Análise de métodos e ferramentas para reconhecimento de palavras relevantes em microblogs

Abstract

Extracting accurate information from the huge volumes of data, much of them unstructured, generated in social media is currently a big challenge. However, it has several relevant applications, some of them latent yet. One of the first and most decisive steps in this information extraction process is the recognition of relevant words in texts. This article presents a comparative study of methods and tools for recognizing relevant words on microblog posts. Among several analyzed tools, five have been selected for experments with 100,000 tweets. These experiments showed high variability of the results generated by different tools, suggesting a need for improvements.