Detalhes bibliográficos
| Ano de defesa: |
2020 |
| Autor(a) principal: |
Brandão, Jhonathan de Godoi
 |
| Orientador(a): |
Calixto, Wesley Pacheco
|
| Banca de defesa: |
Calixto, Wesley Pacheco,
Ribeiro, Luiz Eduardo Bento,
Reis, Márcio Rodrigues da Cunha,
Cruz Junior, Gelson da,
Marques, Leonardo Garcia |
| Tipo de documento: |
Dissertação
|
| Tipo de acesso: |
Acesso aberto |
| Idioma: |
por |
| Instituição de defesa: |
Universidade Federal de Goiás
|
| Programa de Pós-Graduação: |
Programa de Pós-graduação em Engenharia Elétrica e da Computação (EMC)
|
| Departamento: |
Escola de Engenharia Elétrica, Mecânica e de Computação - EMC (RG)
|
| País: |
Brasil
|
| Palavras-chave em Português: |
|
| Palavras-chave em Inglês: |
|
| Área do conhecimento CNPq: |
|
| Link de acesso: |
http://repositorio.bc.ufg.br/tede/handle/tede/11185
|
Resumo: |
The aim of this work is to develop a tool for that optimizes supervised machine learning models in order to classify polarity of opinions in tweets. Five different datasets are used, which are prepared, preprocessed and then used as input for the training and evaluation stage of machine learning models. The best accuracy results obtained in the training and evaluation of the models are 82.45% for the data without preprocessing × 78.83% with all the proposed preprocessing for the dataset using the Naive Bayes classifier. Finally, hyperparametric optimization of the classifiers and selection of the model that obtains the best accuracy is performed. The optimized model achieves an accuracy greater than 90% for some data sets. The supervised learning techniques depend on labeled data for training, the proposed method produces similar performances for datasets of varying sizes, which allows the development of optimized classification models with reduced amount of labeled data. |
