to enhance information systems that support the surveillance and control of such diseases. Despite its importance, many current methods of surveillance and control still rely on manual, paper-based processes that are slow and prone to spatial inaccuracies. This work presents the integration of advanced georeferencing resources into the ``Controle Vetorial Paraná'' system, a technological solution designed to modernize entomological surveillance for the state of Paraná, Brazil. The proposed approach shifts the operational focus from individual property addresses to block-based polygons, overcoming the precision limitations of public geocoding services. Key features include the implementation of the Ray Casting algorithm for the mandatory geographic validation of field visits and the generation of dynamic heat maps for strategic decision-making. The solution was validated through user-centered design stages with health coordinators and endemic control agents. Results demonstrate that the transition to a polygon-based model, combined with a 35-meter GPS tolerance zone, significantly enhances the reliability of field data and streamlines the planning of blocking routes, providing a more precise and efficient tool for public health management. The experience reported in this work is worthwhile for a wide range of studies on applying georeferencing in information systems.
Agent-Based Models (ABMs) are widely used to simulate complex systems through emergent behavior. Agent Based Modeling practice is constrained by the lack of systematic methods to evaluate and select frameworks for a given problem, and by the absence of protocols to study differences across computational implementations. We propose a Feature Space Maturity Model (FSMM) to evaluate how well a framework matches the requirements of a modeling task. We present results of four canonical ABM experiments to demonstrate the FSMM applied to five well-known frameworks. Our results suggest that combining the FSMM with the proposed protocol provides valuable information for both framework builders and modelers facing complex technical and research choices. Ensuring ABM reliability requires specification fidelity -- meaning both the conceptual correctness of the model and its faithful implementation. We study implementation fidelity by analyzing statistical similarity across multiple frameworks. We apply a macroscopic statistical approach to evaluate whether different ABM frameworks yield functionally equivalent results when executing the same specification. We designed a Pareto-based experiment and implemented it in six ABM frameworks. We defined three macroscopic observables and analyzed 10,000 simulation runs per framework using coefficient of variation, Fréchet distance, ANOVA, and F-tests. Five of the six frameworks show statistically indistinguishable behavior, while one exhibits consistent deviations. The results highlight the value of statistical validation to identify implementation inconsistencies and optimize resource allocation in large-scale simulations.
Graph classification plays a central role in many scientific disciplines. While classical kernel-based methods and graph neural networks achieve strong predictive performance, they often require substantial computational resources. Hyperdimensional Computing (HDC) has recently emerged as an efficient and noise-resilient alternative, providing lightweight models that are attractive for resource-constrained settings. Within this context, GraphHD is a representative HDC-based approach for graph classification; however, its encoding process can become costly on large graphs and its standard configuration relies on a single centrality choice (PageRank) for node-to-hypervector assignment. In this work, we go beyond PageRank in GraphHD by systematically evaluating alternative centrality measures (degree, closeness, betweenness, Katz, and eigenvector) and by introducing two new encoding variants. GraphHD-Level preserves quantitative structural information by mapping centrality values to level-hypervectors, whereas GraphHD-Order simplifies the algorithm by eliminating edge encoding and aggregating node hypervectors directly. Experiments on six widely used benchmarks from cheminformatics and bioinformatics (MUTAG, ENZYMES, PROTEINS, DD, NCI1, and PTC\_FM) show that replacing PageRank with alternative centralities yields similar F1-scores while offering notable runtime savings, and that GraphHD-Order remains competitive with the original GraphHD baseline while providing consistent speedups in encoding time.
Digital transformation represents a fundamental opportunity for organizations to incorporate sustainability criteria into their strategic decisions. In this context, IT governance and management frameworks—such as COBIT, ITIL, CMMI, and Six Sigma—provide structures to guide technology projects aligned with economic, environmental, and social objectives. This article presents a Systematic Literature Review (SLR) that analyzed 61 primary studies selected from 126 initial records. The research work was based on the Kitchenham methodology and the PRISMA 2020 statement; to identify how these technological governance and management frameworks act as facilitators in the adoption of sustainability indicators under GRI Standards. The results reveal a clear association between cost reduction and increased productivity within the economic dimension; waste reduction reflects progress in the environmental dimension; and the improvement of quality and value delivered to stakeholders aligns with the social dimension. In conclusion, sustainability is not an abstract concept, but a measurable result of mature process management; which allows organizations to make informed strategic decisions and communicate their performance in a comprehensive manner.
