T
E
L
K
O
M
N
I
K
A
T
elec
o
m
m
un
ica
t
io
n,
Co
m
pu
t
ing
,
E
lect
ro
nics
a
nd
Co
ntr
o
l
Vo
l.
19
,
No
.
2
,
A
p
r
il
2
0
2
1
,
p
p
.
471
~
47
8
I
SS
N:
1
6
9
3
-
6
9
3
0
,
ac
cr
ed
ited
First Gr
ad
e
b
y
Kem
en
r
is
tek
d
i
k
ti,
Dec
r
ee
No
: 2
1
/E/KPT
/2
0
1
8
DOI
: 1
0
.
1
2
9
2
8
/TE
L
KOM
NI
K
A.
v
1
9
i2
.
1
8
1
4
9
471
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//jo
u
r
n
a
l.u
a
d
.
a
c.
id
/in
d
ex
.
p
h
p
/TELK
OM
N
I
K
A
Ado
pting e
x
plici
t
a
nd implicit
so
cia
l relatio
ns by
S
V
D++ for
recom
menda
tion
sy
stem impro
v
em
ent
M
o
hs
i
n H
a
s
a
n H
us
s
ei
n,
Ak
ee
l A
bd
ulk
a
re
em
Als
a
k
a
a
,
H
a
y
da
r
A.
M
a
rho
o
n
De
p
a
rtme
n
t
o
f
Co
m
p
u
ter
S
c
ien
c
e
,
Co
ll
e
g
e
o
f
S
c
ien
c
e
,
Un
iv
e
rsit
y
o
f
Ke
rb
a
la,
Ira
q
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
u
l 1
8
,
2
0
2
0
R
ev
is
ed
Oct
1
,
2
0
2
0
Acc
ep
ted
Oct
1
6
,
2
0
2
0
Re
c
o
m
m
e
n
d
e
r
sy
ste
m
s
su
ffe
r
a
se
t
o
f
d
ra
wb
a
c
k
s
su
c
h
a
s
sp
a
rsi
ty
.
S
o
c
ial
re
latio
n
s
p
r
o
v
i
d
e
a
u
se
fu
l
so
u
rc
e
to
o
v
e
rc
o
m
e
th
e
sp
a
rsity
p
ro
b
lem
.
P
re
v
io
u
s
stu
d
ies
h
a
v
e
u
ti
li
z
e
d
s
o
c
ial
re
latio
n
s
o
r
ra
ti
n
g
fe
e
d
b
a
c
k
so
u
rc
e
s.
Ho
we
v
e
r,
th
e
y
ig
n
o
re
d
in
teg
ra
ti
n
g
t
h
e
se
s
o
u
rc
e
s.
In
th
is
p
a
p
e
r,
t
h
e
li
m
it
a
ti
o
n
s
o
f
p
re
v
io
u
s
stu
d
ies
a
re
o
v
e
rc
o
m
e
b
y
e
x
p
l
o
i
ti
n
g
fo
u
r
so
u
rc
e
s
o
f
i
n
fo
rm
a
ti
o
n
,
n
a
m
e
ly
:
e
x
p
li
c
it
so
c
ial
re
latio
n
sh
i
p
s,
imp
l
icit
so
c
ial
re
latio
n
sh
i
p
s,
u
se
rs’
ra
ti
n
g
s,
a
n
d
imp
li
c
it
fe
e
d
b
a
c
k
i
n
fo
rm
a
ti
o
n
.
F
irstl
y
,
im
p
li
c
it
so
c
ial
re
latio
n
sh
ip
s
a
re
e
x
trac
ted
th
r
o
u
g
h
t
h
e
so
u
rc
e
a
ll
o
c
a
ti
o
n
in
d
e
x
a
lg
o
rit
h
m
to
e
sta
b
li
sh
n
e
w
re
latio
n
s
a
m
o
n
g
u
se
rs.
S
e
c
o
n
d
l
y
,
th
e
sim
il
a
rit
y
m
e
th
o
d
is
a
p
p
li
e
d
to
fin
d
t
h
e
sim
il
a
rit
y
b
e
twe
e
n
e
a
c
h
p
a
ir
o
f
u
se
rs
wh
o
h
a
v
e
e
x
p
l
icit
o
r
im
p
l
icit
so
c
ial
re
latio
n
s.
