Navegando por Autor "Nascimento, Eliaquim Moreira do"

Filtrar resultados informando o último nome do autor
Agora exibindo 1 - 1 de 1
  • Imagem de Miniatura
    Item
    Análise comparativa de métodos de aprendizado supervisionado para mineração de opinião dos usuários da plataforma de e-participação votenaweb
    (2019) Nascimento, Eliaquim Moreira do; Souza, Ellen Polliana Ramos; Santos, Diego George da Silva; http://lattes.cnpq.br/1433346297212674; http://lattes.cnpq.br/6593918610781356; http://lattes.cnpq.br/6033431769198675
    With the help of information technologies, it is possible that the media and participatory media are recognized as democratic and participatory, as well as the participation and collaboration platforms, also known as e-participation and e-collaboration. However, although it is a problem, it is necessary to have a broader view on the perspectives, it is a process that is not analyzed and does not have a process of presentation or bill. It is impossible to be human completely in all content in a reasonable amount of time What has sparked an interest in the community of systems able to obtain additional information about the data type automatically. Opinion mining, also known as sentiment analysis, is an area of study that analyzes people’s feelings and opinions about entities, such as products and services, unstructured formal expressions, such as text, for example. This study, this investigate the case of the class comparator and technique to pre-processing the reviews of users of topics-and-collaboration and e-collations available for the Brazilian citizens. As this article was published, it was published in the magazine Aplications of the portal VOTENAWEB, in view of what is widely used, in addition to allowing publications to comment on a set of projects. Supervised learning algorithms with data preprocessing techniques were evaluated: storpwords removal, keywords, N-gram, TF-IDF and word embedding, a set of instructions for keyword mining. . The linear regression algorithm was the best result with accuracy of 88.22% and f-measure of 87.07%, while the deep learning of the old meaning was accuracy of 84.96% and f-measure of 84.90%.