training data 0.00377183
language model 0.003292242
model accuracy 0.003143712
training corpus 0.002837433
accuracy training 0.002792532
markov model 0.002630429
bigram model 0.002603065
wsj training 0.0025885079999999998
based model 0.002577387
model mitchell 0.002568283
wsj data 0.002462898
supervised training 0.002439161
treebank training 0.002389645
training approach 0.002377655
training corpora 0.002370623
additional training 0.002359434
various training 0.002354041
training sets 0.002352884
unlabeled training 0.002346835
model 0.0022999
data size 0.002286172
useful training 0.002249535
unlabeled data 0.002221225
training conditions 0.002220769
unlimited training 0.002216307
testing data 0.0021498109999999997
scale data 0.002149391
nyt data 0.00213875
data dataset 0.00213775
data ngr 0.002121917
data yields 0.002102151
sufficient data 0.002101568
data sparsity 0.002101326
sparse data 0.002094842
language models 0.001958412
training 0.00194872
other models 0.001870915
word preferences 0.001839205
test corpus 0.001806062
other systems 0.00156272
test set 0.0015304390000000002
such models 0.001503545
wsj accuracy 0.0014835999999999998
different domains 0.0014328009999999998
large accuracy 0.001414713
accuracy improvements 0.00138184
various models 0.001371391
parser oov 0.001342472
language generation 0.001332055
domain performance 0.001328065
brown test 0.001326338
msa models 0.001324444
different kinds 0.001313936
other domains 0.001313746
language processing 0.001310583
brown corpus 0.001297702
gram models 0.001285039
natural language 0.001280329
other modifiers 0.001268113
ken language 0.001260802
accuracy brown 0.001252801
brown accuracy 0.001252801
annotated test 0.001237982
corpus likelihood 0.001228273
berkeley parser 0.001227752
other contexts 0.001218051
corpus sections 0.001211114
accuracy improvement 0.001207364
other kinds 0.001194881
switchboard test 0.00118679
string accuracy 0.001169221
corpus yields 0.001167754
swbd corpus 0.0011636069999999999
gigaword corpus 0.0011614099999999999
corpus por 0.0011575259999999999
wsj evaluation 0.00114307
prediction accuracy 0.001126947
emission probabilities 0.001126253
swbd accuracy 0.001118706
accuracy swbd 0.001118706
evaluation set 0.001116372
accuracy frameworks 0.001113498
pairwise accuracy 0.001113498
large improvements 0.0011089289999999998
transition probabilities 0.001088833
wsj treebank 0.001080713
automatic parsers 0.001046099
state transition 0.001044836
absolute performance 0.001024453
performance increase 0.001023943
learning approaches 0.001019713
msa systems 0.001016249
previous systems 0.001010288
prior systems 0.0010027690000000001
recent systems 0.001001806
msa wsj 9.98162E-4
novel systems 9.97442E-4
language 9.92342E-4
ing modifier 9.869380000000001E-4
superior performance 9.80175E-4