T
h
e
n
,
u
se
rs’
ra
ti
n
g
s
a
n
d
imp
li
c
it
ra
ti
n
g
fe
e
d
b
a
c
k
s
o
u
rc
e
s
a
re
e
x
trac
ted
v
ia
a
u
se
r
-
it
e
m
m
a
tri
x
.
F
u
rth
e
rm
o
re
,
a
ll
so
u
rc
e
s
a
re
in
teg
ra
ted
i
n
t
o
th
e
sin
g
u
lar
v
a
lu
e
d
e
c
o
m
p
o
siti
o
n
p
lu
s
(
S
VD
+
+
)
m
e
th
o
d
.
F
i
n
a
ll
y
,
m
issin
g
p
re
d
ictio
n
s
a
re
c
o
m
p
u
ted
.
T
h
e
p
r
o
p
o
se
d
m
e
th
o
d
is
imp
lem
e
n
ted
o
n
th
re
e
re
a
l
-
wo
rld
d
a
tas
e
ts:
Las
t.
F
m
,
F
il
m
Tru
st,
a
n
d
Ciao
.
E
x
p
e
rime
n
tal
re
su
lt
s
re
v
e
a
l
th
a
t
th
e
p
ro
p
o
se
d
m
o
d
e
l
is
su
p
e
rio
r
t
o
o
th
e
r
st
u
d
ies
su
c
h
a
s
S
VD
,
S
VD
+
+
,
EU
-
S
VD
+
+
,
S
o
c
Re
g
,
a
n
d
EI
S
R
in
te
rm
s
o
f
a
c
c
u
ra
c
y
,
wh
e
re
t
h
e
im
p
r
o
v
e
m
e
n
t
o
f
th
e
p
r
o
p
o
se
d
m
e
th
o
d
is
a
b
o
u
t
0
.
0
3
%
fo
r
M
AE
a
n
d
0
.
0
1
%
f
o
r
R
M
S
E
wh
e
n
d
ime
n
sio
n
v
a
l
u
e
(
d
)
=
1
0
.
K
ey
w
o
r
d
s
:
R
ec
o
m
m
en
d
atio
n
s
y
s
tem
So
cial
R
elatio
n
Sp
ar
s
ity
SVD++
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r
th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Mo
h
s
in
Hasan
Hu
s
s
ein
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
Scie
n
ce
Un
iv
er
s
ity
o
f
Ker
b
ala
Ker
b
ala
5
6
0
0
1
,
I
r
aq
E
m
ail:
m
o
h
s
in
.
h
@
u
o
k
e
r
b
ala.
e
d
u
.
iq
1.
I
NT
RO
D
UCT
I
O
N
Du
e
to
th
e
ex
p
o
n
en
tial
g
r
o
wth
o
f
I
n
ter
n
et
e
-
c
o
m
m
er
ce
,
th
e
in
f
o
r
m
atio
n
p
r
o
v
id
ed
b
y
th
e
I
n
ter
n
et
h
as
in
cr
ea
s
ed
an
d
th
is
p
h
en
o
m
en
o
n
is
ca
lled
in
f
o
r
m
atio
n
o
v
er
f
lo
w.
Acc
o
r
d
in
g
ly
,
f
i
n
d
in
g
p
r
ef
er
r
ed
p
r
o
d
u
cts
o
n
web
s
ites
h
as
b
ec
o
m
e
a
p
o
n
d
e
r
o
u
s
task
[
1
,
2
]
.
T
h
er
e
f
o
r
e,
a
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
is
u
tili
ze
d
to
s
o
lv
e
th
e
af
o
r
e
m
en
tio
n
ed
p
r
o
b
lem
b
y
r
ed
u
cin
g
s
ea
r
ch
tim
es
an
d
p
r
ed
ictin
g
a
u
s
er
’
s
p
r
e
f
er
en
c
es
o
f
p
r
o
d
u
cts.
