INF - Artigos publicados em periódicos

URI Permanente para esta coleçãohttp://200.137.215.59//handle/ri/1337

Navegar

Agora exibindo 1 - 20 de 106

Requirements engineering for machine learning-based AI systems: a tertiary study
(2025) Martins, Mariana Crisostomo; Campos, Lívia Mancine Coelho de; Soares, João Lucas Rodrigues; Kudo, Taciana Novo; Bulcão Neto, Renato de Freitas
Context: In the last decade, machine learning (ML) components have become more and more present in contemporary software systems. A number of secondary literature studies reports challenges impacting on the development of ML-based systems, including those for requirements engineering (RE) activities. Motivation/Problem: Synthesizing secondary literature contributes to building knowledge and reaching conclusions about the existing RE approaches for ML-based systems (RE4ML), besides the novelty of a tertiary study on that subject. Objective: Through a tertiary study protocol we elaborated on, this paper synthesizes the body of evidence present in secondary studies on RE4ML systems. Method: We followed well-accepted guidelines about tertiary study protocols, including automatic search, the snowballing technique, selection and quality criteria, and data extraction and synthesis. Results: Nine secondary studies on RE4ML systems were aligned to our tertiary study's goal. We extracted and summarized the requirements elicitation, analysis, specification, validation, and management techniques for ML-based systems as well as the great challenges identified. Finally, we contribute with a nine-item research agenda to direct current and future searches to fill the gaps found. Conclusions: We conclude that RE has not been left aside in ML research, however, there are still challenges to be overcome, such as dealing with non-functional requirements, collaboration between stakeholders, and research in an industrial environment.
Optimizing energy consumption for vRAN placement in O-RAN systems with flexible transport networks
(2025) Pires Junior, William Teixeira; Almeida, Gabriel Matheus Faria de; Correa, Sand Luz; Both, Cristiano Bonato; Pinto, Leizer de Lima; Cardoso, Kleber Vieira
Virtualized RAN (vRAN) matches O-RAN Alliance specifications while transitioning towards virtualized functions on general-purpose computing platforms. However, the energy consumption of these systems remains a major concern. Although this issue has been addressed in the literature, previous works oversimplify routing decisions, overlook the benefits of flexible split choices, or neglect the energy consumption of the transport network. Additionally, most studies employing optimal solutions exhibit very limited scalability due to their high computational time. In this work, we present a comprehensive and efficient Mixed Integer Linear Programming model to minimize the energy consumption of O-RAN systems, addressing the limitations of current approaches. We also design and implement a synthetic data generator to evaluate our model across various network usage profiles, topologies, and reasonablesize networks. We achieved valuable insights and promising results in our evaluation. For example, our results show that when devices require high throughput, the transport network incurs significant energy costs and reduces the centralization rate. We also observed that hierarchical RAN topologies can achieve greater energy efficiency than ring topologies, with our approach enabling up to 15% more centralization while saving around 28% of energy and consuming at least one order of magnitude less time than other strategies.
Advancements in artificial intelligence for colorectal cancer: A comprehensive overview of systematic reviews
(2025) Silva, Áurea Valéria Pereira da; Leitão Júnior, Plínio de Sá
Background: Colorectal cancer (CRC) is a leading cause of cancer-related mortality worldwide. Computational intelligence (CI) has emerged as a promising tool to improve diagnosis, staging, and treatment, but evidence remains scattered across the literature. Objective: This tertiary review aims to synthesize systematic reviews on CI applications in CRC care, highlighting algorithms, datasets, performance metrics, clinical scopes, and methodological gaps. Methods: A structured search in PubMed and EMBASE identified systematic reviews published between 2018 and 2023, following PRISMA guidelines. Twenty-two reviews were included. Extracted data covered CI techniques, evaluation methods, target outcomes, and dataset characteristics. Risk of bias was assessed using AMSTAR 2, and overlap of primary studies was analyzed through a correlation matrix. Results: The reviews addressed four clinical scopes: macroscopic lesion classification (colonoscopy), histological analysis, disease staging, and survival or treatment prediction. Convolutional neural networks (CNNs) were the most commonly used models. While some applications showed high performance (AUC 0.90), most reviews had low to moderate methodological quality. Key limitations included lack of external validation, dataset heterogeneity, and limited generalizability. Significant overlap was observed in studies focused on colonoscopy-based tasks. Conclusion: CI offers valuable contributions to CRC management, but broader clinical adoption is hindered by methodological inconsistencies and insufficient validation. This review provides a comprehensive synthesis to guide future research and promote the development of robust, explainable, and generalizable models for clinical use.
