several data 0.0015357020000000001
language model 0.001448455
different methods 0.001361585
many features 0.001351395
various feature 0.001339784
training set 0.001334816
development data 0.001298429
data sets 0.0012967830000000001
set results 0.001291295
feature selection 0.001276663
population data 0.001268189
method cell 0.001236991
test set 0.001235631
different cells 0.001231176
set document 0.0012310340000000002
model word 0.001228532
good features 0.001209938
high training 0.0012055009999999999
large training 0.00120187
test document 0.0011841450000000002
characteristic features 0.0011681629999999998
different parameter 0.0011625799999999999
feature independence 0.001160865
same cell 0.001151993
hierarchical model 0.001144753
igr feature 0.001126502
cell accuracy 0.001121169
cell size 0.001110942
several methods 0.001100019
feature hash 0.001089038
same corpus 0.001083768
document geolocation 0.001072251
new documents 0.001056846
training process 0.001053214
language modeling 0.0010496589999999999
different fea 0.001049038
various methods 0.00104149
such tweet 0.001038021
test time 0.001021856
geolocation method 0.001021252
ilar language 0.001003843
hierarchical grid 9.99691E-4
wikipedia corpus 9.97378E-4
regression model 9.96584E-4
different scales 9.94807E-4
interpolation model 9.94237E-4
first level 9.92479E-4
node level 9.90656E-4
different meaning 9.89483E-4
such grids 9.85902E-4
selection methods 9.78369E-4
leaf cell 9.73256E-4
training sets 9.728989999999999E-4
twitter users 9.68001E-4
full set 9.63691E-4
cell choice 9.59665E-4
other metrics 9.54821E-4
large dataset 9.537920000000001E-4
cell sizes 9.52371E-4
local classifier 9.52254E-4
user geolocation 9.50198E-4
root cell 9.36477E-4
infrequent words 9.339159999999999E-4
other types 9.32425E-4
cal words 9.291389999999999E-4
salient words 9.27365E-4
high number 9.264729999999999E-4
cell intonxn 9.24824E-4
hierarchical classification 9.23887E-4
hierarchical approach 9.22543E-4
grid size 9.219019999999999E-4
large size 9.2104E-4
such collections 9.20199E-4
such subtree 9.20067E-4
unseen words 9.195E-4
next level 9.18237E-4
training docu 9.15401E-4
training cost 9.13698E-4
first method 9.11522E-4
cell accu 9.090789999999999E-4
poor results 9.07361E-4
user frequency 9.070090000000001E-4
cell croot 9.06947E-4
corpus differences 9.024230000000001E-4
method parameters 9.02095E-4
single user 8.97087E-4
geographic document 8.96051E-4
features 8.93988E-4
dev set 8.92057E-4
tree grid 8.91548E-4
information gain 8.880870000000001E-4
wikipedia table 8.808259999999999E-4
feature 8.80051E-4
certain documents 8.79078E-4
test sets 8.73714E-4
classification approach 8.654E-4
model evaluations 8.63801E-4
ous methods 8.63761E-4
direct method 8.63507E-4
level hierlr 8.63501E-4