A
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
(
R
S)
ca
n
b
e
d
e
f
in
ed
as
a
to
o
l
th
at
ca
n
r
ec
o
m
m
en
d
a
lis
t
o
f
i
tem
s
to
in
ter
ested
u
s
er
s
[
3
]
.
A
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
co
m
es
in
a
v
a
r
iety
o
f
ty
p
es,
co
llab
o
r
ativ
e
f
ilter
in
g
b
ein
g
th
e
m
o
s
t
co
m
m
o
n
an
d
m
o
s
t
wid
ely
u
s
ed
[
4
,
5
]
.
C
o
llab
o
r
ativ
e
f
ilter
in
g
esti
m
ates
u
n
k
n
o
wn
r
atin
g
s
b
ased
o
n
u
s
er
h
is
to
r
y
r
atin
g
s
an
d
b
eh
av
i
o
r
s
o
f
s
im
ilar
u
s
er
s
.
Su
ch
s
y
s
tem
s
ar
e
ca
l
led
u
s
er
-
o
r
ien
ted
.
An
item
-
o
r
ie
n
ted
ap
p
r
o
ac
h
was
p
r
o
p
o
s
ed
by
[
6
]
wh
er
e
p
r
ed
ictio
n
s
ar
e
g
e
n
er
ated
o
n
t
h
e
b
asis
o
f
s
im
ilar
ities
b
etwe
en
item
s
.
Ho
wev
er
,
m
o
s
t
r
ec
en
t
d
atasets
ar
e
r
ath
er
h
u
g
e
an
d
h
av
e
co
m
p
r
is
ed
th
o
u
s
an
d
s
o
f
u
s
er
s
an
d
item
s
,
in
clu
d
in
g
Netf
lix
,
Face
b
o
o
k
,
an
d
Yo
u
T
u
b
e
,
s
o
co
m
p
u
tin
g
s
im
ilar
ity
b
ec
o
m
es
co
s
tly
.
T
h
er
e
ar
e
two
w
ay
s
to
im
p
lem
en
t
th
is
task
:
f
ir
s
t
,
by
f
in
d
in
g
th
e
s
im
ilar
ity
b
etwe
en
u
s
er
s
o
r
ite
m
s
,
an
d
s
ec
o
n
d
b
y
f
in
d
in
g
th
e
laten
t
f
ac
to
r
s
m
o
d
el
o
f
u
s
er
s
an
d
item
s
w
h
ich
ca
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6
9
3
0
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
,
Vo
l.
19
,
No
.
2
,
Ap
r
il 2
0
2
1
:
4
7
1
-
47
8
472
b
e
im
p
lem
en
ted
b
y
co
n
v
er
tin
g
b
o
th
u
s
er
s
an
d
item
s
to
th
e
s
a
m
e
laten
t
f
ac
to
r
f
o
llo
wed
b
y
c
o
m
p
ar
is
o
n
s
b
etwe
en
th
em
[
7
]
.
Ma
tr
ix
f
ac
to
r
izatio
n
is
a
co
m
m
o
n
tech
n
i
q
u
e
o
f
ex
tr
ac
tin
g
l
aten
t
f
ea
tu
r
es
f
r
o
m
a
u
s
er
-
ite
m
m
atr
ix
.
Sev
er
al
m
eth
o
d
s
h
a
v
e
b
ee
n
a
p
p
li
ed
in
th
is
tech
n
iq
u
e
s
u
ch
as
SVD
[
7
]
,
PMF
[
8
]
,
SVD++
[
9
]
,
n
o
n
n
eg
ativ
e
m
atr
ix
f
ac
to
r
izatio
n
[
1
0
-
1
2
]
.
A
R
ec
o
m
m
en
d
atio
n
s
y
s
tem
ai
m
s
to
p
r
ed
ict
item
s
ac
c
u
r
ately
s
o
th
at
u
s
er
s
ca
n
f
in
d
th
eir
p
r
e
f
er
en
ce
s
am
o
n
g
th
o
u
s
an
d
s
o
f
p
r
o
d
u
cts.
