Análise de Sentimentos de Textos do Twitter sobre Mercado de Ações Brasileiro

Abstract

Used to determine the attitude, opinion, emotion expressed by someone on a particular subject, the sentiment analysis has provided several studies on the importance of mining opinion with the price in the stock market. Research shows that there is a strong relationship between the news about a company and its fluctuations in stock prices. The objective of this research is to develop a model that identifies and classifies the polarity in the analysis of feelings using LSTM neural networks, in which they were extracted from specialized profiles in the Brazilian stock market of the social network Twitter, for companies previously defined: Petrobras, Vale, JBS, Eletrobras and Oi. The results obtained allowed to classify positivity, neutrality and negativity of the tweets, as well as perform an analysis and visualize the frequency in which it occurs.