I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
p
ute
r
E
ng
in
ee
ring
(
I
J
E
CE
)
Vo
l.
10
,
No
.
3
,
J
u
n
e
2
0
2
0
,
p
p
.
2
3
6
7
~
2
3
7
4
I
SS
N:
2
0
8
8
-
8708
,
DOI
:
1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v
1
0
i
3
.
p
p
2
3
6
7
-
2
3
7
4
2367
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m/in
d
ex
.
p
h
p
/I
JE
C
E
Co
m
bin
ed cos
ine
-
linea
r re
g
ress
io
n
m
o
del si
m
i
la
rity
w
ith
a
pplica
tion to ha
ndw
ritt
en w
o
rd spo
tt
ing
Yo
us
s
ef
E
lf
a
k
ir
,
G
hizla
ne
K
ha
is
s
i
di
,
M
o
s
t
a
f
a
M
ra
bti
,
Dr
is
s
Che
no
un
i,
M
a
na
l B
o
ua
l
a
m
L
a
b
o
ra
to
ry
o
f
Co
m
p
u
ti
n
g
a
n
d
I
n
t
e
rd
isc
ip
li
n
a
ry
P
h
y
sic
s,
ENS
,
S
id
i
M
o
h
a
m
e
d
Be
n
A
b
d
e
ll
a
h
Un
iv
e
rsi
ty
,
M
o
ro
c
c
o
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
A
p
r
5
,
2
0
1
9
R
ev
i
s
ed
No
v
9
,
2019
A
cc
ep
ted
No
v
2
5
,
2
0
1
9
T
h
e
si
m
il
a
rit
y
o
r
th
e
d
istan
c
e
m
e
a
su
re
h
a
v
e
b
e
e
n
u
se
d
w
id
e
l
y
to
c
a
lcu
late
th
e
si
m
il
a
rit
y
o
r
d
issi
m
il
a
rit
y
b
e
tw
e
e
n
v
e
c
to
r
se
q
u
e
n
c
e
s,
w
h
e
re
th
e
d
o
c
u
m
e
n
t
im
a
g
e
s
si
m
il
a
rit
y
is
k
n
o
w
n
a
s
th
e
d
o
m
a
in
th
a
t
d
e
a
li
n
g
with
im
a
g
e
in
f
o
rm
a
ti
o
n
a
n
d
b
o
th
sim
il
a
ri
t
y
/d
istan
c
e
h
a
s
b
e
e
n
a
n
im
p
o
rtan
t
ro
le
f
o
r
m
a
tch
in
g
a
n
d
p
a
tt
e
rn
re
c
o
g
n
it
i
o
n
.
T
h
e
re
a
re
se
v
e
ra
l
t
y
p
e
s
o
f
si
m
il
a
rit
y
m
e
a
su
re
,
w
e
c
o
v
e
r
in
th
is
p
a
p
e
r
t
h
e
su
rv
e
y
o
f
v
a
rio
u
s
d
istan
c
e
m
e
a
su
re
s
u
se
d
in
th
e
im
a
g
e
s
m
a
tch
in
g
a
n
d
we
e
x
p
lain
th
e
li
m
it
a
ti
o
n
s
a
ss
o
c
iate
d
w
it
h
th
e
e
x
isti
n
g
d
istan
c
e
s.
T
h
e
n
,
w
e
in
tro
d
u
c
e
th
e
c
o
n
c
e
p
t
o
f
th
e
f
lo
a
ti
n
g
d
istan
c
e
w
h
ich
d
e
sc
rib
e
s
t
h
e
v
a
riatio
n
o
f
th
e
t
h
re
sh
o
l
d
’s
se
lec
ti
o
n
f
o
r
e
a
c
h
w
o
rd
in
d
e
c
isio
n
m
a
k
in
g
p
ro
c
e
ss
,
b
a
se
d
o
n
a
c
o
m
b
in
a
ti
o
n
o
f
L
in
e
a
r
Re
g
re
ss
io
n
a
n
d
c
o
sin
e
d
istan
c
e
.
Ex
p
e
rime
n
ts are
c
a
rried
o
u
t
o
n
a
h
a
n
d
w
rit
ten
A
ra
b
ic
i
m
a
g
e
d
o
c
u
m
e
n
ts
o
f
G
a
l
li
c
a
li
b
ra
ry
.
T
h
e
se
e
x
p
e
ri
m
e
n
ts
sh
o
w
th
a
t
th
e
p
ro
p
o
se
d
f
lo
a
ti
n
g
d
istan
c
e
o
u
tp
e
rf
o
rm
s
th
e
trad
it
io
n
a
l
d
istan
c
e
in
w
o
rd
sp
o
tt
i
n
g
sy
ste
m
.
K
ey
w
o
r
d
s
:
B
ag
-
of
-
v
is
u
al
w
o
r
d
Featu
r
es e
x
tr
ac
tio
n
s
Flo
atin
g
th
r
e
s
h
o
ld
Han
d
w
r
itte
n
A
r
ab
ic
d
o
cu
m
en
t
s
Si
m
i
lar
it
y
d
is
tan
ce
Co
p
y
rig
h
t
©
2
0
2
0
In
stit
u
te o
f
A
d
v
a
n
c
e
d
E
n
g
i
n
e
e
rin
g
a
n
d
S
c
ien
c
e
.
Al
l
rig
h
ts re
se
rv
e
d
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Yo
u
s
s
e
f
E
l
f
ak
ir
,
L
ab
o
r
ato
r
y
o
f
C
o
m
p
u
ti
n
g
a
n
d
I
n
ter
d
is
cip
lin
ar
y
P
h
y
s
ics,
Natio
n
al
S
u
p
er
io
r
Sch
o
o
l,
Sid
i M
o
h
a
m
m
ed
B
en
A
b
d
ellah
U
n
iv
er
s
it
y
,
Fes
3
0
0
5
0
,
M
o
r
o
cc
o
.
E
m
ail:
y
o
u
s
s
e
f
.
elf
a
k
ir
1
@
u
s
m
b
a.
ac
.
m
a
1.
I
NT
RO
D
UCT
I
O
N
I
n
th
e
p
atter
n
r
ec
o
g
n
itio
n
f
ie
ld
s
,
th
e
o
b
j
ec
ts
ca
n
b
e
r
e
p
r
e
s
en
ted
as
s
eq
u
e
n
ce
o
f
f
ea
t
u
r
es,
w
h
er
e
a
s
i
m
ilar
it
y
m
ea
s
u
r
e
ca
n
b
e
u
s
ed
as
a
to
o
l
to
j
u
d
g
e
th
e
s
i
m
ilar
it
y
o
r
d
is
s
i
m
ilar
i
t
y
b
et
w
ee
n
t
w
o
s
eq
u
e
n
ce
s
o
f
f
ea
t
u
r
es.
Fro
m
li
ter
atu
r
e,
s
i
m
i
lar
is
d
e
f
i
n
ed
as
lo
o
k
i
n
g
o
r
b
ein
g
n
ea
r
b
et
w
ee
n
t
w
o
o
b
j
ec
ts
,
b
u
t
n
o
t
th
e
s
a
m
e.
I
n
th
is
f
ield
,
s
ev
er
al
s
i
m
ilar
it
y
an
d
d
is
tan
ce
m
ea
s
u
r
e
h
a
v
e
b
ee
n
u
s
ed
w
id
el
y
i
n
v
ar
io
u
s
f
ie
ld
s
o
f
th
e
ar
ea
:
tex
t
s
i
m
ilar
it
y
[
1
]
,
d
o
cu
m
e
n
t
s
i
m
ilar
it
y
[
2
]
,
h
an
d
w
r
iti
n
g
r
ec
o
g
n
itio
n
[
3
-
5
]
,
h
a
n
d
w
r
it
t
en
ch
ar
ac
ter
[
6
,
7
]
,
s
p
ee
ch
r
ec
o
g
n
itio
n
[
8
-
1
0
]
,
v
i
d
eo
an
al
y
s
i
s
[
1
1
].
So
u
s
e
a
g
o
o
d
s
i
m
ilar
it
y
m
ea
s
u
r
e
is
f
u
n
d
a
m
en
tal
s
tep
f
o
r
m
an
y
t
y
p
e
s
o
f
ap
p
licatio
n
an
d
d
o
m
ai
n
s
u
c
h
as
in
f
o
r
m
at
io
n
r
etr
iev
al
an
d
r
ec
o
g
n
itio
n
,
ch
e
m
is
tr
y
,
clu
s
ter
i
n
g
o
r
class
i
f
icatio
n
.
Th
e
ter
m
s
i
m
ilar
it
y
i
n
t
h
e
p
atter
n
r
ec
o
g
n
itio
n
is
m
ea
n
a
s
co
r
e
th
at
r
ep
r
esen
t
s
t
h
e
s
t
r
en
g
t
h
o
f
r
elatio
n
s
h
ip
b
et
w
ee
n
t
w
o
s
eq
u
en
ce
s
o
f
d
ata
ite
m
s
.
T
h
e
p
atte
r
n
r
ec
o
g
n
i
tio
n
o
r
t
h
e
w
o
r
d
s
p
o
r
tin
g
is
d
o
n
e
b
ased
o
n
a
s
i
m
ilar
it
y
m
ea
s
u
r
e
to
s
e
ar
ch
a
s
i
m
ilar
i
m
a
g
e
t
h
at
ar
e
lo
o
k
in
g
o
r
b
ein
g
n
ea
r
to
g
eth
e
r
.