Ho
wev
er
,
th
e
in
cr
ea
s
in
g
s
ize
o
f
th
e
in
f
o
r
m
atio
n
p
r
esen
t
s
o
b
s
tacle
s
to
im
p
lem
e
n
tin
g
an
id
ea
l
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
.
I
n
a
r
ea
l
d
ata
s
et,
th
e
s
p
a
r
s
ity
r
atio
is
s
o
h
ig
h
d
u
e
to
u
s
er
s
leav
e
m
o
s
t
item
s
with
o
u
t
r
atin
g
.
M
o
r
eo
v
e
r
,
c
o
m
p
u
tin
g
p
r
ed
ictio
n
v
alu
es
d
ep
en
d
i
n
g
o
n
u
s
er
s
’
s
im
ilar
ities
wo
u
ld
b
e
f
u
tile.
T
h
er
ef
o
r
e
,
m
atr
ix
f
ac
to
r
izatio
n
m
eth
o
d
s
ar
e
a
s
u
itab
le
c
h
o
ice
to
o
v
er
c
o
m
e
s
u
ch
lim
itati
o
n
s
.
Alth
o
u
g
h
m
atr
ix
f
ac
to
r
izatio
n
tech
n
iq
u
es wo
r
k
well
with
s
p
ar
s
e
d
ata,
th
ey
s
til
l n
ee
d
im
p
r
o
v
em
en
t
to
in
cr
ea
s
e
p
r
ed
ictio
n
ac
cu
r
ac
y
.
C
o
n
s
eq
u
en
tly
,
s
o
cial
in
f
o
r
m
at
io
n
h
as
r
ec
en
tly
s
p
r
ea
d
r
a
p
id
l
y
an
d
m
a
n
y
ap
p
licatio
n
s
n
o
w
u
s
e
s
o
cial
r
elatio
n
s
[
1
3
,
1
4
]
.
A
s
o
cial
n
etwo
r
k
p
r
o
d
u
ce
s
n
ew
f
ac
to
r
s
f
o
r
f
u
r
th
er
p
r
o
m
o
tin
g
th
e
p
er
f
o
r
m
an
ce
o
f
th
e
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
[
1
5
,
1
6
]
.
I
n
r
ea
lity
,
u
s
er
s
ar
e
o
f
ten
i
n
f
lu
en
ce
d
b
y
th
eir
f
r
ien
d
s
’
ch
o
ices,
s
o
f
o
r
ex
am
p
le
wh
en
th
ey
tr
y
to
d
ec
id
e
o
n
a
s
p
ec
if
ic
m
o
v
ie
to
watc
h
,
th
e
y
ch
ec
k
th
o
s
e
m
o
v
ies
th
at
ar
e
alr
ea
d
y
r
ated
b
y
th
eir
f
r
ien
d
s
.
So
cial
r
elatio
n
s
h
ip
s
a
r
e
th
e
m
ai
n
f
ac
to
r
h
er
e
an
d
e
x
p
lo
itin
g
e
x
p
licit
r
elatio
n
s
h
ip
s
ca
n
en
h
a
n
ce
th
e
p
r
ed
ictio
n
p
r
o
ce
s
s
.
Ma
n
y
s
tu
d
ies
h
av
e
b
ee
n
p
er
f
o
r
m
ed
in
th
is
ar
ea
,
s
u
ch
as
[
1
7
-
2
0
]
.
I
n
[
1
7
]
p
r
o
p
o
s
ed
two
m
o
d
els,
n
am
ely
SR
1
an
d
SR
2
.
I
n
th
at
s
tu
d
y
th
e
au
th
o
r
s
in
te
g
r
ated
th
e
m
atr
i
x
f
ac
to
r
izatio
n
m
eth
o
d
a
n
d
s
o
cial
r
elatio
n
s
.
B
o
th
m
o
d
els
wer
e
i
m
p
lem
en
ted
b
y
co
m
p
u
tin
g
th
e
s
im
ilar
ity
b
etwe
en
a
tar
g
et
u
s
er
an
d
a
s
et
o
f
th
at
u
s
er
’
s
f
r
ien
d
s
.
T
h
e
r
esu
lts
en
h
an
ce
d
th
e
ac
cu
r
ac
y
o
f
th
e
p
r
ed
ictio
n
.