Blockchain for the carbon market: a literature review
(2025) Merlo, André Luiz Coutinho; Mendonça, Diogo Silveira; Santos, Joel André Ferreira dos; Carvalho, Sérgio Teixeira de; Guerra, Raphael Pereira de Oliveira; Brandão, Diego Nunes
The increasing urgency of climate change demands innovative solutions to reduce the environmental impact of human activities. While carbon credits have become a pivotal tool in reducing greenhouse gas emissions, their effectiveness is hindered by challenges such as lack of transparency, inefficiencies, and governance issues in carbon markets. This article uniquely explores how blockchain technology addresses these gaps by automating processes like verification and validation through blockchain-based smart contracts. Unlike previous reviews, our work focuses on platform interoperability and the integration of blockchain within the “3D’s” of decentralization, decarbonization, and digitization, providing a structured analysis of its potential to transform carbon markets. This comprehensive review synthesizes findings from academic papers, international reports, and technical documents, showing that blockchain technology improves transparency, reduces fraud, and ensures compliance with regulatory standards for the carbon market. However, interoperability between platforms and governance issues remain a challenge.
IsoPuzzle: development and evaluation of an ethno-educational serious game
(2025) Moura, Thiago Emanuell Vieira; Civardi, Jaqueline Araújo; Silva, Mary Anne Vieira; Berretta, Luciana de Oliveira; Carvalho, Sergio Teixeira de
This study presents the development and evaluation of IsoPuzzle, a digital ethno-educational game that merges progressive isometric geometry challenges with Adinkra symbolism to integrate mathematical concepts and African cultural elements. Based on Game-Based Learning (GBL) and the PEED methodology, IsoPuzzle promotes collaboration and active stakeholder participation. Its evaluation, conducted via a structured questionnaire derived from frameworks such as SUS, MEEGA+, PAJDE, and IAQJEd, focused on three indicators: pedagogical engagement, accessibility and usability, and the impact of cultural immersion on player experience (PX). Results indicate that IsoPuzzle effectively combines mathematical abstraction with engaging visuals, enhancing cultural connections through the use of Adinkra symbolism. The analysis also identified areas for refinement, particularly related to instruction clarity and interface accessibility, indicating the need to improve navigational intuitiveness. The data proved reliable, with the educational dimension averaging 4.16 (SD = 0.91), the cultural dimension 3.78 (SD = 1.02), and the user experience 3.61 (SD = 1.08). Variations in device quality further influenced user experience, underscoring the need for multi-platform optimization. Overall, these findings provide valuable insights for refining digital teaching tools that strike a balance between mathematical rigor, cultural diversity, and interface efficiency.
A metamodeling approach for planning critical IoT systems
(2024-12) Veiga, Ernesto Fonseca; Kudo, Taciana Novo; Bulcão Neto, Renato de Freitas
This paper presents a metamodeling approach to address the lack of methodological support in canvas model development, focusing on planning critical IoT systems. We introduce MM4Canvas - a metamodel that provides a solid foundation for developing structured and standardized canvas models, allowing for consistent reuse and extension across diverse project types. A proof of concept was conducted by instantiating a general-purpose canvas model basedon MM4Canvas for project planning, aiming to establish a connection between this activity and the Requirements Engineering process. We extended this model to incorporate safety and security requirements for critical IoT systems. Our contribution demonstrates the metamodel’s capacity to support standardization, reuse, and extensibility in canvas-based project planning.