Ma
th
e
m
atica
ll
y
,
s
i
m
ilar
it
y
d
is
ta
n
ce
is
u
s
ed
to
ca
lcu
late
h
o
w
f
ar
b
et
w
ee
n
t
w
o
s
eq
u
en
ce
s
o
f
d
ata,
is
also
n
a
m
ed
d
is
s
i
m
ilar
it
y
in
o
th
er
d
o
m
ai
n
s
,
o
r
th
e
co
n
ce
p
t
is
th
o
u
g
h
th
e
s
a
m
e
a
n
d
i
n
v
er
s
el
y
[
1
2
]
.
Fig
u
r
e
1
s
h
o
ws
t
h
e
c
h
r
o
n
o
lo
g
ica
l
o
v
er
v
ie
w
o
f
th
e
s
i
m
ilar
it
y
an
d
d
is
tan
ce
m
ea
s
u
r
e
th
at
i
s
co
m
m
o
n
l
y
u
s
ed
in
v
ar
io
u
s
f
ield
s
,
b
ased
o
n
f
ea
t
u
r
e
s
eq
u
en
ce
s
o
f
d
ata
[
1
3
,
1
4
]
,
w
h
ich
ar
e
d
iv
id
ed
in
to
th
r
ee
d
if
f
er
en
t
g
r
o
u
p
s
,
o
n
e
g
r
o
u
p
d
is
ta
n
ce
b
ased
,
th
e
s
ec
o
n
d
a
n
d
th
e
t
h
ir
d
g
r
o
u
p
s
o
f
s
i
m
ilar
it
y
m
ea
s
u
r
e
ar
e
r
esp
ec
tiv
el
y
co
r
r
elatio
n
a
n
d
n
o
n
-
co
r
r
elatio
n
b
ased
.
Si
m
i
lar
it
y
m
ea
s
u
r
e
is
a
n
i
m
p
o
r
tan
t
to
p
ic
b
ee
n
ex
ten
s
i
v
el
y
ap
p
lied
in
s
o
m
e
f
ield
s
s
u
ch
a
s
d
ec
is
io
n
m
ak
in
g
,
p
atter
n
r
ec
o
g
n
itio
n
,
m
ac
h
in
e
lear
n
i
n
g
a
n
d
m
ar
k
et
p
r
ed
ictio
n
b
ased
o
n
d
is
tan
c
e
f
u
n
ctio
n
s
u
c
h
as
E
u
clid
ea
n
,
Ma
n
h
atta
n
,
Mi
n
k
o
w
s
k
i
an
d
C
o
s
i
n
e
d
is
ta
n
ce
s
i
m
ilar
it
y
,
etc.
Or
,
m
a
n
y
li
m
itatio
n
s
a
s
s
o
ciate
d
w
i
th
d
is
tan
ce
m
ea
s
u
r
es
w
h
ich
h
a
v
e
b
ee
n
h
av
e
m
e
n
tio
n
ed
i
n
th
i
s
p
ap
er
an
d
ar
e
aim
i
n
g
to
o
v
er
co
m
e
in
o
u
r
r
esear
ch
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
10
,
No
.
3
,
J
u
n
e
2
0
2
0
:
2
3
6
7
-
2374
2368
Fig
u
r
e
1
.
C
h
r
o
n
o
lo
g
ica
l
o
v
er
v
ie
w
o
f
th
e
s
i
m
ilar
it
y
a
n
d
d
is
ta
n
ce
m
ea
s
u
r
e
s
T
h
e
o
r
g
an
izatio
n
o
f
t
h
i
s
p
ap
er
is
as
f
o
llo
w
s
.
First
w
e
d
escr
ib
e
o
n
s
o
m
e
o
f
c
h
r
o
n
o
lo
g
ica
l
o
v
er
v
ie
w
o
f
th
e
s
i
m
ilar
it
y
a
n
d
d
is
tan
ce
m
ea
s
u
r
e
s
u
s
ed
f
o
r
s
i
m
ilar
it
y
s
ea
r
ch
in
g
an
d
r
etr
iev
i
n
g
s
y
s
te
m
s
.
T
h
en
,
w
e
in
v
est
ig
ate
t
h
e
li
m
itatio
n
s
o
f
t
h
e
ex
i
s
ti
n
g
d
is
ta
n
ce
s
in
w
o
r
d
s
p
o
t
tin
g
s
y
s
te
m
in
d
etail.
Fi
n
all
y
,
w
e
p
r
o
p
o
s
e
a
f
lo
atin
g
d
is
ta
n
c
e
b
ased
o
n
th
e
co
m
b
i
n
atio
n
o
f
L
in
ea
r
R
e
g
r
ess
io
n
an
d
co
s
i
n
e
d
is
tan
ce
a
n
d
w
e
ap
p
ly
t
h
e
m
o
n
a
h
a
n
d
w
r
itte
n
A
r
ab
ic
i
m
ag
e
d
o
cu
m
e
n
ts
o
f
G
al
lica
.
2.
O
VE
RW
I
E
W
O
F
S
I
M
I
L
AR
I
T
Y
AN
D
DIS
T
AN
CE
M
E
T
RICS
Fo
r
y
ea
r
s
,
d
if
f
er
en
t
d
is
tan
ce
m
ea
s
u
r
es
ar
e
u
s
ed
to
ca
lcu
lat
e
th
e
s
i
m
ilar
it
y
b
et
w
ee
n
t
w
o
d
ata
o
b
j
ec
ts
,
th
e
ai
m
o
f
t
h
ese
m
etr
ics
i
s
to
f
o
u
n
d
a
s
p
ec
if
ic
d
is
ta
n
ce
f
u
n
ct
io
n
th
a
t
allo
w
a
s
ep
ar
ati
o
n
o
r
class
if
icatio
n
b
et
w
ee
n
ele
m
en
t
s
i
n
a
s
et
o
f
d
ata
[
1
5
,
1
6
]
.
I
n
f
ac
t,
th
e
s
e
d
is
tan
ce
s
o
p
er
ate
d
ir
ec
tl
y
i
n
v
ar
io
u
s
f
ield
s
a
s
s
i
m
ilar
it
y
s
ea
r
ch
in
g
a
n
d
r
etr
iev
al
s
y
s
te
m
s
w
h
er
e
th
e
p
er
f
o
r
m
an
ce
d
ep
en
d
s
u
p
o
n
c
h
o
o
s
in
g
f
u
n
ctio
n
.
Her
e,
w
e
p
r
ese
n
t
s
ev
er
al
s
i
m
ilar
it
y
d
is
ta
n
ce
s
co
m
m
o
n
l
y
u
s
ed
i
n
p
atter
n
r
ec
o
g
n
itio
n
an
d
w
o
r
d
s
p
o
ttin
g
s
y
s
te
m
s
.
2
.
1
.
E
uclid
ea
n dis
t
a
nce
T
h
e
E
u
clid
ea
n
d
is
ta
n
ce
(
1
)
is
d
ef
in
ed
as
t
h
e
g
e
n
er
al
an
d
t
h
e
s
tan
d
ar
d
m
etr
ic
f
o
r
u
s
ed
in
g
eo
m
e
tr
ical
ca
s
es,
th
e
m
at
h
e
m
atica
l
f
o
r
m
u
la
o
f
th
i
s
d
is
ta
n
ce
is
d
ef
i
n
ed
as
:
=
√
∑
(
−
)
2
=
1
(
1
)
W
h
ich
is
a
s
i
m
p
le
m
e
tr
ic
b
etw
ee
n
t
w
o
p
o
in
ts
o
f
d
ata
(
,
)
th
at
d
eter
m
i
n
es
th
e
r
o
o
t
o
f
s
q
u
ar
e
d
if
f
er
en
ce
b
et
w
ee
n
th
e
m
,
is
t
h
e
d
ef
a
u
lt
m
etr
ic
in
t
h
e
k
-
m
ea
n
s
al
g
o
r
ith
m
an
d
t
h
e
m
o
s
t
u
s
ed
i
n
d
if
f
er
en
t
cl
u
s
ter
i
n
g
p
r
o
b
lem
s
.
2
.
2
.
M
a
nh
a
t
t
a
n dis
t
a
nce
Ma
n
h
a
ttan
o
r
T
ax
icab
d
is
tan
c
e
(
2
)
is
a
m
etr
ic
t
h
at
ca
lc
u
lat
es
th
e
ab
s
o
l
u
te
d
if
f
er
en
ce
b
et
w
ee
n
t
w
o
p
o
in
ts
o
f
d
ata,
is
also
k
n
o
w
n
a
s
L
1
d
is
ta
n
ce
,
L
1
d
is
ta
n
ce
o
r
r
ec
tili
n
ea
r
d
is
ta
n
ce
,
is
d
ef
in
ed
as:
=
|
−
|
(
2
)
2.
3
.
M
ink
o
w
s
k
i di
s
t
a
nce
T
h
e
Min
k
o
w
s
k
i
d
i
s
tan
ce
(
3
)
k
n
o
w
n
as
t
h
e
g
en
er
al
ized
m
et
r
ic,
ca
n
b
e
u
s
ed
i
n
ca
s
e
o
f
d
ata
th
at
ar
e
o
r
d
in
al
an
d
q
u
an
ti
tativ
e.
T
h
e
d
is
tan
ce
o
f
o
r
d
er
p
b
etw
ee
n
t
w
o
v
ar
iab
les i
s
d
ef
i
n
ed
as:
=
√
∑
(
−
)
p
=
1
p
(
3
)
T
h
is
d
is
tan
ce
is
s
i
m
ilar
to
th
e
E
u
clid
ea
n
d
is
ta
n
ce
w
h
en
p
=2
an
d
th
e
Ma
n
h
at
tan
d
is
ta
n
ce
wh
en
p
=1
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
C
o
mb
in
ed
co
s
in
e
-
lin
ea
r
r
eg
r
ess
io
n
mo
d
el
s
imila
r
ity
w
ith
a
p
p
lica
tio
n
to
h
a
n
d
w
r
itten
…
(
Yo
u
s
s
ef
E
lfa
kir
)
2369
2.