Public administration software systems are increasingly adopting microservices architectures to achieve scalability, flexibility, and resilience. However, the complexity of distributed systems poses challenges to access authorization management. This study presents the SIARE-Artefactos solution, which is designed to automate the registration and authorization of resources within Paraguay’s Integrated State Resource Management System (SIARE). SIARE-Artefactos leverages OAuth 2.0 and Spring Boot Starter to ensure secure and efficient authorization processes. SIARE-Artefactos leverages OAuth 2.0 and Spring Boot Starter to ensure secure and efficient authorization processes. SIARE-Artefactos has been implemented and used by public employees in Paraguay, enabling validation with real users. For the validation we adopted a mixed-methods, combining quantitative with qualitative sources, and a longitudinal observational design. The convergence of evidence allowed us to correlate the objective reduction in deployment times (from minutes to seconds) with the subjective perception of 'system predictability' reported by the DevOps team. Validation of the proposal demonstrates the solution's replicability to other cases with similar challenges and resources, as well as its ability to significantly improve system reliability and reduce configuration time by 74%. This highlights its potential to transform large-scale public administration systems using modern DevOps practices and agile methodologies.
This article presents an in-depth analysis of ArSGam, a modular architecture designed to integrate gamification strategies into serious games for cognitive-affective training. The original work introduced four interconnected modules that guide the process from user profiling to the integration of game mechanics. This extended version strengthens the scientific contribution by conducting additional experiments with ArSGam and employing Large Language Models (LLMs ) as tools for generating synthetic data to expand testing scenarios. Furthermore, new case studies and validation analyses were performed to examine the architecture's operational consistency and applicability in high-risk training contexts. A comparison with related architectures is also provided to contextualize the distinctive features of ArSGam. The results reaffirm the relevance of incorporating cognitive-affective profiling into gamification design and highlight the flexibility of the architecture for supporting personalized serious game experiences. Consequently, this study consolidates ArSGam as a robust and versatile framework for integrating cognitive-affective personalization and gamification within serious games.
Technology played a crucial role in combating the COVID pandemic, both in the rapid development of vaccines and the early detection of the virus. Consequently, numerous studies in the medical field have focused on leveraging the power of artificial intelligence for COVID-19 detection. However, in the medical domain, it is essential to have a clear understanding of the processes and algorithms used in decision-making, as these directly impact people’s health. Therefore, efforts have been made to implement explainable artificial intelligence techniques, enabling humans to understand and explain the deep learning algorithms used in disease detection. In this work, we present an approach to detecting COVID-19 in chest X-rays that combines reconstruction-based anomaly discov- ery with perturbation-based attribution. Specifically, we use a Variational Autoencoder trained on healthy lungs to identify lung anomalies, and we additionally evaluate a Per- ceptual Autoencoder (PAE) that incorporates a perceptual loss to produce sharper recon- structions and more contrastive residual maps. These signals are then used to highlight critical image evidence through both patch-level scoring (grid-based occlusion) and pixel- level masking derived from reconstruction-error maps, enabling healthcare professionals to localize relevant regions more effectively. Moreover, the proposed framework provides clearer explanations of the model’s decisions by quantifying the prediction change after occluding the identified regions, reducing the complexity of the “black boxes” generated by deep learning neural networks. With this methodology, we aim to improve the effec- tiveness and reliability of early COVID-19 detection through chest X-rays.
CalcLogic is a proof assistant adapted for students of Calculational Logic. CalcLogic uses the same syntax and theorems as in Gries and Schneider's book, ``A Logical Approach to Discrete Math''. All proofs can be done by students using only the mouse, without the need to learn any language. To mechanize the assistant, a new proof theory for Calculational Logic is developed in this paper. This theory allows the loading of first-order theories axiomatized with an axiom schemes. Also is presented a study, using statistical evidence, that shows a significant improvement in learning Calculational Logic when the formative activities are made by means of the assistant.
