Cassiana Fagundes da Silva,  Renata Vieira. 

This paper compares Decision Trees (DT) and Support Vector Machines (SVM) for text categorization tasks using linguistic information. We show that linguistic knowledge is useful for term selection for both learning techniques. We also show that DTs perform better than SVM when a reduced number of terms is considered, and they stop improving at a certain point whle SVM continually improves the results when the number of terms increase.

http://www.lbd.dcc.ufmg.br/colecoes/til/2007/0011.pdf

Processando consulta.