4
.
Co
s
ine dis
t
a
nce
C
o
s
i
n
e
d
is
ta
n
ce
(
4
)
is
th
e
co
s
i
n
e
o
f
th
e
an
g
le
b
et
w
ee
n
t
w
o
v
ec
to
r
s
i
n
a
n
n
-
d
i
m
e
n
s
io
n
al
s
p
ac
e
g
iv
e
n
b
y
t
h
e
f
o
llo
w
in
g
f
o
r
m
u
la:
=
X
.
Y
‖
X
‖
∗
‖
Y
‖
=
∑
∗
=
1
√
∑
2
=
1
∗
√
∑
2
=
1
(
4
)
T
h
is
d
is
tan
ce
r
ep
r
esen
ts
t
h
e
d
o
t
p
r
o
d
u
ct
b
etw
ee
n
t
w
o
v
ec
to
r
s
d
iv
id
ed
b
y
th
e
p
r
o
d
u
ct
o
f
th
e
t
w
o
v
ec
to
r
s
'
len
g
th
s
,
an
d
is
o
f
ten
u
s
ed
in
i
n
f
o
r
m
atio
n
r
etr
iev
al
a
n
d
tex
t
m
i
n
in
g
.
2.
5
.
J
a
cc
a
rd
dis
t
a
nce
J
ac
ca
r
d
d
is
tan
ce
(
co
ef
f
icie
n
t)
,
a
ter
m
co
in
ed
b
y
P
au
l
J
ac
ca
r
d
,
m
ea
s
u
r
e
s
s
i
m
ilar
itie
s
o
f
t
h
e
t
w
o
d
ata
ite
m
s
.
T
h
e
J
ac
ca
r
d
d
is
tan
ce
(
5
)
is
m
ea
s
u
r
ed
b
y
th
e
f
o
llo
w
in
g
f
o
r
m
u
la:
=
|
∩
|
|
∪
|
(
5
)
T
h
e
f
u
n
ctio
n
is
b
est
u
s
ed
w
h
e
n
ca
lc
u
lati
n
g
t
h
e
s
i
m
ilar
it
y
b
et
w
ee
n
s
m
all
n
u
m
b
er
s
o
f
s
ets,
an
d
is
o
f
te
n
u
s
ed
t
o
in
th
e
r
ec
o
m
m
e
n
d
atio
n
f
lied
.
2.
6
.
Cheby
s
hev
dis
t
a
nce
C
h
eb
y
s
h
ev
d
is
ta
n
ce
(
6
)
,
is
al
s
o
ca
lled
T
ch
eb
y
ch
e
v
o
r
t
h
e
m
a
x
i
m
u
m
v
a
lu
e
d
is
ta
n
ce
.
T
h
is
m
e
tr
ic
ca
lcu
late
s
th
e
ab
s
o
l
u
te
m
a
g
n
it
u
d
e
o
f
th
e
d
i
f
f
er
e
n
ce
b
et
w
ee
n
t
w
o
p
o
in
t
s
o
f
d
ata
b
y
f
o
llo
w
i
n
g
:
=
(
|
−
|
)
(
6
)
3.
CO
M
P
ARIS
O
N
O
F
DIS
T
A
NCES
M
E
T
R
I
C
I
N
WO
R
D
SPO
T
T
I
N
G
SYS
T
E
M
S
T
h
e
ai
m
o
f
Q
u
er
y
-
by
-
e
x
a
m
p
l
e
w
o
r
d
s
p
o
ttin
g
i
s
s
ea
r
c
h
an
d
lo
ca
lize
th
e
y
r
eg
io
n
s
t
h
at
ar
e
s
i
m
ilar
f
o
r
a
g
i
v
e
n
q
u
er
y
.
Or
to
g
et
t
h
e
b
est
p
er
f
o
r
m
a
n
ce
s
till
as
a
c
h
allen
g
e,
b
ec
au
s
e
s
e
v
er
al
tas
k
s
in
f
l
u
en
ce
o
n
th
is
s
y
s
te
m
,
as
i
m
a
g
e
p
r
o
ce
s
s
in
g
,
f
ea
t
u
r
e
ex
tr
ac
tio
n
s
,
s
i
m
i
lar
it
y
m
ea
s
u
r
es,
clas
s
if
icatio
n
.
W
h
il
e
o
u
r
w
o
r
k
f
o
cu
s
es
o
n
th
e
i
n
f
l
u
en
ce
o
f
s
i
m
i
lar
it
y
m
ea
s
u
r
e
d
is
ta
n
ce
s
i
n
t
h
e
co
n
tex
t
o
f
a
h
a
n
d
w
r
itte
n
d
o
cu
m
e
n
t
i
m
a
g
es
r
etr
iev
a
l
w
h
er
e
m
an
y
ex
i
s
ti
n
g
m
etr
ics
ca
n
b
e
ad
ap
ted
.
W
e
u
n
d
er
li
n
e
t
h
at
th
e
p
r
esen
ted
m
ea
s
u
r
e
o
f
s
i
m
ilar
it
y
i
s
ap
p
licab
le
to
an
y
p
r
o
b
le
m
w
h
er
e
th
e
s
ize
o
f
q
u
er
y
is
d
if
f
er
e
n
t o
n
e
f
r
o
m
th
e
o
t
h
er
.
I
n
th
is
p
ar
t,
w
e
p
r
esen
t
a
co
m
p
ar
is
o
n
o
f
d
is
tan
ce
m
ea
s
u
r
e
s
in
w
o
r
d
s
p
o
ttin
g
s
y
s
te
m
.
Fo
r
th
is
,
w
e
u
s
e
th
e
A
r
ab
ic
h
a
n
d
w
r
it
ten
d
o
cu
m
en
t
i
m
a
g
es
f
o
r
m
Gall
ica,
w
h
ic
h
i
s
t
h
e
d
i
g
ital
lib
r
ar
y
o
f
th
e
Natio
n
a
l
L
ib
r
ar
y
o
f
Fra
n
ce
,
i
n
o
p
en
ac
ce
s
s
.
Firs
t,
w
e
d
iv
id
e
t
h
e
d
o
cu
m
en
t
i
m
a
g
es
i
n
to
a
s
et
o
f
lo
ca
l
r
eg
io
n
s
,
d
en
s
el
y
s
a
m
p
led
;
th
ese
lo
ca
l
r
eg
io
n
s
ar
e
th
e
b
as
ic
s
tr
u
ct
u
r
e
u
s
ed
to
s
p
o
t
th
e
wo
r
d
s
in
th
e
d
o
cu
m
e
n
t,
a
n
d
w
e
ex
tr
ac
t
all
in
ter
est
s
p
o
in
ts
f
r
o
m
th
e
i
m
a
g
es
u
s
i
n
g
SIFT
[
1
7
]
d
etec
to
r
,
an
d
w
e
u
s
e
SIFT
d
escr
ip
to
r
to
r
e
p
r
esen
t
ea
ch
in
ter
e
s
t
p
o
in
t
in
t
h
e
i
m
a
g
es b
y
t
h
eir
d
escr
ip
t
o
r
.
T
h
en
,
i
n
t
h
e
lear
n
in
g
s
tep
,
w
e
s
elec
t t
h
e
4
f
ir
s
t
i
m
ag
e
s
o
f
d
o
cu
m
en
t to
g
r
o
u
p
all
d
escr
ip
to
r
s
to
p
r
o
v
id
e
b
a
g
o
f
v
is
u
al
d
escr
ip
to
r
s
an
d
cl
u
s
t
er
ce
n
ter
s
,
in
t
h
is
s
ta
g
e,
t
h
e
k
-
m
ea
n
s
a
lg
o
r
it
h
m
is
u
s
ed
w
i
th
d
if
f
er
en
t
n
u
m
b
er
o
f
ce
n
ter
s
(
1
0
0
,
2
0
0
,
3
0
0
,
4
0
0
)
.
I
n
th
e
e
n
d
,
w
e
e
n
co
d
in
g
ea
c
h
r
eg
io
n
b
y
a
h
is
to
g
r
a
m
u
s
i
n
g
a
b
ag
-
of
-
v
i
s
u
al
-
d
escr
ip
to
r
s
[
1
8
,
1
9
]
in
s
tead
to
r
ep
r
esen
t
t
h
e
m
b
y
th
eir
d
escr
ip
t
o
r
s
.
A
ll
r
eg
io
n
s
ar
e
d
escr
ib
ed
w
i
th
t
h
eir
h
is
to
g
r
a
m
s
b
y
a
s
s
i
g
n
in
g
ea
c
h
d
escr
ip
to
r
in
th
e
r
eg
io
n
to
th
e
n
ea
r
e
s
t
v
i
s
u
al
d
escr
ip
to
r
in
th
e
co
d
eb
o
o
k
.
So
,
ea
c
h
r
eg
io
n
is
r
ep
r
esen
ted
b
y
a
h
is
to
g
r
a
m
o
f
ac
cu
m
u
lates
f
r
eq
u
e
n
cies
j
=
1
,
,
2
,
,
3
,
,
4
,
,
5
,
,
…
…
,
w
h
er
e
N
r
ep
r
esen
ts
th
e
n
u
m
b
er
o
f
th
e
v
i
s
u
al
d
escr
ip
to
r
in
th
e
co
d
eb
o
o
k
an
d
,
r
ep
r
esen
ts
th
e
cu
m
u
lati
v
e
f
r
eq
u
e
n
c
y
o
f
k
v
is
u
al
d
escr
ip
to
r
in
th
e
j
e
m
e
r
eg
io
n
.
T
h
e
p
r
o
ce
s
s
o
f
t
h
e
u
s
ed
s
y
s
te
m
i
s
p
r
ese
n
ted
in
F
i
g
u
r
e
2
.
