https://repositorio.ifgoiano.edu.br/handle/prefix/250 2026-03-13T08:00:17Z https://repositorio.ifgoiano.edu.br/handle/prefix/6275 Título: PROTÓTIPO DE UMA API REST PARA GERENCIAMENTO E ENTREGA DINÂMICA DE ATIVOS 3D EM REALIDADE AUMENTADA Autor(es): Conrado, Carlos Eduardo Primeiro Orientador: Silva, Alexandre Carvalho Abstract: With the growing competition in the corporate world, arising from the advent of so- called Industry 4.0 or the Fourth Industrial Revolution, there has been a need to leverage industrial processes by providing technological mechanisms that assist and enhance their optimization. Based on this context, this study is grounded in the integration of Lean Manufacturing—which is essentially the optimization of services aimed at maximizing value for the customer—and Industry 4.0, through the implementation of a technology developed using a REST API interface applied to an Augmented Reality system that can assist in the control and maintenance of electrical industry equipment. For this purpose, a complete REST API was developed with an emphasis on project scalability. Initially, the tools to be used were established, all of which were completely free, and from that point onward, the project followed a step-by-step process from modeling to implementation. For application testing, Black Box Testing was adopted, in which different functionalities were selected to analyze the expected and obtained responses. With the execution of this project, it was possible to conclude the high level of practicality and adaptability of systems that use the REST API architectural style, enabling developers to maintain flexibility when facing future maintenance and innovation demands. Editor: Instituto Federal Goiano Tipo: Trabalho de Conclusão de Curso 2026-02-23T00:00:00Z https://repositorio.ifgoiano.edu.br/handle/prefix/6174 Título: GERAÇÃO AUTOMATIZADA DE SELOS PARA PERSONALIZAÇÃO DE GAMIFICAÇÃO NO MOODLE USANDO INTELIGÊNCIA ARTIFICIAL GENERATIVA Autor(es): Barros, Vinicius Henrique Alves Primeiro Orientador: Costa, Newarney Torrezão Primeiro Membro da Banca: Silva, Lais Cândido Rodrigues Segundo Membro da Banca: Cardoso, Luciana Recart Abstract: This work presents the development and evaluation of a native Moodle plugin, called **local_selo**, created with the objective of automating the generation of personalized badges as a gamification strategy in Virtual Learning Environments. The solution integrates individual student preferences, collected through a form within Moodle itself, with a Generative Artificial Intelligence model capable of producing personalized images. Initially, a proof of concept was carried out using a language model based on ChatGPT in order to validate the feasibility of integrating Moodle with external AI services. After this preliminary stage, the development evolved to the use of the Stable Diffusion XL Base 1.0 model, made available by the HuggingFace platform, which allows the generation of visual badges from structured textual descriptions. The plugin was built following Moodle’s modular architecture, employing event observers, storage in dedicated tables, asynchronous queue-based processing, and automatic email delivery containing the generated badge. To evaluate user perception, an experiment was conducted with 16 students from the Computing area, who used the plugin in a real context and responded to the System Usability Scale (SUS) instrument. The results indicated good acceptance and ease of use, reinforcing the technical feasibility and pedagogical potential of the proposal. It is concluded that the **local_selo** plugin is technically viable, well integrated into Moodle, and capable of providing a personalized experience for students, although the results cannot be generalized due to convenience sampling. The work also points to promising perspectives for enhancing personalized gamification in Virtual Learning Environments with the support of generative AI. Editor: Instituto Federal Goiano Tipo: Trabalho de Conclusão de Curso 2025-12-05T00:00:00Z https://repositorio.ifgoiano.edu.br/handle/prefix/6147 Título: SISTEMA INTEGRADO PARA RECONHECIMENTO DE DESENHOS EM SMARTWATCHES E VISUALIZAÇÃO TRIDIMENSIONAL DE MOLÉCULAS EM REALIDADE VIRTUAL Autor(es): Santos, Jamilly Lima Primeiro Orientador: Vieira, Marcos Alves Primeiro Membro