The present issue of CLEI Electronic Journal (CLEIej) is devoted to extended and revised papers of selected works presented at the 51st Latin American Informatics Conference (CLEI 2025), held in Valparaíso, Chile, from October 27th to 31st, 2025. This is the second volume of selected articles from the conference. The CLEI 2025 conference included five major tracks on different computer science and informatics areas, particularly: Computer Technology, Intelligent Systems, Software Systems, Systems In Practice, and Computer Science Education. A total of 275 submissions were received in these tracks; 91 were accepted as full papers and 23 as work-in-progress. It resulted in an acceptance rate of 33,1% for full papers. Based on the recommendations of the track chairs, we invited the authors of the 20 papers with the highest score to extend their work and submit it to this special issue. The authors of 14 papers accepted the invitation, and submitted an extended version of the previous article. These articles were reviewed by external researchers and revised by the authors accordingly. This process resulted in five papers being accepted and already published in the first volume of this special issue, and six papers accepted for publication in this second volume. Next we introduce the articles included in this volume of the special issue: “Practical Machine Learning for Self-Tuning Buffer Pools in DBMS Mainframe Systems”. This work focuses on the automated optimization of buffer pools in database management systems running on mainframes, a critical area for performance in high-demand enterprise environments. Its main contribution is an empirical validation in a real production setting, applying a methodology that integrates exploratory factor analysis, LASSO regression, and Bayesian optimization with Gaussian processes to tune buffer pool configurations based on actual workloads. Tested in a large-scale financial system, the approach achieved substantial performance improvements, including up to a 95% reduction in maximum synchronous I/O wait time, demonstrating that machine learning techniques are both practical and effective in mission-critical mainframe environments. “Enhancing an Arbovirus Control System with Georeferencing and Geolocation Resources”. This paper addresses the enhancement of arbovirus control systems, through the integration of advanced georeferencing and geolocation technologies. Focusing on dengue and other mosquito-borne diseases in Brazil, the study introduces innovative features that shift operational focus from postal addresses of individual properties to block-based polygons of areas, significantly improving spatial accuracy and data reliability. The system, tailored specifically for field agents and health coordinators, facilitates precise visit validation, allows building dynamic heat maps, and optimized route plans, ultimately enabling more targeted and efficient vector control efforts. The main contribution of this work lies in developing a modular, scalable geospatial architecture embedded within existing health management platforms, which has demonstrated substantial improvements in operational effectiveness and strategic decision-making in disease surveillance. “Beyond PageRank in GraphHD: Centrality Metrics and Efficient Hyperdimensional Encodings”: This article addresses graph classification within the field of Hyperdimensional Computing (HDC). It proposes efficient hypervector encoding methods that extend the GraphHD framework, which is the basis of this proposal. The authors systematically evaluate alternative graph centrality indicators—e.g., degree, closeness, betweenness, Katz, and eigenvector—in place of the standard PageRank to assign node hypervectors. Additionally, they introduce two novel encoding variants: GraphHD-Level, which preserves quantitative centrality information, and GraphHD-Order, a simplified method that omits edge encoding altogether. Experimental results on six benchmark datasets demonstrate that replacing PageRank with alternative centralities maintains comparable classification performance, while significantly reducing encoding time. Notably, GraphHD-Order achieves similar accuracy to the baseline, but offers consistent speedups (especially on large graphs), making it an attractive lightweight alternative for resource-constrained scenarios. This study highlights the potential of HDC for efficient and robust graph representation and classification tasks. “IT Governance and Corporate Sustainability: Decision Frameworks for Digital Transformation Aligned with GRI Standards”: This study explores the role of IT governance and management frameworks—such as COBIT, ITIL, CMMI, and Six Sigma—in facilitating organizational sustainability within the context of digital transformation, aligning with Global Reporting Initiative standards. Through a systematic literature review that analyzed 61 primary studies, this work identifies how these frameworks support economic, environmental, and social sustainability indicators, by improving operational efficiency, waste reduction, and quality enhancement. The research highlights that IT frameworks not only optimize internal processes, but also enable the conversion of technical data into measurable sustainability metrics, thereby informing strategic decision-making and promoting long-term sustainable value creation. This work contributes with a structured decision-making guide for organizations seeking to integrate digital transformation initiatives with sustainability objectives. “Early Computing Education in Brazilian Public Schools: Insights from Teachers”: This paper addresses the Computing Education field, with a particular focus on K–12 computer science education, teacher professional development, and gender inclusion in STEM. It presents a mixed-methods study that investigates teachers’ perceptions of the long-running Meninas.comp outreach initiative, which introduces programming, robotics, and computational thinking into Brazilian public schools. Unlike most previous studies that primarily assess students’ experiences, this work places teachers at the center of the evaluation, providing new insights into their professional development, pedagogical practices, and perceived impacts on students, schools, and local communities. The study demonstrates that participation in the project strengthens teachers’ technical and pedagogical skills, increases student engagement—particularly among girls—and fosters broader institutional and social benefits, highlighting the crucial role of teachers as agents of