T
o
ev
al
u
ate
t
h
e
e
f
f
ec
t
o
f
s
i
m
i
lar
i
t
y
m
ea
s
u
r
e
an
d
d
eter
m
i
n
e
t
h
e
ap
p
r
o
p
r
iate
d
is
tan
ce
in
th
e
co
n
te
x
t
o
f
w
o
r
d
s
p
o
tti
n
g
,
w
e
c
h
an
g
e
t
h
is
m
etr
ic
an
d
w
e
ca
lcu
late
th
e
d
etec
tio
n
ac
c
u
r
ac
y
r
ep
r
esen
ted
b
y
F
-
s
co
r
e
f
o
r
d
if
f
er
en
t
q
u
er
ies.
T
h
e
F
-
s
co
r
e
is
ca
lcu
lated
as f
o
llo
w
s
:
F
S
co
r
e
=
2
.
1
1
r
ec
al
l
+
1
pr
ec
is
io
n
=
2
.
r
ecal
l
+
p
r
e
cis
i
o
n
r
ecal
l
+
p
r
e
cis
i
o
n
(
7
)
T
ab
le
1
s
h
o
w
s
th
e
F
-
s
co
r
e
r
e
s
u
lt
s
f
o
r
Gallica
d
ata
s
et,
w
e
o
b
s
er
v
e
th
at
t
h
e
F
-
s
co
r
e
d
ep
en
d
in
g
o
n
th
e
co
d
eb
o
o
k
s
ize
a
n
d
al
s
o
th
e
m
etr
ic
d
i
s
ta
n
ce
u
s
ed
.
Fo
r
t
h
e
co
s
in
e
m
etr
ic
an
d
co
d
e
b
o
o
k
s
ize
3
0
0
w
e
o
b
tain
F
-
s
co
r
e=
0
,
7
8
.
W
h
ile
f
o
r
t
h
e
o
th
er
m
etr
ic
w
e
r
ep
o
r
ted
F
-
s
o
cr
e
=0
,
6
2
f
o
r
E
u
clid
ea
n
d
is
ta
n
ce
,
F
-
s
o
cr
e
=
0
,
5
2
f
o
r
C
h
eb
y
s
h
ev
d
is
tan
ce
.
So
,
th
e
b
est r
esu
lt is
o
b
tai
n
ed
f
o
r
th
e
co
s
in
e
m
e
tr
ic
an
d
3
0
0
clu
s
ter
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
10
,
No
.
3
,
J
u
n
e
2
0
2
0
:
2
3
6
7
-
2374
2370
Fig
u
r
e
2
.
P
r
o
ce
s
s
o
f
th
e
p
r
o
p
o
s
ed
s
y
s
te
m
T
ab
le
1
.
F
-
score
f
o
r
d
if
f
er
e
n
t c
o
d
eb
o
o
k
s
izes a
n
d
d
if
f
er
en
t
m
et
r
ic
d
is
tan
ce
s
C
o
d
e
b
o
o
k
s
i
z
e
1
0
0
2
0
0
3
0
0
4
0
0
Eu
c
l
i
d
e
a
n
d
i
st
a
n
c
e
0
.
27
0
.
58
0
.
62
0
.
53
M
a
n
h
a
t
t
a
n
d
i
s
t
a
n
c
e
0
.
4
0
.
46
0
.
5
0
.
48
M
i
n
k
o
w
sk
i
d
i
st
a
n
c
e
0
.
15
0
.
23
0
.
34
0
.
30
C
o
si
n
e
d
i
s
t
a
n
c
e
0
.
6
2
7
0
.
76
0
.
78
0
.
6
2
9
Jac
c
a
r
d
d
i
st
a
n
c
e
0
.
27
0
.
36
0
.
5
0
.
41
C
h
e
b
y
sh
e
v
d
i
s
t
a
n
c
e
0
.
37
0
.
42
0
.
52
0
.
45
4.
L
I
N
E
AR
R
E
G
RE
SS
I
O
N
M
O
DE
L
S
I
M
I
L
ARI
T
Y
W
I
T
H
AP
P
L
I
CA
T
I
O
N
T
O
H
ANDWR
I
T
T
E
N
WO
RD
SPO
T
T
I
N
G
4
.
1
.
Ana
ly
s
is
o
f
c
o
s
ine dis
t
a
nce
m
ea
s
ure
f
o
r
w
o
rd
s
po
t
t
i
ng
As
s
h
o
w
n
i
n
th
e
p
r
ev
io
u
s
p
ar
t,
th
e
m
etr
ic
t
h
at
g
i
v
es
t
h
e
b
est
r
esu
lt
is
t
h
e
co
s
in
e
d
is
tan
ce
.
Ho
w
e
v
er
,
T
ab
le
1
s
h
o
w
s
t
h
at
th
e
s
ize
o
f
th
e
co
d
eb
o
o
k
in
f
lu
e
n
ce
d
i
r
ec
tl
y
o
n
th
e
p
er
f
o
r
m
an
ce
.
S
o
w
e
b
elie
v
e
it
is
th
er
ef
o
r
e
n
ec
ess
ar
i
l
y
to
an
al
y
s
es
d
i
v
er
s
e
p
h
e
n
o
m
e
n
a
th
at
a
p
p
ea
r
w
h
e
n
u
s
i
n
g
co
d
eb
o
o
k
in
h
ig
h
-
d
i
m
e
n
s
io
n
al
s
p
ac
es,
th
i
s
p
h
e
n
o
m
en
a
ca
n
b
e
ex
p
lain
b
y
t
h
e
cu
r
s
e
o
f
d
i
m
en
s
io
n
a
lit
y
,
th
e
ter
m
w
as
i
n
v
e
n
ted
b
y
R
ic
h
ar
d
E
.
B
ellm
a
n
i
n
[
2
0
,
2
1
]
.
Fo
r
th
is
,
t
h
e
s
y
s
te
m
p
r
esen
ted
i
n
F
ig
u
r
e
2
is
th
er
ef
o
r
e
u
s
ed
to
ev
alu
ate
i
n
d
etail
d
if
f
er
e
n
t size
s
o
f
t
h
e
co
d
eb
o
o
k
f
r
o
m
1
0
0
to
9
0
0
ce
n
ter
s
.
T
h
en
,
to
ca
lcu
late
th
e
s
i
m
ilar
i
t
y
b
et
w
ee
n
t
h
e
h
is
to
g
r
a
m
s
o
f
ea
ch
r
eg
io
n
i
n
t
h
e
i
m
ag
e
d
o
c
u
m
e
n
t
a
n
d
th
e
q
u
er
y
’
s
h
i
s
to
g
r
a
m
,
w
e
u
s
e
th
e
co
s
i
n
e
s
i
m
ilar
it
y
d
ef
i
n
ed
b
y
:
S
=
1
−
∑
H
i
,
j
R
i
N
i
=
1
√
∑
H
i
,
j
2
N
i
=
1
√
∑
R
i
2
N
i
=
1
(
8
)
W
h
er
e
H
i
,
j
r
ep
r
esen
ts
t
h
e
o
cc
u
r
r
en
ce
o
f
t
h
e
ie
m
e
ce
n
ter
o
f
th
e
c
o
d
eb
o
o
k
in
t
h
e
j
em
e
r
e
g
io
n
,
a
n
d
R
i
r
ep
r
esen
t
s
th
e
o
cc
u
r
r
en
ce
o
f
th
e
ie
m
e
ce
n
ter
o
f
t
h
e
co
d
eb
o
o
k
.
Or
,
to
s
elec
t
th
e
m
r
eg
io
n
s
t
h
at
ar
e
s
i
m
ilar
to
th
e
q
u
er
y
,
th
e
co
s
in
e
s
i
m
i
lar
it
y
“
S”
s
h
o
u
ld
b
e
in
f
er
io
r
to
a
ce
r
tai
n
th
r
es
h
o
ld
.
Fo
r
tw
o
r
eg
io
n
s
p
er
f
ec
tl
y
s
i
m
ilar
∑
H
i
,
j
R
i
N
i
=
1
√
∑
H
i
,
j
2
N
i
=
1
√
∑
R
i
2
N
i
=
1
=
1
,
s
o
S=0
.
T
h
er
ef
o
r
e,
as
m
u
c
h
as
th
e
th
r
es
h
o
ld
is
les
s
,
th
e
r
eg
io
n
s
ar
e
m
o
r
e
s
i
m
ilar
.
Fo
r
th
is
,
f
o
r
ea
ch
co
d
eb
o
o
k
s
ize,
w
e
ca
lcu
la
te
th
e
F
-
Sco
r
e
b
y
ch
o
o
s
i
n
g
th
e
b
est
t
h
r
esh
o
ld
o
f
ea
ch
s
ize
as
s
h
o
w
n
i
n
Fi
g
u
r
e
3
.
W
e
o
b
s
er
v
e
th
at
th
e
s
ize
i
n
f
l
u
en
ce
o
n
t
h
e
p
er
f
o
r
m
an
ce
an
d
b
est
r
esu
lt
is
g
iv
e
n
f
o
r
k
=3
0
0
an
d
d
ec
r
ea
s
e
b
ey
o
n
d
t
h
is
s
ize.
Fig
u
r
e
3
.
T
h
e
F
-
S
c
ore
cu
r
v
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
C
o
mb
in
ed
co
s
in
e
-
lin
ea
r
r
eg
r
ess
io
n
mo
d
el
s
imila
r
ity
w
ith
a
p
p
lica
tio
n
to
h
a
n
d
w
r
itten
…
(
Yo
u
s
s
ef
E
lfa
kir
)
2371
No
w
,
w
e
s
ea
r
ch
th
e
i
n
f
lu
e
n
c
e
o
f
th
e
co
d
eb
o
o
k
s
ize
o
n
th
e
th
r
esh
o
ld
.