Estratégias de inteligência artificial e Business Intelligenceaplicadas à imunização no contexto da pandemia de COVID-19
(2025) Lux, Eduardo; Silva, Laíla Pereira Gomes da; Santos, Wender Ferreira dos; Leitão Júnior, Plínio de Sá
Introduction: The application of Business Intelligence and Artificial Intelligence has been used in the health sector, especially in the area of immunization, which generates large amounts of data and health records. Objective: Investigate how artificial intelligence and business intelligence have been used to increase immunization. Method:Integrative review of studies selected through a search strategy in the databases, PubMed/MEDLINE, LILACS, SciELO, after meeting the eligibility criteria of the proposed protocol. Results: We obtained a total of 608 records, of which 75 had their titles and summaries read, 25 included. Of these, 5 duplicates were excluded, totaling 20 articles, of which 12 had a direct or indirect relationship with the theme of study in the context of the COVID-19 pandemic. Articles published between 2020 and 2023 were categorized according to the theme addressed as follows: Monitoring (4 studies); Vaccine Development (3 studies); and Health Communication (5 studies). Conclusion: The use of these strategies favors the context of immunization with a focus on the COVID-19 pandemic, pointing to the democratization of data to promote public policies, generating ease of access and reading of data to assist decision-making by management, as wellas helping to reduce the time for vaccine production.
Energy efficiency in network slicing: survey and taxonomy
(2025) Donatti , Adnei Willian; Machado, Marcia Cristina; López Martínez, Marvin Alexander; Antunes, Sabino Rogério da Silva; Souza, Eli Carlos Figueiredo; Correa, Sand Luz; Ferreto, Tiago Coelho; Monteiro, José Augusto Suruagy; Martins, Joberto Sérgio Barbosa; Carvalho, Tereza Cristina Melo de Brito
Network Slicing (NS) is a fundamental feature of 5G, 6G, and future mobile networks, enabling logically isolated virtual networks over shared infrastructure. As data demand increases and services diversify, ensuring Energy Efficiency (EE) in NS is vital (not only for operational cost savings but also to reduce the Information and Communication Technology (ICT) sector’s environmental footprint). This survey addresses the need for a comprehensive and holistic perspective on energy-efficient NS by reviewing and classifying recent strategies across the NS life cycle. Our contributions are threefold: (i) a thorough review of state-of-the-art techniques aimed at reducing energy consumption in NS; (ii) a novel taxonomy that organizes strategies into infrastructure, path/route, and slice operation levels; and (iii) the identification of open challenges and research directions, with a focus on systemic, cross-layer, and AI-driven approaches. By consolidating insights from recent developments, our work bridges existing gaps in the literature, offering a structured foundation for researchers and practitioners to design, evaluate, and improve energy-efficient network slicing systems.
Optimizing energy consumption for vRAN placement in O-RAN systems with flexible transport networks
(2024) Pires Junior, William Teixeira; Almeida, Gabriel Matheus Faria de; Correa, Sand Luz; Both, Cristiano Bonato; Pinto, Leizer de Lima; Cardoso, Kleber Vieira
Virtualized RAN (vRAN) matches O-RAN Alliance specifications while transitioning towards virtualized functions on general-purpose computing platforms. However, the energy consumption of these systems remains a major concern. Although this issue has been addressed in the literature, previous works oversimplify routing decisions, overlook the benefits of flexible split choices, or neglect the energy consumption of the transport network. Additionally, most studies employing optimal solutions exhibit very limited scalability due to their high computational time. In this work, we present a comprehensive and efficient Mixed Integer Linear Programming model to minimize the energy consumption of O-RAN systems, addressing the limitations of current approaches. We also design and implement a synthetic data generator to evaluate our model across various network usage profiles, topologies, and reasonable-size networks. We achieved valuable insights and promising results in our evaluation. For example, our results show that when devices require high throughput, the transport network incurs significant energy costs and reduces the centralization rate. We also observed that hierarchical RAN topologies can achieve greater energy efficiency than ring topologies, with our approach enabling up to 15% more centralization while saving around 28% of energy and consuming at least one order of magnitude less time than other strategies.