In
[
1
8
]
p
r
o
p
o
s
ed
a
m
o
d
el
th
at
d
ep
e
n
d
ed
o
n
co
m
m
u
n
ity
d
etec
tio
n
to
allev
iate
th
e
co
ld
-
s
tar
t
p
r
o
b
lem
.
Ad
d
itio
n
ally
,
u
s
er
s
’
p
r
ef
er
e
n
ce
s
m
ay
b
e
in
f
lu
en
ce
d
b
y
an
o
t
h
er
k
in
d
o
f
r
elatio
n
s
h
ip
k
n
o
wn
as
“f
r
ie
n
d
s
o
f
f
r
ien
d
s
”
o
r
im
p
lici
t
r
elatio
n
s
h
ip
s
.
R
esear
ch
[
1
9
]
i
n
tr
o
d
u
ce
d
a
n
ew
s
y
s
tem
t
h
at
ex
p
lo
ited
th
e
im
p
licit
r
elatio
n
s
h
ip
o
f
u
s
er
s
.
I
n
th
at
s
tu
d
y
,
ex
p
licit
r
elatio
n
s
h
ip
s
wer
e
ig
n
o
r
ed
.
In
[
2
0
]
,
ex
p
licit
an
d
im
p
licit
r
elatio
n
s
h
ip
s
wer
e
u
tili
ze
d
wit
h
an
d
in
co
r
p
o
r
ated
with
th
e
u
s
er
-
ite
m
m
atr
ix
in
to
th
e
p
r
o
b
ab
ilis
tic
m
atr
ix
f
ac
to
r
i
za
tio
n
(
PMF)
m
eth
o
d
.
Alth
o
u
g
h
th
e
p
r
ev
io
u
s
s
tu
d
ies
en
h
an
ce
d
th
e
R
S
p
er
f
o
r
m
a
n
ce
,
s
till
th
er
e
is
a
s
h
o
r
tag
e
in
ter
m
s
o
f
p
r
ed
ictio
n
a
cc
u
r
ac
y
.
As
m
en
ti
o
n
ed
i
n
[
2
0
]
,
th
e
au
t
h
o
r
s
p
r
o
p
o
s
ed
th
e
E
I
SR
m
o
d
el,
w
h
ich
ex
p
l
o
i
ted
im
p
licit
s
o
cial
r
elatio
n
s
,
ex
p
licit
s
o
cial
r
elatio
n
s
a
n
d
ex
p
licit
u
s
er
s
’
r
a
tin
g
s
.
I
n
th
at
s
tu
d
y
,
th
ey
o
v
er
c
am
e
th
e
d
r
awb
ac
k
s
of
[
1
7
,
2
1
]
b
y
u
s
in
g
im
p
licit
s
o
cial
r
elatio
n
s
as
an
ex
tr
a
s
o
u
r
ce
,
b
u
t
th
ey
o
v
e
r
lo
o
k
e
d
th
e
im
p
licit
f
ee
d
b
ac
k
in
f
o
r
m
atio
n
th
at
i
s
av
aila
b
le
in
th
e
u
s
er
-
item
m
atr
ix
.
I
m
p
licit
f
ee
d
b
ac
k
in
f
o
r
m
atio
n
is
v
ital
a
s
it
ca
n
r
ed
u
ce
th
e
s
p
ar
s
ity
r
atio
,
wh
ich
is
co
n
s
id
er
ed
a
m
ajo
r
p
r
o
b
lem
in
R
S
.
I
n
th
is
p
ap
er
,
t
h
e
lim
itatio
n
o
f
[
2
0
]
is
o
v
er
c
o
m
e
an
d
th
e
s
p
ar
s
ity
p
r
o
b
lem
is
allev
iated
b
y
ex
p
lo
itin
g
th
e
im
p
licit
f
ee
d
b
ac
k
in
f
o
r
m
atio
n
s
o
u
r
ce
with
th
e
o
th
er
th
r
ee
s
o
u
r
ce
s
th
at
ar
e
alr
ea
d
y
u
s
ed
i
n
[
2
0
]
.