Fo
r
m
(
2
)
,
w
h
en
th
e
s
ize
o
f
th
e
co
d
eb
o
o
k
in
cr
ea
s
e
(
N)
,
th
e
p
r
o
b
ab
ilit
y
to
f
i
n
d
ze
r
o
d
esc
r
ip
to
r
in
ea
ch
ce
llu
le
in
th
e
h
i
s
to
g
r
a
m
f
o
r
a
g
iv
e
n
r
eg
io
n
i
n
cr
ea
s
e
as s
h
o
w
n
i
n
Fi
g
u
r
e
3
,
th
at
m
ea
n
:
∀
i
≤
N,
W
h
en
N
in
cr
ea
s
e,
th
e
p
r
o
b
ab
ilit
y
P
(
R
i
)
=0
in
cr
ea
s
e
W
h
en
N
d
ec
r
ea
s
e,
th
e
p
r
o
b
ab
ilit
y
P
(
R
i
)
=0
d
ec
r
ea
s
e
Su
b
s
eq
u
e
n
tl
y
,
∀
i
≤
N,
W
h
en
N
in
cr
ea
s
e,
th
e
p
r
o
b
ab
ilit
y
∑
P
(
H
i
,
j
)
*P
(
R
i
)
=
1
d
ec
r
ea
s
e
W
h
en
N
d
ec
r
ea
s
e,
th
e
p
r
o
b
ab
ilit
y
∑
P
(
H
i
,
j
)
*P
(
R
i
)
=
1
in
cr
ea
s
e
So
,
whe
n
N
in
c
r
e
a
s
e
,
∑
H
i
,
j
R
i
N
i
=
1
√
∑
H
i
,
j
2
N
i
=
1
√
∑
R
i
2
N
i
=
1
S
=
1
−
∑
H
i
,
j
R
i
N
i
=
1
√
∑
H
i
,
j
2
N
i
=
1
√
∑
R
i
2
N
i
=
1
W
h
er
e
R
i
r
ep
r
esen
ts
th
e
h
is
to
g
r
a
m
o
f
t
h
e
q
u
er
y
,
Hi,
j
r
ep
r
esen
ts
t
h
e
h
is
to
g
r
a
m
o
f
j
e
m
e
r
eg
io
n
in
th
e
d
o
cu
m
en
t
an
d
∑
P
(
H
i
,
j
)
=
=
1
is
th
e
n
u
m
b
er
o
f
in
ter
e
s
t
p
o
in
ts
in
j
e
m
e
r
eg
io
n
.
So
w
h
e
n
th
e
s
ize
o
f
h
is
to
g
r
a
m
(
N)
in
cr
ea
s
e,
th
e
n
u
m
b
er
o
f
ze
r
o
c
ellu
le
i
n
cr
ea
s
e
to
o
.
C
o
n
s
eq
u
e
n
tl
y
,
t
h
e
th
r
e
s
h
o
ld
s
h
o
u
ld
tak
e
a
co
u
n
t
th
e
s
ize
o
f
th
e
co
d
eb
o
o
k
(
n
u
m
b
er
o
f
clu
s
ter
)
.
Fi
g
u
r
e
4
s
h
o
w
s
e
x
a
m
p
le
o
f
q
u
er
y
an
d
r
eg
io
n
h
is
to
g
r
a
m
s
.
Fo
r
th
e
e
x
p
er
i
m
e
n
ts
s
tep
,
F
ig
u
r
e
5
s
h
o
w
s
t
h
e
c
u
r
v
e
o
f
t
h
e
b
est
th
r
es
h
o
ld
f
o
r
ea
ch
co
d
eb
o
o
k
s
ize.
T
h
e
th
r
esh
o
ld
d
ep
en
d
s
o
n
th
e
s
ize
o
f
th
e
co
d
eb
o
o
k
an
d
c
an
r
ep
r
esen
ted
b
y
:
y
=0
.
0
0
0
2
6
5
x
+
0
.
3
9
9
3
.
Fig
u
r
e
4
.
E
x
a
m
p
le
o
f
q
u
er
y
an
d
r
eg
io
n
h
i
s
to
g
r
a
m
s
Fig
u
r
e
5
.
B
est th
r
esh
o
ld
f
o
r
ea
ch
co
d
eb
o
o
k
s
ize
A
t
th
i
s
s
tag
e,
a
n
d
to
p
er
f
o
r
m
th
e
e
x
p
er
i
m
e
n
ts
,
w
e
s
h
o
u
ld
s
e
ar
ch
th
e
in
f
l
u
e
n
ce
o
f
t
h
e
i
n
ter
est
p
o
in
t
s
n
u
m
b
er
(
n
u
m
b
er
o
f
d
escr
ip
to
r
s
n
)
o
n
th
e
th
r
e
s
h
o
ld
.
Fo
r
m
(
2
)
,
w
h
e
n
t
h
e
n
u
m
b
er
o
f
i
n
ter
est
p
o
in
ts
n
in
cr
ea
s
e,
th
e
p
r
o
b
ab
ilit
y
to
f
in
d
ze
r
o
d
escr
ip
to
r
in
ea
c
h
ce
llu
le
i
n
th
e
h
is
to
g
r
a
m
f
o
r
a
g
iv
e
n
r
eg
io
n
d
ec
r
ea
s
e
as
s
h
o
w
n
in
Fig
u
r
e
3
,
th
at
m
ea
n
:
∀
i
≤
N,
W
h
en
n
in
cr
ea
s
e,
th
e
p
r
o
b
ab
ilit
y
P
(
R
i
)
=0
d
ec
r
ea
s
e
W
h
en
n
d
ec
r
ea
s
e,
th
e
p
r
o
b
ab
ilit
y
P
(
R
i
)
=0
in
cr
ea
s
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
10
,
No
.
3
,
J
u
n
e
2
0
2
0
:
2
3
6
7
-
2374
2372
Su
b
s
eq
u
e
n
tl
y
,
∀
i
≤
N,
W
h
en
n
in
cr
ea
s
e,
th
e
p
r
o
b
ab
ilit
y
∑
P
(
H
i
,
j
)
*P
(
R
i
)
=
1
in
cr
ea
s
e
W
h
en
n
d
ec
r
ea
s
e,
th
e
p
r
o
b
ab
ilit
y
∑
P
(
H
i
,
j
)
*P
(
R
i
)
=
1
d
ec
r
ea
s
e
So
,
whe
n
n
in
c
r
e
a
s
e
,
∑
H
i
,
j
R
i
N
i
=
1
√
∑
H
i
,
j
2
N
i
=
1
√
∑
R
i
2
N
i
=
1
S
=
1
−
∑
H
i
,
j
R
i
N
i
=
1
√
∑
H
i
,
j
2
N
i
=
1
√
∑
R
i
2
N
i
=
1
I
n
f
ac
t,
w
h
e
n
th
e
n
u
m
b
er
o
f
in
ter
est
p
o
in
ts
(
n
)
i
n
cr
ea
s
e,
t
h
e
n
u
m
b
er
o
f
ze
r
o
ce
llu
le
d
e
cr
ea
s
e
to
o
.
C
o
n
s
eq
u
en
tl
y
,
t
h
e
t
h
r
es
h
o
ld
s
h
o
u
ld
tak
e
a
co
u
n
t
th
e
n
u
m
b
er
o
f
i
n
ter
est
p
o
in
t
s
(
n
u
m
b
er
o
f
d
escr
ip
to
r
s
)
,
b
ec
au
s
e
t
h
is
n
u
m
b
er
ch
a
n
g
e
f
r
o
m
r
eg
io
n
to
o
th
er
.
E
x
p
e
r
i
m
en
tl
y
,
Fig
u
r
e
6
s
h
o
w
s
t
h
e
th
r
es
h
o
ld
cu
r
v
e
ac
co
r
d
in
g
to
th
e
n
u
m
b
er
o
f
d
escr
ip
to
r
s
w
h
er
e
w
e
ca
n
s
e
e
th
at
th
e
t
h
r
es
h
o
ld
d
ep
en
d
o
n
th
e
d
escr
ip
to
r
s
,
th
is
d
ep
en
d
e
n
ce
ca
n
b
e
r
ep
r
esen
ted
b
y
:
y=
-
0
.
0
0
0
9
x
+0
.
5
2
1
9
.
Fig
u
r
e
6
.
T
h
r
esh
o
ld
ac
co
r
d
in
g
to
th
e
n
u
m
b
er
o
f
d
escr
ip
to
r
s
Giv
e
n
t
h
is
co
n
d
itio
n
,
t
w
o
r
e
g
i
o
n
s
ar
e
s
i
m
ilar
if
th
e
th
r
es
h
o
l
d
S
is
le
s
s
t
h
a
n
a
g
iv
e
n
v
al
u
e,
o
r
if
t
w
o
h
av
i
n
g
e
v
en
n
u
m
b
er
o
f
in
ter
est
p
o
in
ts
w
ill
h
a
v
e
s
a
m
e
t
h
r
esh
o
ld
,
b
u
t
t
h
is
n
o
t
m
ea
n
th
at
th
e
y
ar
e
s
i
m
ilar
,
b
ec
au
s
e
m
a
y
b
e
th
e
y
h
a
v
e
d
i
f
f
er
en
t
h
is
to
g
r
a
m
s
an
d
t
h
er
ea
f
ter
d
if
f
er
en
t
d
is
tan
ce
s
o
f
s
i
m
i
lar
it
y
.
