sentence compression 0.00468329
original sentence 0.002986677
sentence compres 0.002981287
sentence level 0.002944475
ture sentence 0.0028367329999999997
sentence candidates 0.00283467
different compression 0.0027465169999999995
first compression 0.00262416
sentence 0.00258525
compression step 0.0025583729999999996
tence compression 0.0024189009999999998
compression meth 0.002344911
compression ratio 0.002343834
compression 0.00209804
summary sentences 0.0018214989999999999
word level 0.0015486649999999999
important sentences 0.001472218
human compressions 0.001437993
fields model 0.0014260730000000001
same corpus 0.001414121
same data 0.00133216
global information 0.001297939
unnecessary words 0.001263978
incomplete words 0.001260933
system performance 0.0012515529999999999
new features 0.001249751
data set 0.001215905
portant information 0.001208359
model 0.00114732
meeting corpus 0.001129608
human references 0.001115689
human annota 0.001106779
summarization task 0.001070124
sentences 0.00106492
feature set 0.001046479
words 0.00102027
tree approaches 0.0010132309999999999
system output 0.001011912
first step 9.86453E-4
information 9.46423E-4
labeling method 9.267979999999999E-4
text domain 9.255839999999999E-4
previous step 9.23813E-4
meeting summarization 9.23643E-4
labeling problem 9.170300000000001E-4
previous work 9.07974E-4
speech utterances 9.040739999999999E-4
programming approach 9.030099999999999E-4
evaluation metrics 8.91663E-4
speech domain 8.852039999999999E-4
training approach 8.80299E-4
based method 8.79314E-4
speech domains 8.731819999999999E-4
decision tree 8.65943E-4
formal text 8.60777E-4
spontaneous speech 8.32993E-4
corpus 8.28806E-4
taneous speech 8.232049999999999E-4
broadcast news 8.190770000000001E-4
second step 8.13376E-4
features 8.12157E-4
linear programming 8.05284E-4
bleu scores 7.82856E-4
sequence labeling 7.704680000000001E-4
grammar rules 7.626359999999999E-4
summary 7.56579E-4
level labeling 7.36233E-4
discriminative training 7.34521E-4
ond step 7.110689999999999E-4
discriminative reranking 6.73146E-4
reference com 6.68869E-4
random fields 6.3923E-4
rates grammar 6.37172E-4
performance 6.29922E-4
summarization 6.22841E-4
multiple references 6.2248E-4
system 6.21631E-4
quence labeling 6.182550000000001E-4
abstractive summa 6.12885E-4
much redundancy 6.066019999999999E-4
global lev 5.96479E-4
evaluation 5.91016E-4
speech 5.81009E-4
annotation procedure 5.80353E-4
feature 5.77419E-4
reranking module 5.75994E-4
compressions 5.73974E-4
icsi meeting 5.56556E-4
method 5.4979E-4
approach 5.47612E-4
only things 5.47395E-4
problem 5.40022E-4
scores 5.33851E-4
additional gain 5.283810000000001E-4
source 5.24834E-4
example 4.54181E-4
approaches 4.41448E-4
study 4.23694E-4
methods 4.00175E-4
compres 3.96037E-4
