Análise de sentimentos de tweets em português-brasileiro utilizando Inteligência Artificial

Abstract

With the rapid growth of social media worldwide, the vast amount of information circulating there has become a valuable database for research and behavioral analysis. In recent years, Twitter has emerged as one of the largest sources of opinion on the internet, with an average of 500 million tweets created daily, each with up to 280 characters. Sentiment analysis refers to the study of emotions, opinions, and feelings expressed in a text, carried out through the analysis of the polarities present in the sentences. Given this context, the aim of this work is to develop and implement a web application for data annotation via manual classification, which will serve as a basis for training a model generated by machine learning algorithms. This model uses feature extraction techniques and Bayes' Theorem for the automatic classification of texts in Brazilian Portuguese. In the end, part of the dataset created using the web application is used to train the model, while the remaining part is used for its evaluation. Subsequently, accuracy is analyzed using a table that shows the classification frequencies for each class of the model, known as a confusion matrix.