Online resource-aware video content recommendation in edge-caches for mobile users
(2023) Monção, Ana Cláudia Bastos Loureiro; Rodrigues, Karlla Bianca Chaves; Correa, Sand Luz; Cardoso, Kleber Vieira
The coupling of content caching at the wireless network edge and video streaming recommendation systems has been thoroughly investigated to enhance the cache hit and improve the user quality of experience (QoE). However, the existing literature lacks studies addressing the joint problem of QoE and cache hit ratio maximization while considering device characteristics and dynamic network resources of mobile users. This study introduces On-RAViR, an online framework comprising a Channel Quality Indicator (CQI) prediction module and a heuristic algorithm. This framework aims to maximize both cache hit ratio and user QoE under two constraints: the quality of the user equipment (UE) wireless link and the computing capabilities of the UE.We evaluate our framework employing a real-world video content dataset and a thirdparty 5G trace dataset. The results demonstrate that our framework produces rapid and high-quality solutions, increasing user QoE by 20% on average when compared to a state-of-the-art caching and recommendation heuristic unaware of computing and network resources.
Professors' perspective on a pedagogical architecture to requirements engineering education: a qualitative study
(2025) Santana, Thalia Santos de; Kudo, Taciana Novo; Santos, Davi Viana dos; Bulcão Neto, Renato de Freitas
Given the fundamental connection between requirements analyst training and high-quality software development, the disparity between requirements engineering (RE) education in academic institutions and the software industry needs is an ongoing concern. In this context, the conceptual framework of Pedagogical Architecture (PA), which organizes educational practice using digital technologies, is an alternative to enhance RE teaching and learning. We developed a PA for RE education to aid in directing the development of hard and soft skills crucial for field practitioners regarding requirements specification and validation. This paper describes a qualitative analysis from the professors' point of view who reproduced our PA, using Grounded Theory procedures as a data analysis technique. Three educators who have instructed four iterations of RE courses in higher education, following the guidelines set down by the PA, were questioned. The analysis confirmed that the PA assisted educators in working on practical tasks, incorporating hard and soft skills in line with RE professional practice. Moreover, the PA increased student participation and engagement and made educators aware of the value of support materials for replicating the PA.
Data sharing-based approach for Federated Learning tasks onEdge Devices
(2025) Oliveira, Renan Rodrigues de; Freitas, Leandro Alexandre; Moreira, Waldir; Ribeiro, Maria do Rosário Campos; Oliveira Junior, Antonio Carlos de
Federated Learning (FL) enables edge devices to collaboratively train a global machine learning model. In this paradigm, the data is maintained on the devices themselves and a server is responsible for aggregating the parameters of the local models. However, the aggregated model may present convergence difficulties when the device data are non-independent and identically distributed (non-IID), that is, when they present a heterogeneous distribution. This work proposes an algorithm that extends data sharing-based solutions from the literature by considering privacy-flexible environment, where users agree to share a small percentage of their private, and privacy-sensitive environment, where it is assumed that the aggregator server has a set of public global data that is shared with users in the initial phase of the FL process. The proposed algorithm is evaluated in a distributed and centralized way considering a Human Activity Recognition (HAR) application. The results show that data sharing strategies indicate improved global model performance in non-IID scenarios.
Fake news: a brief tertiary review through health, deep learning, and emerging perspectives
(2025) Gomes, Juliana Resplande Sant'Anna; Graciano Neto, Valdemar Vicente; Barbosa, Jacson Rodrigues; Lima, Eliomar Araújo de; Galvão Filho, Arlindo Rodrigues
Context: The proliferation of fake news represents a significant social threat, especially regarding health in-formation, a problem exacerbated by the COVID-19 pandemic. Deep Learning (DL) techniques are central to detection efforts, with increasing focus on health-related misinformation. Objective: This paper extends our previous work, synthesizing secondary studies (SS) on fake news detection, focusing on DL roles, the health domain, and recent trends (2022-2023). Method: A rapid tertiary review was conducted, analyzing 15 SS published between 2013 and August 2023, categorized by emphasis: DL applications, health misinformation, or recent publications. Results: A consistent dependence on DL and Natural Language Processing for text classification and fabricated media detection was identified. Health-focused or recent trend studies addressed challenges using specific datasets. Key challenges include echo chambers, cross-domain applications, early detection needs, and threats from generative models. Demands for transparency, blocking mechanisms, and Explainable Artificial Intelligence were highlighted. Conclusion: This review provides a synthesized view of research on fake news detection, emphasizing intersections with DL and health contexts, confirming the prevalence of core techniques despite diverse methodologies, and pointing to challenges requiring urgent attention.
