Application of text mining to analysis of social groups in blogosphere

Abstract

The paper concerns analysis of social groups in blogosphere using text mining methods to discover additional knowledge about groups and users. Two methods to distinguish messages (the first one - between messages from main and secondary thread, the second one - between facts and opinions) in blogosphere were proposed and their quality was assessed on manually annotated dataset. Both tasks are very important and proposed methods deal with them in a fully automatic way. The results were obtained on real-world data from Polish blogosphere. © 2014 Springer International Publishing Switzerland.