sustainable and inclusive computing education. “Agent-Based Modeling and Simulation - Feature Spaces, Macroscopic Statistical Analysis, and Specification Fidelity”. This work focuses on evaluating Agent-Based Modeling (ABM) frameworks, by proposing the Feature Space Maturity Model to systematically assess how well frameworks meet modeling task requirements. It also introduces a macroscopic statistical protocol to validate implementation fidelity across multiple frameworks. Testing six frameworks on a standardized experiment, the study finds that five of them produce statistically equivalent results, highlighting the value of statistical validation for ensuring reliable and reproducible ABM simulations. After concluding the process of preparing, processing and publishing these two-volume special issue, we would like to express our deep appreciation to all the CLEI 2025 Track Chairs for their collaboration throughout the paper review process, and their vital role in evaluating and selecting the papers invited to this special issue. We also extend our sincere thanks to the reviewers, who selflessly dedicated their time and expertise to assess the content and quality of the papers featured in this issue. Finally, we are grateful to the authors for accepting the invitation to extend their conference papers, and incorporate the reviewers' feedback to create these high-quality extended versions of the manuscripts. We hope you find this special issue as insightful and enjoyable as we do, as it helps to disseminate significant research from the Latin American Computer Science community. Ignacio Araya, Escuela de Ingeniería Informática, Pontificia Universidad Católica de Valparaíso (Chile) Carlos Luna, Instituto de Computación, Universidad de la República (Uruguay) Sergio Ochoa, Departamento de Ciencias de la Computación, Universidad de Chile (Chile) CLEIej Guest Editors for CLEI 2025 Special Issue - Volume II
The Semantic Web enables data exchange and knowledge exploitation through SPARQL queries; however, users often encounter difficulties when formulating such queries, leading to unexpected results and iterative trial-and-error processes. These issues are aggravated by incomplete knowledge of the application domain and the underlying ontological schema, frequently causing unnecessary query executions and increased computational load on RDF repositories. This work proposes a method for the prior syntactic and semantic validation of SPARQL queries, based on an ontology and independent of its instances. The method operates before query execution and detects inconsistencies arising from literal writing errors and violations of object property axioms, providing explicit feedback to users. The development of the proposed approach follows a classical methodological framework grounded in the software engineering process. Experimental results show that the additional computational overhead introduced by validation is minimal, while the benefits are significant in scenarios involving incorrect queries. In particular, prior validation prevents the execution of erroneous queries, reduces processing time, and maintains negligible overhead for correct queries, demonstrating a favorable trade-off between computational efficiency and reduced manual intervention. Overall, the results indicate that prior validation improves SPARQL query quality and enhances developer productivity in semantic data–driven applications.
Multi-access Edge Computing (MEC) extends cloud computing capabilities to the network edge to reduce latency and enhance user experience. However, the constrained resources of edge devices and their exposure to cyber attacks make secure and efficient authentication a critical challenge. Although numerous authentication protocols have been proposed for MEC, the absence of standardized evaluation frameworks has hindered systematic comparison among existing solutions. Based on prior literature, this paper presents a comparative analysis of representative authentication protocols for MEC in an organized manner, evaluating and contrasting them according to their cryptographic foundations, computational and communication overhead, support for features such as mutual authentication and single sign-on, and resilience against established attack models. Our findings reveal that more recent protocols based on elliptic-curve cryptography generally offer improved security and performance compared to earlier approaches. Nevertheless, several protocols continue to exhibit weaknesses, including susceptibility to impersonation attacks, lack of perfect forward secrecy, and reliance on centralized authentication entities. Furthermore, our study also exposes significant inconsistencies in experimental setups and performance metrics across prior studies, which complicate fair and reproducible comparisons. Ultimately, our study concludes by identifying key gaps and research directions for the development of more robust and lightweight authentication mechanisms, as well as the establishment of standardized criteria for comparative evaluation in MEC environments.
Conventional fire detection systems, which depend on smoke and fire sensors, frequently have problems with calibration, susceptibility to environmental changes, and the requirement for ongoing maintenance. These flaws can result in unfavorable false alarms. Reducing the amount of damage caused by fire requires prompt and precise identification of potentially dangerous fire incidents. In response, this paper proposes a vision-based fire and smoke detection system based on computer vision and convolutional neural networks (CNNs). The main goal is to create a novel model that is built on three different architectures: Inception-v3, VGG-19, and VGG-16, each of which is customized by adjusting layers and hyperparameters. For a thorough assessment, the research integrates two publicly available datasets, BoWfire and Fire-Flame, and presents a brand-new NoFiSe dataset. Results show that the Inception-v3 model is quite effective; it outperforms other techniques with 91.11%, 94.99%, and 98.03% accuracy rates in the Fire-Flame, NoFiSe, and BoWfire datasets, respectively. The main novelty of this work is the integration of customized multi-scale CNN architectures with a newly developed multi-class fire/smoke dataset to achieve robust and generalized early fire detection under diverse real-world conditions. The study highlights the potential of intelligent fire suppression and emphasizes the need for continued research to improve current models in order to reduce future fatalities.
