http://repositorio.ufc.br/handle/riufc/22343 2026-04-09T23:05:35Z http://repositorio.ufc.br/handle/riufc/85078 Título: Desenvolvimento de um Largue Language Model: uma estratégia de aprimoramento de transcrições de anamneses médicas Autor(es): Gonçalves, Yanna Torres Abstract: This work aims to explore the development of a Large Language Model (LLM) to enhance the transcription of patient records. The research involved two main stages: the first consisted of an initial benchmark of three language models (Phi3, Llama, and Mistral), evaluating their performance in the task of formatting anamneses, and the second focused on fine-tuning the Mistral model, based on qualitative and quantitative feedback from the evaluators. The Mistral model stood out in the benchmark due to its consistency and better results in agreement metrics, such as Kendall’s tau and Fleiss’ Kappa coefficients, although it was criticized for overloading details in its responses, impairing its objectivity. The Llama model performed the worst, with responses that were often irrelevant, while Phi3 showed moderate performance, but with issues in precision and excessive use of generic language. The fine-tuning of the Mistral model, despite attempts to correct the flaws, did not result in significant improvements, as the average evaluation score was slightly lower than the original model. The main contributions of this work include the comparison of language models for a specific task in the medical field and the analysis of the limitations of the fine-tuning process. The study also identified important limitations, such as data quality and potential bias among the evaluators, which may have influenced the results. As future work, the intention is to conduct longer training sessions, expand and diversify the dataset, and explore new machine learning approaches to further improve the model’s performance. Tipo: TCC 2024-01-01T00:00:00Z http://repositorio.ufc.br/handle/riufc/83894 Título: Ceará à vista: uma análise de artefatos gráficos cearenses do século XX em relação aos rótulos pernambucanos de Coutinho Autor(es): Cavalcante, João Victor Teixeira Abstract: This study compares graphic artifacts and informational systems of labels for cachaça and soap from Ceará with artifacts from Pernambuco analyzed by Coutinho (2011), considering the geographical proximity of the states and the origins of the Ceará press in Pernambuco. It also seeks to identify graphic artifacts from the 20th century that make up the graphic memory of Ceará and to analyze the organization and informational hierarchy of the labels from the perspective of information design. This is an exploratory qualitative-quantitative research with documentary research procedures, based on the methodology of Coutinho (2011). Twenty-four labels were collected from the collection of the Chico Albuquerque Museum of Image and Sound, and the labels for sugarcane brandy and soap, found in greater quantity in the collection, were analyzed using the Mijksenaar matrix model through a comparative procedure based on the analyses of the Pernambuco labels. Similar characteristics were evidenced between the labels of the two states, such as the influence of the theme in the creation of the artifact. With this, it was possible to perceive the approximation of the artifacts from Ceará to those from Pernambuco in terms of the themes addressed and the organization of the informational systems. Finally, it was concluded that the recognition of these printed materials as evidence of the expression of Ceará's graphic identity is a contribution to the consolidation of the field of studies of local graphic memory. Tipo: TCC 2025-01-01T00:00:00Z http://repositorio.ufc.br/handle/riufc/83874 Título: Stella Calendar: uma ferramenta de apoio à seleção das trilhas do curso de Sistemas e Mídias Digitais Autor(es): Cavalcante Filho, Jackson Pontes Abstract: This paper describes the refinement of the Stella web application, designed to assist students of the Digital Media Systems program at the Federal University of Ceará (UFC) in organizing their class schedules based on formative itineraries (tracks). The tool allows students to select courses, check for schedule conflicts, and view the weekly distribution of classes. It also offers export options in Portable Document Format (PDF) and Internet Calendaring and Scheduling Core Object Specification (iCalendar) formats for integration with digital calendars. The User Interface (UI) was developed using React for the frontend, Node.js for the backend, and MongoDB as the database, following a feature-based folder code organization pattern. Usability concepts were applied through prototyping in Figma, and the user experience was evaluated using the System Usability Scale (SUS). The results indicated a high level of usability and user acceptance, confirming the tool’s effectiveness in supporting academic planning. Tipo: TCC 2025-01-01T00:00:00Z http://repositorio.ufc.br/handle/riufc/83833 Título: Mecânicas de virada como suporte da espetacularização dos esports Autor(es): Araújo, Alan César Vieira de Abstract: eSports have established themselves as one of the most profitable segments of the digital gaming industry, with an estimated market value of $1.38 billion in 2022 and projected growth to $1.87 billion by 2025. The continuous development of these games involves frequent updates for balance adjustments and the introduction of new mechanics, ensuring technical quality and fostering active player communities, which are essential for organizing competitions. However, one of the biggest challenges eSports faces is maintaining audience interest. Repeating patterns can reduce viewer engagement, making broadcasts and tournaments predictable and less ap- pealing. In this context, Comeback Mechanics emerge as a crucial element in sustaining the unpredictability of matches, allowing disadvantaged players or teams to turn the game around and create exciting moments. This study aims to analyze Comeback Mechanics and their role in enhancing the spectacle of eSports, highlighting their impact on viewer experience and the growth of the industry. The game League of Legends was chosen for this investigation due to its relevance and influence in the competitive eSports scene. Its characteristics were analyzed, with a focus on initial choices, symmetry aspects, and how neutral objectives can influence a match to make it more thrilling. Tipo: TCC 2024-01-01T00:00:00Z