A
ll
s
o
u
r
ce
s
ar
e
in
co
r
p
o
r
ated
in
to
th
e
SVD++
m
eth
o
d
(
r
ath
er
th
a
n
PMF
m
eth
o
d
)
.
Acc
o
r
d
in
g
ly
,
th
e
co
n
tr
ib
u
tio
n
s
o
f
t
h
is
p
ap
er
in
cl
u
d
e;
f
ir
s
t,
e
x
p
lo
itin
g
f
o
u
r
s
o
u
r
ce
s
o
f
i
n
f
o
r
m
atio
n
n
am
ely
:
ex
p
licit
s
o
cial
r
elatio
n
s
,
im
p
licit
s
o
cial
r
elatio
n
s
,
e
x
p
licit
r
atin
g
,
an
d
im
p
licit
f
ee
d
b
ac
k
.
Seco
n
d
,
th
e
s
o
u
r
ce
s
o
f
in
f
o
r
m
atio
n
ar
e
in
c
o
r
p
o
r
ate
d
in
to
th
e
SVD++
m
eth
o
d
.
T
h
ir
d
,
th
e
p
r
o
p
o
s
ed
m
eth
o
d
is
ev
alu
ated
v
i
a
th
r
ee
r
ea
l
-
d
atasets
:
C
iao
,
Fil
m
T
r
u
s
t,
an
d
L
ast.Fm
.
T
h
e
r
em
ai
n
d
er
o
f
th
is
p
a
p
er
i
n
clu
d
es
r
ela
ted
wo
r
k
in
th
e
n
ex
t
s
ec
tio
n
an
d
th
e
p
r
o
p
o
s
ed
m
eth
o
d
is
ex
p
lain
ed
in
th
e
th
ir
d
s
ec
tio
n
f
o
llo
wed
b
y
th
e
r
esu
lts
an
d
an
aly
s
is
.
Fin
ally
,
th
e
co
n
clu
s
io
n
s
o
f
th
e
s
tu
d
y
ar
e
p
r
esen
ted
in
t
h
e
last
s
ec
tio
n
.
2.
RE
S
E
ARCH
M
E
T
H
O
D
T
h
e
b
ac
k
b
o
n
e
o
f
t
h
e
p
r
o
p
o
s
e
d
m
eth
o
d
is
th
e
SVD++
m
eth
o
d
.
SVD++
is
an
ex
ten
s
io
n
o
f
th
e
SVD
p
r
o
p
o
s
ed
b
y
[
9
]
.
SVD++
e
x
p
l
o
its
ex
tr
a
in
f
o
r
m
atio
n
ca
lled
i
m
p
licit
f
ee
d
b
a
c
k
as
well
as
th
e
ex
p
licit
r
atin
g
.
I
n
th
is
s
tu
d
y
f
o
u
r
s
o
u
r
ce
s
o
f
i
n
f
o
r
m
atio
n
ar
e
a
d
o
p
te
d
:
ex
p
licit
s
o
cial
r
elatio
n
s
,
im
p
licit
s
o
cial
r
elatio
n
s
,
r
atin
g
v
alu
es,
an
d
im
p
licit
f
ee
d
b
ac
k
r
atin
g
.
T
h
e
u
s
er
s
’
r
atin
g
s
ar
e
d
iv
id
ed
in
to
two
p
a
r
ts
:
tr
ain
in
g
an
d
test
in
g
.
T
h
e
test
in
g
p
ar
t
is
u
s
ed
in
th
e
ev
a
lu
atio
n
p
r
o
ce
s
s
,
wh
ile
th
e
tr
ain
in
g
p
a
r
t
is
u
tili
ze
d
to
im
p
le
m
en
t
th
e
p
r
o
p
o
s
e
d
m
eth
o
d
.
Fig
u
r
e
1
s
h
o
ws a
s
u
m
m
ar
y
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
.
Af
ter
d
eter
m
in
in
g
th
e
tr
ai
n
in
g
s
et,
ex
p
licit
u
s
er
s
’
r
atin
g
s
(
ex
p
licit
f
ee
d
b
ac
k
)
is
cr
ea
ted
in
a
u
s
er
-
item
m
atr
ix
.
