A Comparative Analysis of Strategies for Semantic Short-Text Categorization
Veronica Perez Rosas, Marcelo Luis Errecalde, Paolo Rosso · 2010
Nowadays, short-texts categorization is an important research area be- cause most of the information we usually receive and work with have this characte- ristic (e-mails, text messages, news, etc.). Difierent studies have reported interesting results in text categorization by adding semantic information to documents' repre- sentation. However, these studies have not focused on the particularities that short texts introduce. Furthermore, the most popular disambiguation methods (corpus- based methods) not always are feasible to use in these domains. Thus, knowledge- based disambiguation methods become interesting alternatives in these cases. In this article, we study the efiectiveness of short text categorization, when semantic infor- mation, obtained by knowledge-based methods, is used. The results obtained with this approach show interesting improvements that encourage to continue this line of research.