language model 0.00271835
language word 0.00250381
translation model 0.002326077
model data 0.0021672510000000002
target word 0.002025315
word pairs 0.001997476
linear model 0.0019565750000000003
statistical model 0.001899775
target language 0.001876365
pair features 0.001811156
lexical features 0.001803878
channel model 0.001802907
word pair 0.001789766
guage model 0.001781477
labeling model 0.0017664240000000002
quence model 0.001766285
language models 0.0017452739999999998
english language 0.001696598
target words 0.0016945100000000002
word candidates 0.001684444
several word 0.001675883
arbitrary features 0.0016301640000000001
similarity features 0.001626836
oov word 0.001610726
standard language 0.00161002
binary features 0.001608826
features lexnorm 0.0016080030000000002
feature function 0.00160762
candidate word 0.001606831
word sequences 0.001605964
related word 0.001605743
active features 0.0015742640000000001
language modeling 0.00156942
word formation 0.0015615400000000002
sider word 0.001550566
word something 0.00154962
word suttin 0.00154962
model 0.00154092
statistical language 0.0015362849999999999
feature weights 0.001503926
bigram language 0.001469731
training approach 0.0014480510000000001
online language 0.001441373
media language 0.00144041
standard words 0.0014281650000000001
language variation 0.001428069
training algorithm 0.001415978
internet language 0.001413091
language tweets 0.0014064
dia language 0.0014005509999999999
language risk 0.0014005509999999999
language varieties 0.0014005509999999999
language changes 0.0014005509999999999
training data 0.001395914
features 0.00134777
similar words 0.001332841
training method 0.0012891460000000001
oov words 0.001279921
candidate words 0.001276026
nonstandard words 0.001257552
target sentence 0.0012573010000000002
feature types 0.001254051
such information 0.001243422
feature functions 0.0012397279999999998
feature counts 0.0012224599999999999
source sentence 0.001215797
feature name 0.001209477
likelihood function 0.001207517
feature expectations 0.001199116
corresponding feature 0.001195325
necessary feature 0.001189009
feature expecta 0.0011847609999999999
computing feature 0.0011847609999999999
language 0.00117743
different approach 0.001152287
machine translation 0.001151357
other values 0.001146158
new training 0.001121236
such tokens 0.001109507
text normalization 0.0011050970000000002
training con 0.001089096
carlo training 0.001087881
possible target 0.001086638
target sequence 0.00106175
ing pairs 0.001059084
trigram target 0.001058295
new target 0.001050588
statistical approach 0.001037323
target tokens 0.001037205
such trigrams 0.001032836
available training 0.001032006
target domain 0.001031713
standard normalization 0.001030406
such resources 0.001028874
source sequence 0.001020246
normalization system 0.001016179
ing data 0.001014319
proposal distribution 0.001014285
unlabeled training 0.001007613
training tweets 9.985530000000001E-4
