Processamento de linguagem natural para detecção de padrões no mercado de ações brasileiro

Abstract

Through social media, influencers are able to reach thousands of people. In the financial market it is no different, tweets from such people can come to influence the shares’s value. In this work, the use of Natural Language Processing (NLP) and the Apriori algorithm is proposed to identify associations between the behavior of stock prices and the sentiment expressed on Twitter. The main approach is to analyze structured and unstructured data, using NLP concepts and association rules to identify the main players capable of influen- cing the masses and having some impact on stock prices, or even the anomalous social movements capable of causing some impact. Associations were found only in short peri- ods, where in all stocks some periods were identified in which the predominant sentiment is expressed after the change in price, not exceeding a period of 15 days.