SENTIMENT ANALYSIS OF NEWS TWEET MESSAGES

Abstract

The curiosity of knowing what people think and how they feel about daily events has always existed. With the advent of Web 2.0 and the wide dissemination of people’s opinions through the World Wide Web, this interest has become even greater, leaving the personal level and reaching companies’ marketing activities. In this study, we aim to please this curiosity by examining people’s reaction to news published in the media. To achieve this goal, we developed a tool capable of determining the polarity of texts collected via the microblogging service Twitter and analyzing these opinions, with respect to if people tend to classify news related to three previously selected topics as positive or negative. For this, we used different classifiers based on language models. By the end of the experiment, it was also possible to evaluate the performance of these classifiers when working with tweets written in Brazilian Portuguese.