https://repositorio.uema.br/jspui/handle/123456789/1915 2026-07-24T02:26:43Z https://repositorio.uema.br/jspui/handle/123456789/6316 Título: Alucinações em modelos de IA Generativa: limitações e desafios na compreensão da linguagem humana Abstact: Chatbots are artificial intelligence systems that simulate human interactions and have been widely used in diverse areas, from customer service to content creation. However, despite advances in technology, chatbots still present limitations and challenges, such as the tendency to make mistakes and the difficulty in dealing with complex questions and nuances of human language. These errors are known as hallucinations and can be caused by several factors, such as a lack of common sense, biased data and limitations in the models' ability to deal with non-existent or unrepresented information in the data. Chatbot hallucinations can have significant impacts in several areas, such as education, customer service and content creation. For example, a chatbot that cannot understand the nuances of human language may provide inappropriate or incorrect responses, which can lead to a negative user experience. Additionally, chatbots can perpetuate social bias and discrimination if they are trained on biased data. To deal with these limitations in the educational context, it was proposed the development of a chatbot for teaching programming in Python that uses advanced natural language processing techniques to reduce hallucinations when learning the language. 2024-05-31T00:00:00Z https://repositorio.uema.br/jspui/handle/123456789/6279 Título: Estimativas de arrecadação do ICMS do Estado do Maranhão usando algoritmos de machine learning Abstact: Tax collection forecasting is a cornerstone of fiscal planning and efficient public management. The Tax on Circulation of Goods and Services (ICMS) constitutes the main source of revenue for Brazilian states, and its accurate projection is crucial for allocating resources to strategic areas. However, the complexity of its dynamics, influenced by non-linear macroeconomic variables, and the lack of studies applied to the reality of the state of Maranhão pose a challenge for public administrators. This work aims to address this gap by investigating how machine learning techniques can improve the accuracy of forecasting monthly ICMS revenue in Maranhão. The overall objective is to develop and validate advanced computational models using a historical series of economic and social data from January 1997 to April 2024. This quantitative and applied research adopted the CRISP-DM framework. Data were collected from public sources such as SEFAZ-MA, IBGE, and the Central Bank. Initially, nineteen independent variables were considered, and a Multiple Linear Regression model was used to select the most relevant ones, such as GDP, diesel consumption, and electricity consumption indicators. Four machine learning algorithms were implemented, compared, and validated: Random Forest, Decision Tree, Linear Regression, and XGBoost. Performance evaluation was performed using the RMSE, MAE, MAPE, SMAPE, and R² metrics, using the k-fold cross-validation technique (with k=10) and a data split of 80% for training and 20% for testing. This study contributes a practical and validated model that can be integrated into the state's budget planning process, promoting more transparent, efficient, and data-driven fiscal management 2025-01-01T00:00:00Z https://repositorio.uema.br/jspui/handle/123456789/6273 Título: Método computacional para auxiliar o diagnóstico precoce da Granulomatose de Wegener Abstact: This paper presents a proteomic pattern recognition system aimed at assisting in the early diagnosis of Wegener's Granulomatosis (WG), a rare idiopathic vasculitis that is difficult to detect and carries a high mortality rate for untreated individuals. The proposed method involves extracting features from proteomic signals and classifying them as belonging to individuals with or without WG. To achieve this, Independent Component Analysis is used for feature extraction, the Minimum Redundancy Maximum Relevance algorithm is employed to reduce the number of features and computational costs, and a Support Vector Machine is used for classification. The method's performance was evaluated using a dataset of 335 proteomic signals, comprising 75 active cases, 101 negative cases, and 159 cases in remission. The best result was obtained using a twenty-feature vector, yielding accuracy, specificity, and sensitivity of 98.24%, 99.73%, and 99.50%, respectively. These results demonstrate that the proposed system is efficient for diagnosing WG and outperforms the current methodology, which is based on clinical, serological, and radiological examinations proposed by the American College of Rheumatology. 2016-07-22T00:00:00Z https://repositorio.uema.br/jspui/handle/123456789/6247 Título: Sistemas de gerenciamento e monitoramento de rack’s outdoors de telecomunicações, baseado em internet of things Abstact: The present work aims to correlate concepts, technologies and application of a solution based on embedded systems, Internet Of Things and 5G, to monitor and ensure data persistence in telecommunications companies. First, a survey of the current scenario of the Telecommunications industry is carried out. In addition, they present intrinsic problems, in which consumers are the most affected, such as the lack of reliability and interactivity of the product offered by energy concessionaires. The bibliographic research was through a literary review based on articles, books and works by several authors from the period 2019 to 2022. The collection of information on the topic took place through the databases: Google Scholar and IEEE. In this way, conceptual information was presented on technologies such as ESP32, 5G, sensing and especially the MQTT protocol, for long-distance solutions, which belong to the concept of Internet of Things (IoT). Therefore, we propose the development of a device capable of monitoring and assisting telecommunications companies in the management to maintain their external assets. 2022-10-31T00:00:00Z