Classificação de sentimentos usando máquina de vetor de suporte (SVM)

Abstract

Sentiment analysis is an expanding field within Natural Language Processing (NLP), enabling the understanding of emotions and opinions expressed in text. This work aims to develop a sentiment classification model capable of automatically identifying positive, negative, or neutral tones in short texts. To achieve this, the Support Vector Machine (SVM) technique is employed together with TF-IDF vectorization, applied to a public dataset of Twitter messages. The model was trained, compared with other classification models such as Naive Bayes and Logistic Regression, and rigorously evaluated using performance metrics such as accuracy, precision, recall, and F1-score in order to validate its effectiveness. The results obtained demonstrate that the method shows excellent performance in analyzing short texts, achieving an accuracy of 92%, highlighting the potential of SVM as a supporting tool for opinion mining and social media monitoring applications.