T
o
co
n
clu
d
e
th
is
p
ar
t,
ea
ch
w
o
r
d
/
r
eg
io
n
i
n
d
o
cu
m
en
t
h
a
s
a
ce
r
tain
n
u
m
b
er
o
f
in
ter
est
p
o
in
t
s
/d
escr
ip
to
r
s
,
s
o
a
co
m
p
r
o
m
i
s
e
b
et
w
ee
n
th
e
n
u
m
b
er
o
f
d
escr
i
p
to
r
s
an
d
th
e
th
r
es
h
o
ld
m
u
s
t b
e
s
o
u
g
h
t.
4
.
2
.
Co
m
bi
na
t
io
n o
f
co
s
ine
-
lin
ea
r
re
g
re
s
s
io
n si
m
ila
rit
y
R
eg
r
es
s
io
n
f
o
cu
s
es
o
n
th
e
r
elatio
n
s
h
ip
b
et
w
ee
n
t
h
e
o
u
tc
o
m
e
s
a
n
d
th
e
i
n
p
u
t
s
.
I
t
also
p
r
o
v
id
es
a
m
o
d
el
th
a
t
h
as
s
o
m
e
ex
p
la
n
ato
r
y
v
alu
e,
i
n
o
u
r
ca
s
e,
th
e
in
p
u
ts
ar
e
s
ize
o
f
co
d
eb
o
o
k
an
d
n
u
m
b
er
o
f
in
ter
es
t
p
o
in
ts
,
th
e
o
u
t
co
m
e
is
t
h
e
t
h
r
esh
o
ld
v
al
u
e
t
h
at
d
ef
i
n
e
if
t
wo
r
eg
io
n
s
ar
e
s
i
m
ilar
o
r
n
o
t.
L
i
n
ea
r
r
eg
r
ess
io
n
is
a
co
m
m
o
n
l
y
u
s
ed
tech
n
iq
u
e
f
o
r
m
o
d
elin
g
w
h
e
n
th
e
o
u
tco
m
e
v
ar
iab
le
is
ex
p
r
ess
ed
as
li
n
ea
r
co
m
b
i
n
atio
n
o
f
th
e
i
n
p
u
t
v
ar
iab
les,
f
o
r
a
g
iv
en
s
e
t
o
f
i
n
p
u
t
v
ar
iab
les
,
th
e
lin
ea
r
r
e
g
r
ess
io
n
m
o
d
el
p
r
o
v
id
es
t
h
e
e
x
p
ec
ted
o
u
tco
m
e
v
a
lu
e.
=
0
+
1
1
+
2
2
+
⋯
+
−
1
−
1
+
(
9
)
w
h
er
e
is
th
e
th
r
esh
o
ld
v
ar
iab
le
ar
e
th
e
in
p
u
t v
ar
iab
les,
f
o
r
j
=1
,
2
,
…,
p
-
1
0
is
th
e
v
alu
e
of
wh
en
each
eq
u
als
ze
r
o
{Of
f
s
et}
is
th
e
ch
an
g
e
in
b
ased
on
a
u
n
it
ch
an
g
e
in
{Co
ef
f
icie
n
t}
~N(
0
,
2
)
an
d
th
e
’
s
ar
e
in
d
ep
en
d
en
t
o
f
ea
ch
o
th
er
T
h
is
r
an
d
o
m
er
r
o
r
(
)
,
d
en
o
ted
b
y
ɛ,
is
ass
u
m
ed
to
b
e
n
o
r
m
all
y
d
is
tr
ib
u
ted
w
i
th
a
m
ea
n
o
f
ze
r
o
an
d
a
co
n
s
ta
n
t v
ar
ia
n
ce
(
2
).
So
,
th
e
u
s
e
d
E
q
u
atio
n
is
: T
=
-
0
,
0
0
0
9
*
n
+0
,
5
2
1
9
f
o
r
N=
3
0
0
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
C
o
mb
in
ed
co
s
in
e
-
lin
ea
r
r
eg
r
ess
io
n
mo
d
el
s
imila
r
ity
w
ith
a
p
p
lica
tio
n
to
h
a
n
d
w
r
itten
…
(
Yo
u
s
s
ef
E
lfa
kir
)
2373
Ho
w
e
v
er
,
th
e
p
er
f
o
r
m
an
ce
o
f
r
eg
r
ess
io
n
an
al
y
s
is
m
et
h
o
d
s
i
n
p
r
ac
tice
d
ep
en
d
s
o
n
t
h
e
t
y
p
e
an
d
f
o
r
m
o
f
th
e
u
s
ed
d
ata,
an
d
h
o
w
it
r
elate
s
to
th
e
u
s
ed
r
eg
r
ess
i
o
n
m
et
h
o
d
.
A
s
is
s
h
o
w
n
i
n
F
ig
u
r
e
s
4
an
d
5
,
th
e
t
h
r
es
h
o
ld
d
ep
en
d
s
o
n
th
e
co
d
eb
o
o
k
s
ize
an
d
t
h
e
n
u
m
b
er
o
f
d
escr
ip
to
r
s
at
ea
ch
w
o
r
d
,
o
r
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
is
a
g
e
n
er
ic
s
i
m
i
lar
it
y
m
ea
s
u
r
e
t
h
at
ta
k
e
a
co
u
n
t
t
h
e
s
e
p
ar
a
m
eter
s
a
n
d
it
d
o
es
n
o
t
m
ak
e
an
y
s
p
ec
i
f
ic
co
n
d
itio
n
to
h
an
d
w
r
itte
n
d
o
cu
m
e
n
t
s
.
T
h
e
p
r
o
p
o
s
ed
s
im
ilar
it
y
m
ea
s
u
r
e
is
ev
al
u
ated
in
t
h
e
co
n
tex
t
o
f
w
o
r
d
s
p
o
ttin
g
ta
s
k
.
T
h
e
ai
m
co
n
s
is
ts
in
q
u
er
y
in
g
A
r
ab
ic
o
f
h
an
d
w
r
itte
n
d
o
cu
m
e
n
t
s
w
it
h
q
u
er
y
-
by
-
e
x
a
m
p
l
e
an
ag
r
a
m
an
d
i
n
s
elec
ti
n
g
/r
ep
o
r
tin
g
r
e
g
io
n
s
th
at
s
i
m
ilar
to
th
e
q
u
er
y
.
W
e
r
ep
o
r
t
h
er
e
s
o
m
e
q
u
er
ies
w
h
er
e
th
e
s
y
s
te
m
y
ie
ld
s
au
to
m
atica
ll
y
as
s
h
o
w
n
i
n
Fig
u
r
e
7
,
w
h
ic
h
ar
e
s
i
m
ilar
to
t
h
e
q
u
er
y
an
d
w
i
th
o
u
t
c
h
o
s
e
t
h
e
th
r
eso
l
d
,
w
h
ic
h
i
s
a
p
r
o
b
lem
i
n
o
th
e
r
s
y
s
te
m
[
2
2
,
2
3
]
.
T
h
en
,
w
e
u
s
e
a
f
i
lter
in
g
s
tep
to
s
elec
t
o
n
e
b
est
r
es
u
lt
w
h
e
n
co
n
f
u
s
io
n
r
eg
io
n
s
ar
e
r
et
u
r
n
i
n
g
,
b
ased
o
n
th
eir
s
s
i
m
ilar
it
y
s
co
r
es
a
n
d
th
eir
s
p
o
s
itio
n
s
.
T
ab
le
2
s
h
o
w
s
a
co
m
p
ar
i
s
o
n
o
f
t
h
e
p
r
o
p
o
s
ed
m
e
th
o
d
w
it
h
s
tate
-
of
-
th
e
-
ar
t
in
ter
m
o
f
P
r
ec
is
io
n
,
t
s
eted
in
th
e
Gallica
d
atase
t.
An
i
m
p
r
o
v
e
m
en
t
is
o
b
s
er
v
ed
w
i
th
th
e
p
r
o
p
o
s
ed
s
i
m
ilar
it
y
d
i
s
tan
ce
o
v
er
o
th
er
m
et
h
o
d
s
.
T
ab
le
2
.
P
er
f
o
r
m
a
n
ce
o
f
t
h
e
p
r
o
p
o
s
ed
m
et
h
o
d
an
d
o
th
er
w
o
r
k
s
M
e
t
h
o
d
P
r
e
c
i
si
o
n
A
l
maz
o
n
e
t
a
l
.
[
2
3
]
68
.
4%
R
u
s
i
n
o
l
e
t
a
l
.
[
2
4
]
8
1
%
P
r
o
p
o
se
d
me
t
h
o
d
8
3
%
H
o
w
e
e
t
a
l
.
[
2
5
]
7
9
%
L
i
a
n
g
e
t
a
l
.
[
2
6
]
6
7
%
F
i
sch
e
r
e
t
a
l
.
[
2
7
]
6
2
%
T
e
r
a
saw
a
e
t
a
l
.
[
2
8
]
7
9
%
Fig
u
r
e
7
.
T
h
e
r
etr
iev
ed
r
eg
io
n
s
f
o
r
s
o
m
e
q
u
er
ie
s
5.
CO
NCLU
SI
O
N
I
n
th
i
s
p
ap
er
,
w
e
p
r
ese
n
t
a
g
en
er
ic
s
i
m
ilar
it
y
m
ea
s
u
r
e
f
o
r
w
o
r
d
s
p
o
ttin
g
s
y
s
te
m
th
a
t
ta
k
e
a
co
u
n
t
th
e
co
d
eb
o
o
k
s
ize
an
d
t
h
e
n
u
m
b
er
o
f
d
escr
ip
to
r
s
at
q
u
er
y
,
an
d
it
d
o
es
n
o
t
m
a
k
e
an
y
s
p
ec
if
ic
co
n
d
it
io
n
to
h
an
d
w
r
itte
n
d
o
cu
m
en
t
s
,
a
co
m
p
ar
i
s
o
n
o
f
t
h
e
p
r
o
p
o
s
ed
m
et
h
o
d
w
it
h
o
th
er
m
e
th
o
d
s
s
h
o
w
s
t
h
at
g
e
n
er
ic
s
i
m
ilar
it
y
m
ea
s
u
r
e
g
i
v
es
a
n
e
x
ce
lle
n
t
r
es
u
lt
i
n
ter
m
o
f
P
r
ec
is
io
n
.