A mini-map alignment approach using landmark observation information intrinsic to the feature-based visual SLAM pipeline
(2025) Saraiva, Felipe Pires; Laureano, Gustavo Teodoro; Oliveira, Thiago Henrique de
This work proposes a map alignment approach based on a simple modification of the Iterative Closest Point algorithm to use point confidences based on metrics usually available in feature-based visual SLAM pipelines. In the context of a hierarchical map composed of mini-maps, aligning these mini-maps is an important task to allow metric information to be related between them. This research enumerates three possible SLAM metrics that could be used for representing landmark confidence, and investigate the potential of using these metrics to improve the ICP algorithm. The experiments show evidence that the usage of the confidence metrics might help to improve the convergence of ICP with only a small modification to the data association step.
Perfect matching cuts partitioning a graph into complementary subgraphs
(2025) Castonguay, Diane; Coelho, Erika Morais Martins; Nascimento, Julliano Rosa; Silva, Hebert Coelho da; Souza, Uéverton dos Santos
In PARTITION INTO COMPLEMENTARY SUBGRAPHS (COMP-SUB) we are given a graph G = (V, E), and an edge set property Π, and asked whether G can be decomposed into two graphs, H and its complement H̄, for some graph H, in such a way that the edge cut [V(H), V(H̄)] satisfies the property Π. Motivated by previous work, we consider COMP-SUB(Π) when the property Π=PM specifies that the edge cut of the decomposition is a perfect matching. We prove that COMP-SUB(PM) is GI-hard when the graph G is C5-free or G is {Ck ≥ 7, C̄k ≥ 7}-free. On the other hand, we show that COMP-SUB(PM) is polynomial-time solvable on hole-free graphs and on P5-free graphs. Furthermore, we present characterizations of COMP-SUB(PM) on chordal, distance-hereditary, and extended P4-laden graphs.
Smart farming for poultry: enhancing growth and efficiency with low-cost internet of things solutions
(2025-04) Oliveira, Roberto Felício de; Hanau, Carla Nébele Ferreira; Graciano Neto, Valdemar Vicente; Lima, Eliomar Araújo de; Lopes, Vinícius Carvalho; David, José Maria Nazar; Villela, Regina Maria Maciel Braga; Arbex, Wagner Antonio; Kassab, Mohamad
This article investigates the impact of a low-cost Internet of Things system for autonomous environmental regulation in poultry farming, demonstrating its potential to optimize growth, welfare, and operational efficiency in small-scale production within the Brazilian agricultural context.
BM25 x Vila Sésamo: avaliando modelos Sentence-BERT para Recuperação de Informação no cenário legislativo brasileiro
(2025) Vitório, Douglas Álisson Marques de Sá; Pereira, Ellen Polliana Ramos Souza; Santos, José Antônio Pedro dos; Carvalho, André Carlos Ponce de Leon Ferreira de; Oliveira, Adriano Lorena Inácio de; Silva, Nadia Felix Felipe da
BERT-based models have been largely used, becoming the state-of-the-art for many Natural Language Processing tasks and for Information Retrieval. The Sentence-BERT architecture allowed these models to be easily used for the semantic search of documents, as it generates contextual embeddings that can be compared using similairty measures. To further investigate the application of BERT-based models for Information Retrieval, this work assessed 12 publicly available Sentence-BERT models for documents re- trieval within the Brazilian legislative scenario. Two BM25 variants were used as baseline: Okapi BM25 and BM25L. BM25L achieved better results, considering statistical significance, even in the scenario in which the documents were not preprocessed, while only one language model, fine-tuned using Brazilian legislative data, could reach a similar performance for one of the three used datasets.
