https://repositorio.pucgoias.edu.br/jspui/handle/123456789/38 2026-09-23T01:53:37Z https://repositorio.pucgoias.edu.br/jspui/handle/123456789/11065 Título: Sistema de registro de frequência para monitores acadêmicos em instituições de ensino superior Abstract: The attendance control of academic monitors is currently carried out in a predominantly manual manner, through electronic spreadsheets and physical forms, making the process susceptible to errors, inconsistencies, and weaknesses in record authenticity. Furthermore, the lack of centralized information hinders real-time monitoring by supervising professors and the monitoring coordination, overloading CAEME with administrative demands that could be resolved more autonomously among those directly involved. Faced with these limitations, this work presents the development of the Academic Monitoring Registration System (SRMA), a web application that automates attendance registration through biometric facial recognition, acting as a support tool among supervising professors, monitoring coordination, and CAEME. The system is structured into three access profiles (monitor, professor, and coordination) each with specific functionalities for managing monitoring activities. Grounded in the principles of Software Engineering, SRMA eliminates the need for manual records and trips to the administrative office, centralizing information in a single platform and ensuring greater reliability in the generated data. The results demonstrate that the solution significantly reduces operational rework, inhibits fraud in attendance registration, and promotes greater transparency and efficiency in academic monitoring management. Tipo: Trabalho de Conclusão de Curso 2026-06-09T00:00:00Z https://repositorio.pucgoias.edu.br/jspui/handle/123456789/11019 Título: LaborAI: agentes para geração automatizada de documentos em processos de licitação Abstract: This research project aims to develop a prototype system for the automated generation of documents in public procurement processes, focusing on the Terms of Reference and the Preliminary Technical Study, based on the guidelines of Brazilian Law 14.133/2021. The application, called LaborAI, was developed using React and Vite on the front-end, NestJS on the back-end, Supabase with PostgreSQL and pgvector as data infrastructure, and LangChain.js with LangGraph for the creation and orchestration of artificial intelligence agents. The architecture follows the client-server model with REST API, integrated with a knowledge base built using the RAG (Retrieval-Augmented Generation) technique, consisting of 522 indexed excerpts from Law 14.133/2021 and real procurement documents used in municipalities. The research is applied and exploratory-descriptive in nature, proposing a technological solution that guides civil servants through a structured conversation until the final document is generated in DOCX format. The system aims to assist in reducing errors, standardizing documents and decreasing repetitive data entry across multiple documents, without replacing subsequent technical and legal review. The results obtained reinforce the technical feasibility of using conversational agents in the context of public procurement, highlighting the potential of artificial intelligence in the modernization and efficiency of public administration. Tipo: Trabalho de Conclusão de Curso 2026-06-11T00:00:00Z https://repositorio.pucgoias.edu.br/jspui/handle/123456789/11018 Título: Fashionboost: sistema de gestão e fidelização para lojas de moda Abstract: This work aims to develop a SaaS (Software as a Service) system for small and medium- sized fashion store owners, called FashionBoost. The proposal arises from the observation that small retailers in the fashion sector generally do not have access to affordable customer management and loyalty tools, such as CRM systems and loyalty programs. The system was developed with NestJS and TypeScript on the back-end, TypeORM and PostgreSQL on the persistence layer, and Next.js 15 with Tailwind CSS v4 on the front-end. The adopted architecture follows the multi-tenant model, in which each retailer operates in a completely isolated environment within the same platform. Key implemented features include: product, inventory and category management; sales recording with discount coupon application; a loyalty program with point accumulation and automatic tier progression; and an artificial intelligence dashboard integrated with the Groq API using the Llama 3.3 70B model. The results demonstrate that FashionBoost is capable of supporting operational management and customer loyalty in independent fashion stores, answering the proposed research question with a viable and accessible technical solution. Tipo: Trabalho de Conclusão de Curso 2026-06-08T00:00:00Z https://repositorio.pucgoias.edu.br/jspui/handle/123456789/11004 Título: Desenvolvimento de IA para correção de avaliações de questões subjetivas Abstract: This study develops and evaluates an Artificial Intelligence system for the automated grading of short-answer conceptual questions, utilizing Natural Language Processing and Deep Learning techniques. The system employs the BERTimbau model—a BERT variant pre-trained on Brazilian Portuguese—combined with a Multi-Layer Perceptron classifier layer, fine-tuned to categorize answers into three classes: correct, partially correct, and incorrect. To overcome the scarcity of labeled data, a local Retrieval-Augmented Generation (RAG) architecture is adopted, based on TF-IDF vectorization and cosine similarity, which selects relevant excerpts from the textbook to guide the generation of synthetic data via the Gemini model API. The system's interface was developed using the Streamlit library, allowing instructors to configure the question, the answer key, and the reference material, initiate on-demand training, and obtain the class grade report within a single session. An Active Learning cycle enables instructor-corrected discrepancies to be reinjected into the training set, promoting the continuous evolution of the model. Experiments demonstrated that the pipeline is technically feasible, achieving an accuracy close to 100% on synthetic data, a grade distribution semantically coherent with the answer key, and a total execution time between two and five minutes on consumer hardware. It is concluded that the solution represents a relevant proof of concept for the automation of the assessment process, with the potential to reduce the time spent by teachers on routine grading and to be adapted to different disciplines without requiring a prior database. Tipo: Trabalho de Conclusão de Curso 2026-06-12T00:00:00Z