sequence features 0.002097676
quote features 0.001998662
similar features 0.001896672
different speaker 0.001886557
binary features 0.0018683150000000002
distance features 0.00185807
individual features 0.0018520770000000002
paragraph features 0.00184637
weighted features 0.0018177850000000001
tic features 0.0017981
realistic features 0.001784212
nearby features 0.001767713
quence features 0.001763952
class model 0.0016593789999999999
features 0.00155283
correct speaker 0.001466791
binary model 0.0014414150000000001
way model 0.0014310590000000002
chain model 0.001427503
corpus sequence 0.0014085999999999999
speaker label 0.0014065640000000001
viterbi model 0.001391043
decoding model 0.001391013
markov model 0.001388346
candidate speaker 0.001387196
speaker prediction 0.001384601
crf model 0.001378107
speaker attributions 0.0013723070000000001
attribution corpus 0.0013716190000000001
speaker candidates 0.001358674
other quotes 0.001349399
pdtb speaker 0.001344002
first corpus 0.001342901
target speaker 0.001341369
gle speaker 0.0013397930000000001
appropriate speaker 0.001339354
rect speaker 0.001337431
corpus gold 0.001233207
different speakers 0.001233029
previous quotes 0.0012190629999999998
other words 0.001176898
final corpus 0.001163777
accuracy results 0.001160301
literature corpus 0.001157466
only corpus 0.001154847
other speakers 0.001152748
next corpus 0.001141572
third corpus 0.001140585
model 0.00112593
speaker 0.00112023
corpus comparisons 0.0011182499999999999
ment corpus 0.001117836
standard information 0.0011168390000000001
lit corpus 0.001114013
wsj corpus 0.001113895
class models 0.001109469
smh corpus 0.001102517
herald corpus 0.001095988
corpus extension 0.001089647
whole corpus 0.001081317
other approaches 0.001073358
standard sequence 0.0010717959999999999
previous attribution 0.001063575
final results 0.001054032
feature 0.00104419
same text 0.0010293799999999999
newswire results 0.001024667
many quotes 0.001023635
previous speakers 0.0010224119999999999
other corpora 0.0010127550000000002
crf results 0.001006186
full approach 0.001002714
attribution system 9.9847E-4
different reconciliation 9.926450000000001E-4
classification task 9.918560000000002E-4
language quote 9.9175E-4
speech attribution 9.80848E-4
different char 9.7701E-4
different voices 9.75736E-4
direct quotes 9.72452E-4
core results 9.688720000000001E-4
bold results 9.66878E-4
italicised results 9.66878E-4
other parts 9.626140000000001E-4
quote attribution 9.536970000000001E-4
other candidate 9.53012E-4
previous work 9.364929999999999E-4
mixed quotes 9.30726E-4
word distance 9.25377E-4
other candidates 9.2449E-4
other trigram 9.095990000000001E-4
prior quotes 9.07093E-4
indirect quotes 9.02842E-4
additional information 9.02556E-4
test set 9.00297E-4
training set 8.96834E-4
gold standard 8.96403E-4
tional quotes 8.96083E-4
other anaphors 8.93787E-4
tract quotes 8.87418E-4
