same topic 0.003957448
many topic 0.0036834719999999997
potential topic 0.0035860069999999996
topic boundaries 0.0035746949999999997
topic boundary 0.0035529959999999997
topic length 0.003548062
topic detection 0.003521728
topic segmentation 0.003519142
topic segment 0.00351837
primary topic 0.0035144649999999996
topic shifts 0.0034866489999999997
likely topic 0.0034588609999999997
putative topic 0.003454754
topic shift 0.0034520899999999997
topic segmentations 0.003451951
legitimate topic 0.0034484249999999998
topic 0.00324541
word sense 0.002876725
content word 0.002724353
word frequency 0.002605802
word plant 0.00257658
use word 0.002518541
particular word 0.002518051
word types 0.002487785
segments word 0.002469776
word sequences 0.002468423
word sequence 0.002459445
word bigrams 0.0024382
approximate word 0.002431459
word bigram 0.002423227
word repetition 0.002422114
word header 0.002422114
new topics 0.001919853
good model 0.00188892
first model 0.00187777
multiple topics 0.0017977179999999998
frequency model 0.001747602
first words 0.0017266199999999999
parameter model 0.0017164290000000002
content words 0.001715003
model average 0.001683709
space model 0.001682801
disjoint topics 0.001678085
statistical model 0.0016585520000000002
same data 0.001655064
entropy model 0.001602054
model max 0.001583692
channel model 0.001576775
model our 0.0015746500000000001
topical words 0.001569407
training data 0.001569276
cue words 0.0014957220000000001
topics 0.00147377
weighted words 0.001429523
unknown words 0.0014191450000000001
surrounding words 0.0014140350000000001
news data 0.001395033
test data 0.0013627700000000001
model 0.00136044
different corpus 0.001313805
broadcast data 0.001216279
words 0.00120929
the data 0.001187084
sufficient data 0.001184486
language models 0.001180916
newswire data 0.001159743
training corpus 0.001146811
same parameter 0.001068027
language modeling 0.001065365
such boundaries 0.001064383
such phrases 0.001061202
different subjects 0.0010302430000000001
common sense 0.001013334
different stories 0.001011558
different hints 9.98095E-4
document length 9.95655E-4
relevant document 9.81844E-4
other section 9.782139999999998E-4
other cohesion 9.770999999999998E-4
document retrieval 9.74646E-4
news corpus 9.72568E-4
content analysis 9.5789E-4
news documents 9.567409999999999E-4
spoken document 9.46358E-4
many documents 9.42796E-4
other sources 9.40366E-4
such signoffs 9.40059E-4
same entities 9.360740000000001E-4
long document 9.33055E-4
same story 9.26577E-4
frequency algorithm 9.139E-4
sense disambiguation 9.11637E-4
other hand 9.11298E-4
second algorithm 9.03523E-4
information retrieval 9.013879999999999E-4
method documents 8.99346E-4
new news 8.9809E-4
local content 8.923329999999999E-4
little training 8.865609999999999E-4
sense decline 8.837679999999999E-4
