word model 0.006374909999999999
word error 0.005409196
word errors 0.0053519449999999994
word segmentation 0.0053131689999999995
unknown word 0.005178518
word boundary 0.0051202669999999995
word sequence 0.005095158999999999
candidate word 0.005032437999999999
target word 0.004994424999999999
correct word 0.004956069
good word 0.004948444
specific word 0.004944294
original word 0.004911899999999999
word sequences 0.004898998
explicit word 0.004898369
probable word 0.004869113
particular word 0.00486845
real word 0.004864467999999999
appropriate word 0.004864004999999999
word collocation 0.004861124
centroid word 0.004859522
corrected word 0.004858487
word lattice 0.004857451
troid word 0.004857451
plicit word 0.004857451
wilunknown word 0.004857451
huge word 0.004857451
word delimiter 0.004857451
finding word 0.004857451
error words 0.002570586
correction words 0.002413218
character segmentation 0.002401929
unknown words 0.002339908
first character 0.002274002
possible words 0.002202528
candidate words 0.002193828
context words 0.002186946
trigram model 0.0021391220000000002
above model 0.002089514
ror words 0.00206628
known words 0.002027675
contiguous words 0.002021619
gram model 0.001998013
guage model 0.001992702
model losses 0.001983202
optical character 0.001948817
character ecognition 0.001946891
gle character 0.001946891
words 0.00178597
model 0.00175033
character 0.00171334
segmentation algorithm 0.001486675
error correction 0.001411864
other characters 0.001403975
error string 0.001391232
spelling error 0.001368139
new errors 0.0012331339999999999
spelling correction 0.0012107709999999998
large number 0.0012032509999999998
unknown features 0.001187286
other approaches 0.0011775169999999999
error trigram 0.001173408
other languages 0.001168811
ocr error 0.0011663860000000002
unknown string 0.001160554
total error 0.0011476960000000001
error total 0.0011476960000000001
different features 0.0011416450000000002
ocr errors 0.001109135
input sentence 0.001107633
several characters 0.001101039
whole corpus 0.001099989
edit distance 0.00109608
prepared corpus 0.001081795
recognition algorithm 0.0010714370000000002
approximate error 0.0010638520000000001
error strings 0.001058056
level characters 0.001055808
winnow algorithm 0.001054733
global algorithm 0.001049831
test set 0.0010487679999999998
other substrings 0.001037443
other characteristic 0.001036487
other sources 0.001036282
algorithm updates 0.001035735
candidate correction 0.001035106
correction candidate 0.001035106
error introduced 0.0010185860000000001
introduced error 0.0010185860000000001
low probability 0.001012984
substitution errors 9.98429E-4
deletion errors 9.764039999999999E-4
space characters 9.63924E-4
first task 9.620639999999999E-4
nonword errors 9.61915E-4
sertion errors 9.61915E-4
spelling rules 9.54777E-4
entire sentence 9.42944E-4
put sentence 9.361899999999999E-4
correction candidates 9.29095E-4
