Um framework para mineração de textos de redes sociais

Abstract

Big advances in storage technology and the popularization of the internet have enabled organizations to generate and store big amounts of data. The data produced by social networks grows each day. Its users share information, feelings and opinions about a wide range of matters that happen over the world. This increasing production of data gives the opportunity and necessity of performing analyses to extract useful knowledge from it. Therefore, many researches were already performed for the purposes of extracting knowledge from social network data. Most of these researches have shared a really similar methodology. The stages of data collect, preprocessing, analyses and visualization are common stages in text mining and social network researches. However, these researches need to make their own implementations to perform each one of its stages. Consequently, these researches need to spend a lot of time in the intermediate stages of the process, leaving less time for the real data analyses. Motivated by this problem, this research had as the objective the creation of a framework capable of helping and accelerating studies with emphasis on text mining in social network. The framework was created including the stages of data collect, preprocessing, analyses and visualization through a web system. For this research it was also performed a case study with emphasis on data collected from twitter related to the impeachment of the former president of Brazil, Dilma Rousseff. This research was performed with the intention to validate the created framework and analyze the repercussion of the impeachment in the social network Twitter.