queri drelev document represent text cluster filter dbase data partit train multipl classifi system automat extract domain dspecif stopword label document learn term depend link inform theoret inclus measur learn social network multipl resampl method combin featur rank text classif automat taxonomi extract googl term depend learn social network web document support vector classifi text classif small number featur data partit evalu measur classifi ensembl inform dtheoret measur evalu data partit multipl classifi gener fuzzi membership function inform theori measur genet algorithm 