Modelo assistente para classificação de dados provenientes de redes sociais: um estudo de caso com dados do twitter

Abstract

Since its inception, virtual social networks like Twitter have reached exorbitant amount of users worldwide, making it an immeasurable potential environment for social research, economic, cultural and etc. Increasingly researchers have turned their attention to the great mass of data generated daily in this environment. However, handling large amounts of data is a costly task when performed manually. The objective of this research is to propose a set of tools and methodology that it can reduce the human effort spent in the organization of large masses of data from social networks. To achieve this goal, we propose an iterative work model that makes the most of existing knowledge in a small amount of data manually analyzed by experts. The working model combines information retrieval techniques such as classification and clustering algorithms in order to make the result of the most similar process to what the expert would get if carried out completely manually. The proposed model was put to the test with use of two sets of extracted data from Twitter and manually classified before this research. The results were promising.