Mineração de texto aplicada à identificação de sentimentos e intenções

Abstract

The human being is surrounded by data and information technology has revolutionized its access and efficient processing. What was once presente in printed materials like books, newspapers, magazines and so on, is now available, at the front of the hand. Sometimes, this information can, for example, be extracted from music, lyrics and songs. Music can be presented in written, spoken or audio-visual form and has a long relationship with the human society, being one of the oldest artistic and cultural manifestations. The benefit of Text Mining is given by the large amount of information contained in the texts. The application of this work is to apply to a set of musical lyrics a series of text mining techniques, in order to obtain patterns of interest regarding word frequency, sentimento anaysis, relationship between words and the most prevalente analysis. This work is classified as a case study, descriptive, exploratory and experimental one. As a result of applying knowledge discovery concepts and text mining techniques, it was possible to identify the content and sentiment involved in the dataset used in order to observe negative and egocentric feelings, such as: I don't want, I won't, I don't know, won't, for me, among others.