da Banca: Pereira Júnior, Cleon Xavier Segundo Membro da Banca: Lopes, Laís Candido Rodrigues da Silva Abstract: Three-dimensional molecular structure comprehension is a recurring challenge in Chemistry education, especially because many students experience difficulties in relating two-dimensional representations to more complex spatial models. This limitation affects the construction of fundamental concepts and may restrict the ability to visually interpret and manipulate different chemical forms. In this context, tools that combine gestural interaction and three-dimensional visualization offer relevant alternatives to support the development of spatial reasoning and to make the learning process more meaningful. This work presents the development and evaluation of an integrated system that combines drawing recognition on smartwatches with the three-dimensional visualization of molecules in a virtual reality environment to support Chemistry teaching. The smartwatch records user-drawn gestures representing simple chemical structures and sends the images to a remote cloud server, responsible for identifying the drawing through a Convolutional Neural Network. After identification, the corresponding molecule is displayed in three dimensions in a virtual reality environment on the smartphone, using Google Cardboard for visualization, enabling interactive exploration of the molecular structure through wrist movements. The adopted methodology included a literature review, prototype development, and an experiment conducted with 11 participants, including 4 professors and 7 undergraduate Chemistry teacher-training students. The system evaluation was carried out using the System Usability Scale (SUS), which provides a standardized measure of system acceptability from a functional perspective, and the User Experience Questionnaire – Short Version (UEQ-S), associated with interest, attractiveness, and user perception of the proposed solution. The results indicated high levels of usability and a positive user experience, suggesting that the system is functional and suitable for supporting the exploration of abstract content through gestures and three-dimensional visualization. These findings contribute to discussions on the use of wearable technologies and immersive environments in education, highlighting their potential to enrich teaching practices. The study conclusions point to opportunities for improvement, such as expanding the set of recognized molecules, enhancing image processing algorithms, and conducting tests in real school contexts in order to evaluate pedagogical impact on a larger scale. Editor: Instituto Federal Goiano Tipo: Trabalho de Conclusão de Curso 2025-12-05T00:00:00Z https://repositorio.ifgoiano.edu.br/handle/prefix/6143 Título: AVALIAÇÃO AUTOMÁTICA DE RESPOSTAS CURTAS EM PORTUGUÊS BRASILEIRO: UM ESTUDO SOBRE MODELOS DE LINGUAGEM, ENGENHARIA DE PROMPT E CARACTERÍSTICAS TEXTUAIS Autor(es): Santos, Heder Filho Silva Primeiro Orientador: Pereira Junior, Cleon Xavier Primeiro Membro da Banca: Vieira, Marcos Alves Segundo Membro da Banca: Lopes, Lais Candido Rodrigues da Silva Abstract: Automatic Short Answer Grading (ASAG) has emerged as a promising approach to reducing human effort in large-scale educational assessments, but studies focused on Brazilian Portuguese remain limited. This work evaluates the performance of three Large Language Models (GPT-4o-mini, Sabiazinho-3, and Gemini 2.0-Flash) in ASAG, testing all 128 possible combinations of seven prompt engineering components and examining how textual characteristics—such as word count and lexical richness—affect model accuracy. Results show that combining few-shot examples with explicit rubrics was the most effective strategy, while step-by-step reasoning particularly benefited GPT-4o-mini. Sabiazinho-3 achieved the highest agreement with human evaluators, Gemini 2.0-Flash obtained the lowest mean absolute error but exhibited a high hallucination rate, and GPT-4o-mini produced the cleanest and most consistent numeric outputs. Furthermore, the lexical profile of student responses significantly influenced model performance, with medium levels of lexical richness posing the greatest challenge across all models. Editor: Instituto Federal Goiano Tipo: Trabalho de Conclusão de Curso 2025-12-03T00:00:00Z