W
e
test
ed
o
u
r
m
et
h
o
d
u
s
i
n
g
ex
p
er
i
m
en
ta
l
s
etu
p
b
ased
o
n
M
A
T
L
A
B
co
d
e
ap
p
lied
to
Gallica
d
atab
ase
RE
F
E
R
E
NC
E
S
[1
]
M
ih
a
lce
a
R.
,
Co
rley
C.
a
n
d
S
t
ra
p
p
a
ra
v
a
C.
,
"
Co
rp
u
s
-
b
a
se
d
a
n
d
k
n
o
w
led
g
e
-
b
a
se
d
m
e
a
su
re
s
o
f
tex
t
se
m
a
n
ti
c
sim
il
a
rit
y
,
"
In
AAAI
,
v
o
l
.
6
,
pp
.
7
7
5
-
7
8
0
,
2
0
0
6
.
[2
]
Hu
a
n
g
,
A
n
n
a
,
"
S
im
il
a
rit
y
m
e
a
s
u
r
e
s
f
o
r
tex
t
d
o
c
u
m
e
n
t
c
lu
ste
rin
g
,
"
In
:
Pro
c
e
e
d
in
g
s
o
f
th
e
six
th
n
e
w
ze
a
l
a
n
d
c
o
mp
u
ter
sc
ien
c
e
re
se
a
rc
h
stu
d
e
n
t
c
o
n
fer
e
n
c
e
(
NZ
CS
RS
C
2
0
0
8
)
,
C
h
ristch
u
rc
h
,
Ne
w
Zea
lan
d
,
p
p
.
9
-
56
,
2
0
0
8
.
[3
]
M
a
rti
,
U.
V
.
,
Bu
n
k
e
,
H
.
,
"
Us
in
g
a
sta
ti
stica
l
lan
g
u
a
g
e
m
o
d
e
l
to
im
p
ro
v
e
th
e
p
e
rf
o
rm
a
n
c
e
o
f
a
n
HMM
-
b
a
se
d
c
u
rsiv
e
h
a
n
d
w
rit
in
g
re
c
o
g
n
it
i
o
n
sy
ste
m
,
"
In
Hid
d
e
n
M
a
rk
o
v
m
o
d
e
ls:
a
p
p
li
c
a
ti
o
n
s i
n
c
o
mp
u
ter
v
isio
n
,
p
p
.
6
5
-
90,
2
0
0
1
.
[4
]
H.
Bu
n
k
e
,
e
t.
a
l.
,
"
Of
f
li
n
e
re
c
o
g
n
it
io
n
o
f
u
n
c
o
n
stra
i
n
e
d
h
a
n
d
w
rit
ten
tex
ts
u
sin
g
HMM
s
a
n
d
sta
ti
stica
l
lan
g
u
a
g
e
m
o
d
e
ls,"
in
IEE
E
T
r
a
n
sa
c
ti
o
n
s
o
n
Pa
tt
e
rn
A
n
a
lys
is
a
n
d
M
a
c
h
i
n
e
I
n
telli
g
e
n
c
e
, v
o
l.
2
6
,
n
o
.
6
,
p
p
.
7
0
9
-
7
2
0
,
2
0
0
4
.
[5
]
C.
Ba
h
lm
a
n
n
a
n
d
H.
Bu
rk
h
a
rd
t,
"
M
e
a
su
rin
g
HMM
sim
il
a
rit
y
w
it
h
th
e
Ba
y
e
s
p
ro
b
a
b
il
i
ty
o
f
e
rro
r
a
n
d
it
s
a
p
p
li
c
a
ti
o
n
to
o
n
li
n
e
h
a
n
d
w
rit
in
g
re
c
o
g
n
it
io
n
,
"
Pro
c
e
e
d
in
g
s
o
f
S
ixth
I
n
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
Do
c
u
me
n
t
An
a
lys
is
a
n
d
Rec
o
g
n
it
io
n
,
S
e
a
tt
le,
W
A
,
USA
,
p
p
.
4
0
6
-
411
,
2
0
0
1
.
[6
]
J.A
.
R
-
S
e
rra
n
o
,
F
.
P
e
rro
n
n
i
n
,
"
A
M
o
d
e
l
-
b
a
se
d
se
q
u
e
n
c
e
sim
il
a
rit
y
w
it
h
a
p
p
li
c
a
ti
o
n
t
o
h
a
n
d
w
rit
ten
w
o
rd
sp
o
tt
i
n
g
,
"
in
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
Pa
tt
e
rn
An
a
lys
is
a
n
d
M
a
c
h
i
n
e
In
tell
ig
e
n
c
e
,
v
o
l.
3
4
,
n
o
.
1
1
,
p
p
.
2
1
0
8
-
2
1
2
0
,
No
v
.
2
0
1
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
10
,
No
.
3
,
J
u
n
e
2
0
2
0
:
2
3
6
7
-
2374
2374
[7
]
Q.
W
a
n
g
,
F
.
Yin
a
n
d
C.
L
iu
,
"
H
a
n
d
w
rit
ten
Ch
in
e
se
T
e
x
t
Re
c
o
g
n
it
io
n
b
y
In
teg
ra
ti
n
g
M
u
lt
ip
le
Co
n
t
e
x
ts,"
in
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
Pa
tt
e
rn
A
n
a
lys
is
a
n
d
M
a
c
h
i
n
e
In
telli
g
e
n
c
e
,
v
o
l.
3
4
,
n
o
.
8
,
p
p
.
1
4
6
9
-
1
4
8
1
,
A
u
g
.
2
0
1
2
.
[8
]
Si
n
g
h
,
R
.
,
Ra
j,
B.
,
S
tern
,
R.
M
.
,
"
S
tru
c
tu
re
d
re
d
e
f
in
it
i
o
n
o
f
so
u
n
d
u
n
it
s
b
y
m
e
rg
in
g
a
n
d
sp
li
tt
i
n
g
f
o
r
im
p
ro
v
e
d
sp
e
e
c
h
re
c
o
g
n
it
io
n
,"
In
S
ixth
In
te
rn
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
S
p
o
k
e
n
L
a
n
g
u
a
g
e
Pr
o
c
e
ss
in
g
,
2
0
0
0
.
[9
]
Hu
o
,
Q.,
L
i,
W
.
,
"
A
DTW
-
b
a
s
e
d
d
issim
il
a
rit
y
m
e
a
su
re
f
o
r
le
f
t
-
to
-
rig
h
t
h
id
d
e
n
M
a
rk
o
v
m
o
d
e
ls an
d
it
s a
p
p
li
c
a
ti
o
n
t
o
w
o
rd
c
o
n
f
u
sa
b
il
it
y
a
n
a
l
y
sis,"
In
Nin
th
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
S
p
o
k
e
n
L
a
n
g
u
a
g
e
Pro
c
e
ss
in
g
,
2
0
0
6
.
[1
0
]
J.
He
r
sh
e
y
,
P
.
Olse
n
,
a
n
d
S
.
Re
n
n
ie,
"
V
a
riati
o
n
a
l
Ku
l
l
b
a
c
k
-
L
e
i
b
ler
Div
e
r
g
e
n
c
e
f
o
r
Hid
d
e
n
M
a
rk
o
v
M
o
d
e
ls,"
Pro
c
.
W
o
rk
sh
o
p
Au
to
ma
ti
c
S
p
e
e
c
h
Rec
o
g
n
it
i
o
n
Un
d
e
rs
ta
n
d
in
g
,
2
0
0
7
.
[1
1
]
M
.
Bra
n
d
a
n
d
V.
Ke
tt
n
a
k
e
r,
"
Disc
o
v
e
r
y
a
n
d
se
g
m
e
n
tatio
n
o
f
a
c
ti
v
it
ies
in
v
id
e
o
,
"
in
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
P
a
tt
e
rn
An
a
lys
is
a
n
d
M
a
c
h
i
n
e
In
tell
ig
e
n
c
e
,
v
o
l.
2
2
,
n
o
.
8
,
p
p
.
8
4
4
-
8
5
1
,
A
u
g
.
2
0
0
0
.
[1
2
]
W
il
lett,
P
.
e
t
a
l.
,
"
G
ro
u
p
i
n
g
o
f
c
o
e
ff
icie
n
ts
f
o
r
th
e
c
a
lcu
latio
n
o
f
in
ter
-
m
o
lec
u
lar
si
m
il
a
rit
y
a
n
d
d
issim
il
a
rit
y
u
sin
g
2
D f
ra
g
m
e
n
t
b
it
-
stri
n
g
s,"
Co
mb
in
a
to
ri
a
l
c
h
e
mistry
&
h
ig
h
th
ro
u
g
h
p
u
t
sc
re
e
n
in
g
,
v
o
l
.
5
,
n
o
.
2
,
p
p
.
1
5
5
-
1
6
6
,
2
0
0
2
.
[1
3
]
S
e
u
n
g
-
Se
o
k
C,
S
u
n
g
-
Hy
u
k
C
a
n
d
T
a
p
p
e
rt
C.
,
"
A
S
u
rv
e
y
o
f
Bin
a
ry
S
i
m
il
a
rit
y
a
n
d
Dista
n
c
e
M
e
a
su
re
s
,
J
o
u
r
n
a
l
o
f
S
y
ste
mic
s,
Cy
b
e
rn
e
ti
c
s
&
In
fo
rm
a
ti
c
s
,
v
o
l.
8
,
c
h
a
p
ter
1
,
pp
.