Evolutionary Algorithms (EAs) are population-based, stochastic search algorithms that emulate natural evolution. Over the years, EAs have been successfully applied to numerous classification problems. This paper proposes an enhanced evolutionary algorithm for synthesizing classifiers in supervised data scenarios. The algorithm is rooted in Genetic Programming (GP), a form of optimization inspired by biological evolution. GP evolves computer programs that perform specific tasks, represented as sets of genes manipulated by a Genetic Algorithm (GA). Our approach generates a Directed Acyclic Graph (DAG) for each training class to align each test class with one of them. Additionally, the approach incorporates a fitness evaluation function with dual objectives: the cumulative evaluation values of test samples and the count of prediction errors in classes. We benchmark our approach against established machine learning methods such as Support Vector Machine (SVM), K-Nearest Neighbor (K-NN), and Particle Swarm Optimization (PSO). The enhanced algorithm demonstrates competitive performance, achieving an accuracy of over 70% when evaluating evolving classifiers across twelve public UCI datasets.
The credibility and efficiency of REDD+ projects rely substantially on the implementation of transparent MRV mechanisms. To tackle this, Blockchain technology has been considered in a plethora of REDD+ projects, since it may enhance trust, traceability, and automation for carbon-credit management. Within this context, this paper particularly presents a comparative analysis of three well-matured blockchain-based REDD+ projects: Ambify, MCO2, and TreeCycle. We assess three pivotal performance metrics: block-finalization traffic, block-finalization time, and systemic security. The first two metrics are derived from computational simulations, whereas the third is obtained through a theoretical security analysis. Overall, the experimental findings reveal that the analyzed projects exhibit distinct performance and security profiles that are strongly influenced by their underlying Blockchain platforms and consensus mechanisms. In particular, architectures optimized for high throughput demonstrate lower block-finalization traffic and latency under transaction-intensive scenarios, while more decentralized designs offer different security and governance trade-offs. The primary contribution of this work lies in providing both theoretical and empirical insights to support the evaluation and design of future Blockchain-based REDD+ initiatives under varying operational contexts. General findings and recommendations for future research conclude this paper.
India is grappling with a serious air pollution crisis that threatens the health and well-being of its large population. Primary sources of air pollution include industrial emissions, vehicular exhaust, agricultural residue burning, and dust. Fine particulate matter (PM2.5), in particular, contributes to millions of deaths and substantial economic losses. Accurate prediction of the Air Quality Index (AQI) is vital for tackling air pollution, as it enables early warnings, supports data-driven decision-making, and facilitates the formulation of effective control measures. To improve AQI forecasting, this paper presents a novel hybrid approach combining Random Forest (RF) and Long Short-Term Memory (LSTM) techniques. The hybrid RF-LSTM model utilizes the RF algorithm to extract relevant features from the AQI dataset, which are then input into the LSTM network to capture temporal dependencies and long-term patterns in the sequential data. The AQI data, with extracted features for five Indian cities Delhi, Lucknow, Chandigarh, Jaipur and Gurugram, is input into the LSTM model, and the prediction results of the hybrid RF-LSTM model are then analysed and presented. The results indicate that the hybrid RF-LSTM approach outperforms both individual RF and LSTM models in predicting various AQI constituents, thereby offering improved support for addressing air pollution.
Industry 4.0 emphasizes connectivity and digitalization, transforming business operations worldwide. In Brazil, however, the adoption of these technologies remains relatively limited, with only a small proportion of small and medium-sized enterprises (SMEs) fully engaging in digital transformation initiatives. Such engagement is critical to the national economy, yet SMEs face substantial barriers, including high implementation costs and limited technical expertise. This study focuses on SMEs in the food sector in S\~{a}o Paulo, aiming to provide strategic guidance for adopting Industry 4.0 practices and Information and Communication Technologies (ICTs) to support their technological transformation and improve digital maturity and competitiveness. To establish the theoretical foundation, a comprehensive literature review was conducted. This was followed by interviews with 10 food sector SMEs, using an online questionnaire of 21 questions organized into five categories: Technological Infrastructure, Automation, Management, Strategy, and Market. The results revealed that, while SMEs are interested in technologies such as QR codes, pagers, and social media, they face obstacles related to training, knowledge, and implementation costs. Nonetheless, they acknowledge the importance of digital transformation in their operations. Based on the data collected, a set of tailored recommendations was developed to support SMEs in planning and implementing effective digital transformation strategies.
