Darlinton B. F. Carvalho,  Ricardo M. Marcacini,  Carlos J. P. de Lucena,  Solange O. Rezende. 

There are plenty of public content available on the Internet, especiallyin online communities, enabling researchers to study society in new ways.Since the qualitative content analysis of online forums is very time consuming,the following problem arises: how to select the content to be analyzed? Thispaper introduces a new process to support solving this problem. This process isbased on unsupervised machine learning techniques and provides consolidatedand structured results. This includes measurements and a content explorationmethod. A tool that helps to apply the proposed process was created and ispresented as well.

http://www.lbd.dcc.ufmg.br/colecoes/brasnam/2012/0028.pdf

Processando consulta.