43
–
48
,
2
0
1
0
.
[1
4
]
Ch
a
S
.,
"
Co
m
p
re
h
e
n
siv
e
S
u
rv
e
y
o
n
Dis
tan
c
e
/
S
i
m
il
a
rit
y
M
e
a
su
re
s
b
e
twe
e
n
P
ro
b
a
b
il
it
y
De
n
sity
F
u
n
c
ti
o
n
s
,"
In
t.
J
.
M
a
th
.
M
o
d
e
l.
M
e
th
o
d
s
Ap
p
l.
S
c
i
,
v
o
l.
1
,
c
h
a
p
ter 4
,
pp
.
3
0
0
–
3
0
7
,
2
0
0
7
.
[1
5
]
Ch
a
n
g
,
W
.
L
.
,
Ka
n
e
sa
n
,
J
.
,
Ku
lk
a
rn
i,
A
.
J.,
Ra
m
iah
,
H.,
"
Da
ta
c
lu
ste
rin
g
u
si
n
g
se
e
d
d
isp
e
rse
r
a
n
t
a
lg
o
rit
h
m
,
"
T
u
rk
ish
J
o
u
rn
a
l
o
f
El
e
c
trica
l
En
g
in
e
e
rin
g
&
Co
mp
u
ter
S
c
ien
c
e
s
,
v
o
l.
2
5
,
n
o
.
6
,
p
p
.
4
5
2
2
-
4
5
3
2
,
2
0
1
7
.
[1
6
]
Orte
g
a
,
J.
P
.
,
De
l,
M
.
,
Ro
jas
,
R.
B.
,
S
o
m
o
d
e
v
il
la,
M
.
J.,
"
Re
se
a
rc
h
issu
e
s
o
n
k
-
m
e
a
n
s
a
lg
o
rit
h
m
:
A
n
e
x
p
e
ri
m
e
n
tal
tri
a
l
u
sin
g
m
a
tl
a
b
,
"
In
CEUR
W
o
r
k
sh
o
p
Pr
o
c
e
e
d
in
g
s: S
e
ma
n
ti
c
W
e
b
a
n
d
Ne
w
T
e
c
h
n
o
lo
g
ies
,
p
p
.
8
3
-
9
6
,
M
a
r.
2
0
0
9
.
[1
7
]
L
o
we
,
D.
G
.
,
"
Distin
c
ti
v
e
i
m
a
g
e
f
e
a
tu
re
s
f
ro
m
sc
a
le
-
in
v
a
rian
t
k
e
y
p
o
i
n
ts,"
In
ter
n
a
ti
o
n
a
l
jo
u
rn
a
l
o
f
c
o
mp
u
ter
v
isio
n
,
v
o
l.
60
,
n
o
.
2
,
p
p
.
91
-
1
1
0
,
2
0
0
4
.
[1
8
]
M
a
rti
,
U
-
V
.
,
B
u
n
k
e
,
H.,
"
Us
in
g
a
sta
ti
stica
l
lan
g
u
a
g
e
m
o
d
e
l
to
i
m
p
ro
v
e
th
e
p
e
rf
o
r
m
a
n
c
e
o
f
a
n
HM
M
-
b
a
se
d
c
u
rsiv
e
h
a
n
d
w
rit
in
g
re
c
o
g
n
it
i
o
n
sy
ste
m
,
"
In
t
J
P
a
tt
e
rn
Rec
o
g
n
i
t
Arti
f
In
tell
,
v
o
l.
1
5
,
n
o
.
0
1
,
p
p
.
65
-
90
,
2
0
0
1
.
[1
9
]
N
o
w
a
k
,
E.
,
Ju
rie,
F
.
,
T
rig
g
s,
B
.
,
"
S
a
m
p
li
n
g
stra
teg
ies
f
o
r
b
a
g
-
o
f
-
fe
a
tu
re
s
i
m
a
g
e
c
las
s
i
f
ic
a
ti
o
n
,
"
in
Eu
r
o
p
e
a
n
Co
n
fer
e
n
c
e
o
n
Co
m
p
u
ter
Vi
si
o
n
,
L
e
c
tu
re
No
tes
in
C
o
mp
u
ter
S
c
ien
c
e
(
L
NCS
)
3
9
5
4
,
p
p
.
4
9
0
-
5
0
3
,
2
0
0
6
.
[2
0
]
Ern
e
st Bellm
a
n
,
R.
,
"
D
y
n
a
m
ic p
ro
g
ra
m
m
in
g
,
”
P
rin
c
e
to
n
U
n
iv
e
rsity
P
re
ss
,
Ra
n
d
Co
r
p
o
ra
ti
o
n
,
1
9
5
7
,
[2
1
]
Ern
e
st Bellm
a
n
,
R.
,
"
A
d
a
p
ti
v
e
c
o
n
tro
l
p
r
o
c
e
ss
e
s: a g
u
id
e
d
to
u
r
,
"
P
r
in
c
e
to
n
U
n
iv
e
rsity
P
re
ss
.
[2
2
]
S
a
n
k
a
r,
K.
P
.
,
M
a
n
m
a
th
a
,
R.
,
Ja
wa
h
a
r,
C.
V
.
,
"
L
a
rg
e
-
sc
a
l
e
d
o
c
u
m
e
n
t
ima
g
e
re
tri
e
v
a
l
b
y
a
u
to
m
a
ti
c
w
o
rd
a
n
n
o
tati
o
n
,
"
In
ter
n
a
t
io
n
a
l
J
o
u
rn
a
l
o
n
Do
c
u
me
n
t
A
n
a
lys
is a
n
d
Rec
o
g
n
it
i
o
n
(
IJ
D
AR
)
,
v
o
l.
1
7
,
n
o
.
1
,
p
p
.
1
-
17
,
2
0
1
3
.
[2
3
]
A
l
m
a
z
á
n
,
J.,
G
o
rd
o
,
A
.
,
F
o
rn
é
s
,
A
.
,
e
t
a
l
.,
"
S
e
g
m
e
n
tatio
n
-
f
re
e
w
o
rd
sp
o
tt
i
n
g
w
it
h
e
x
e
m
p
lar
S
V
M
s
,
"
P
a
tt
e
r
n
Rec
o
g
n
it
io
n
,
v
o
l.
4
7
,
n
o
.
1
2
,
p
p
.
3
9
6
7
-
3
9
7
8
,
2
0
1
4
.
[2
4
]
Ru
siñ
o
l,
M
.
,
A
ld
a
v
e
rt,
D.,
L
lad
o
s,
R.
J.,
"
Eff
icie
n
t
se
g
m
e
n
tatio
n
-
f
re
e
k
e
y
w
o
rd
sp
o
tt
i
n
g
in
h
ist
o
rica
l
d
o
c
u
m
e
n
t
c
o
ll
e
c
ti
o
n
s
,
"
P
a
tt
e
rn
Rec
o
g
n
i
ti
o
n
,
v
o
l.
4
8
,
p
p
.
5
4
5
-
5
5
5
,
2
0
1
5
.
[2
5
]
H
o
w
e
,
H.,
Ra
th
,
T
.
,
R
a
n
m
a
th
a
,
R.
,
"
Bo
o
ste
d
d
e
c
isio
n
tree
s
fo
r
w
o
rd
re
c
o
g
n
it
io
n
i
n
h
a
n
d
w
rit
ten
d
o
c
u
m
e
n
t
re
tri
e
v
a
l,
"
in
Pro
c
e
e
d
in
g
s
o
f
th
e
An
n
u
a
l
I
n
ter
n
a
ti
o
n
a
l
ACM
S
IGI
R
Co
n
fer
e
n
c
e
o
n
Res
e
a
rc
h
a
n
d
De
v
e
lo
p
me
n
t
in
In
fo
rm
a
t
io
n
Retrie
v
a
l
,
p
p
.
3
7
7
-
3
8
3
,
2
0
0
5
.
[2
6
]
L
ian
g
,
Y.,
F
a
irh
u
rst,
M
.
,
G
u
e
st,
R.
.
,
"
A
s
y
n
th
e
siz
e
d
w
o
rd
a
p
p
r
o
a
c
h
to
w
o
rd
re
t
riev
a
l
in
h
a
n
d
w
rit
t
e
n
d
o
c
u
m
e
n
ts
,
"
Pa
tt
e
rn
Rec
o
g
n
it
,
v
o
l
.
4
5
,
n
o
.
1
2
,
p
p
.
4
2
2
5
-
4
2
3
6
,
2
0
1
2
.
[2
7
]
F
isc
h
e
r,
A
.
,
Ke
ll
e
r,
A
.
,
F
rin
k
e
n
,
V
.
,
e
t
a
l
.
,
"
L
e
x
ico
n
-
f
re
e
h
a
n
d
w
rit
ten
w
o
r
d
sp
o
tt
i
n
g
u
sin
g
c
h
a
ra
c
ter
HMM
s
,
"
Pa
tt
e
rn
Rec
o
g
n
it
.
L
e
tt
,
v
o
l.
3
3
,
n
o
.
7
,
p
p
.
9
3
4
-
9
4
2
,
2
0
1
2
.
[2
8
]
T
e
ra
sa
w
a
,
K.,
Tan
a
k
a
,
y
.
,
"
S
li
t
st
y
le
HO
G
fe
a
tu
re
f
o
r
d
o
c
u
m
e
n
t
im
a
g
e
w
o
rd
sp
o
tt
i
n
g
,
"
in
P
ro
c
e
e
d
in
g
o
f
th
e
In
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
Do
c
u
me
n
t
A
n
a
lys
is
a
n
d
Rec
o
g
n
it
i
o
n
,
p
p
.
1
1
6
-
1
2
0
,
2
0
0
9
.
Evaluation Warning : The document was created with Spire.PDF for Python.