Transformer Models improve the acoustic recognition of buzz-pollinating bee species
(2025) Ferreira, Alef Iury Siqueira; Silva, Nadia Felix Felipe da; Mesquita, Fernanda Neiva; Rosa, Thierson Couto; Buchmann, Stephen L.; Mesquita Neto, José Neiva
Buzz-pollinated crops, such as tomatoes, potatoes, kiwifruit, and blueberries, are among the highest-yielding agricultural products. The flowers of these cultivated plants are characterized by having a specialized flower morphology with poricidal anthers that require vibration to achieve a full seed set. At least 446 bee species, in 82 genera, use floral sonication (buzz pollination) to collect pollen grains as food. Identifying and classifying these diverse often look-alike bee species poses a challenge for taxonomists. Automated classification systems, based upon audible bee floral buzzes, have been investigated to meet this need. Recently, convolutional neural network (CNN) models have demonstrated superior performance in recognizing and distinguishing bee-buzzing sounds compared to classical Machine-Learning (ML) classifiers. Nonetheless, the performance of CNNs remains unsatisfactory and can be improved. Therefore, we applied a novel transformer-based neural network architecture for the task of acoustic recognition of blueberry-pollinating bee species. We further compared the performance of the Audio Spectrogram Transformer (AST) model and its variants, including Self-Supervised AST (SSAST) and Masked Autoencoding AST (MAE-AST), to that of strong baseline CNN models based on previous work, at the task of bee species recognition. We also employed data augmentation techniques and evaluated these models with a data set of bee sounds recorded during visits to blueberry flowers in Chile (518 audio samples of 15 bee species). Our results revealed that Transformer-based Neural Networks combined with pre-training and data augmentation outperformed CNN models (maximum F1-score: 64.5% ± 2; Accuracy: 82.2% ± 0.8). These innovative attention-based neural network architectures have demonstrated exceptional performance in assigning bee buzzing sounds to their respective taxonomic categories, outperforming prior deep learning models. However, transformer approaches face challenges related to small dataset size and class imbalance, similar to CNNs and classical ML algorithms. Combining pre-training with data augmentation is crucial to increase the diversity and robustness of training data sets for the acoustic recognition of bee species. We document the potential of transformer architectures to improve the performance of audible bee species identification, offering promising new avenues for bioacoustic research and pollination ecology.
Adoção da Inteligência Artificial no Schema Matching: um levantamento sistemático do Estado da Arte
(2025) Borges, Ricardo Henricki Dias; Graciano Neto, Valdemar Vicente; Ribeiro, Leonardo Andrade
The following text presents a synopsis of the abstract. Given the increasing intricacy of data integra-tion, attributable to both the proliferation of data and the diversification of its characteristics, Schema Matching is a critical component of this process. In the context of this challenging scenario, the applica-tion of Artificial Intelligence. The advent of Artificial Intelligence (AI) has emerged as a promising solution to enhance the efficiency of Schema Matching. The present article This paper presents the findings of a systematic literature review that investigated Artificial Intelligence (AI) techniques and algorithms. This is the most common usage in Schema Matching applications. The insights obtained provide valuable guidance. This text is intended for researchers and professionals who are seeking to improve data inte-gration through Schema Matching.
Risk-sensitive optimization of neural deep learning ranking models with applications in ad-hoc retrieval and recommender systems
(2025-07) Rodrigues, Pedro Henrique Silva; Sousa, Daniel Xavier de; França Júnior, Celso Renato; Magalhães, Gestefane Rabbi; Rosa, Thierson Couto; Gonçalves, Marcos André
We answer open research questions regarding the (hard) problem of incorporating risk-sensitiveness measures into Deep Neural Networks for ranking models of retrieval and recommender systems. Risk-sensitive measures are important for controlling the bias towards the average when optimizing ranking solutions’ effectiveness. In previous work, we proposed the RiskLoss function which presents two important adaptations for neural network ranking in ad-hoc retrieval: a differentiable loss function and the use of networks’ sub-portions, obtained via dropout, as baseline systems for optimizing risk sensitiveness. However, questions remained to be answered regarding the generality, cost, and applicability of our solution. In this article, we respond to these questions by (i) applying RiskLoss to ranking in recommender systems, (ii) analyzing the execution cost of RiskLoss and (iii) providing an experimental evaluation of RiskLoss’ resilience to overfitting. Our experiments, comparing seven loss functions on three benchmark recommendation datasets (AIV, ML35M, ML25M, ML100K and ML1M) and four Learning To Rank datasets (WEB30K, WEB10K, YAHOO and MQ2007), with thousands to millions of interactions, reveal that RiskLoss presents the most consistent risk sensitiveness behavior, with gains up to 4.5% in GeoRisk@10 without significant losses in effectiveness. In particular, RiskLoss can reduce the number of bad recommendations by over 11% for “hard to recommend” users. We also show that RiskLoss is not much affected by overfitting.

Navegar

Submissões Recentes