I
m
p
licit
f
ee
d
b
ac
k
ca
n
b
e
o
b
tain
ed
f
r
o
m
u
s
er
s
’
r
atin
g
s
r
eg
ar
d
less
o
f
wh
eth
er
an
ite
m
is
o
f
in
ter
est
to
th
e
u
s
er
.
Acc
o
r
d
in
g
ly
,
an
o
th
er
m
atr
ix
wi
ll
b
e
cr
ea
ted
with
th
e
s
am
e
s
ize
as
th
e
p
r
ev
io
u
s
u
s
er
-
item
m
atr
ix
;
th
is
m
atr
ix
will
b
e
a
b
in
a
r
y
m
atr
ix
.
T
h
e
o
n
e
v
alu
e
r
ef
er
s
to
th
e
item
r
ated
b
y
a
u
s
er
an
d
ze
r
o
s
f
o
r
u
n
r
ated
item
s
.
I
n
[
2
2
]
a
s
s
er
ts
th
at
ev
en
if
u
s
er
s
d
o
n
o
t
lik
e
an
item
,
th
er
e
is
s
till
an
o
p
p
o
r
tu
n
ity
f
o
r
th
em
to
lik
e
an
item
s
im
ilar
to
th
e
p
r
e
v
io
u
s
o
n
e
.
Seco
n
d
,
t
h
e
im
p
licit
s
o
cial
r
elatio
n
s
h
ip
s
ar
e
ex
tr
ac
ted
f
r
o
m
h
id
d
e
n
in
f
o
r
m
atio
n
in
s
o
cial
n
etwo
r
k
s
,
wh
ic
h
m
a
y
c
o
n
tain
v
ital
in
f
o
r
m
atio
n
.
T
h
e
im
p
lici
t
r
elatio
n
s
h
ip
s
a
r
e
also
ca
lled
Frie
n
d
s
o
f
Frie
n
d
s
.
C
o
n
s
eq
u
en
tly
,
th
e
lin
k
p
r
ed
ict
io
n
tech
n
iq
u
e
u
s
es
th
e
r
eso
u
r
c
e
allo
ca
tio
n
in
d
ex
(
R
AI
)
alg
o
r
ith
m
to
d
er
iv
e
th
e
h
id
d
en
in
f
o
r
m
atio
n
u
s
in
g
as
s
h
o
wn
in
(
1
)
.
Fo
r
ex
am
p
le,
f
o
r
g
r
ap
h
G
(
V
,
E
)
,
V
d
en
o
tes
th
e
u
s
er
s
in
th
e
s
o
cial
n
etwo
r
k
a
n
d
E
r
ef
e
r
s
to
t
h
e
ex
p
licit
r
elatio
n
s
h
ip
b
etwe
en
th
e
u
s
er
s
.
L
et
(
x,
y
)
b
e
two
u
n
f
r
ien
d
ed
u
s
er
s
;
R
AI
will
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
A
d
o
p
tin
g
ex
p
licit a
n
d
imp
licit so
cia
l rela
tio
n
s
b
y
S
V
D++
fo
r
r
ec
o
mme
n
d
a
tio
n
.
.
.
(
Mo
h
s
in
Ha
s
a
n
Hu
s
s
ein
)
473
co
m
p
u
te
th
e
p
r
o
b
ab
ilit
y
o
f
x
an
d
y
b
ein
g
f
r
ien
d
s
b
y
u
s
in
g
th
e
co
m
m
o
n
u
s
er
s
b
etwe
en
th
em
.
I
f
th
e
r
esu
lt
is
g
r
ea
ter
th
an
a
ce
r
tain
th
r
esh
o
l
d
,
x
an
d
y
ar
e
d
ee
m
ed
to
b
e
im
p
l
icit f
r
ien
d
s
.
(
,
)
=
∑
1
|
ᴦ
(
)
|
∈
|
ᴦ
(
)
⋂
ᴦ
(
)
|
(
1
)
W
h
er
e
ᴦ
(
x
)
an
d
ᴦ
(
y
)
r
ef
er
to
t
h
e
s
ets
o
f
f
r
ien
d
s
o
f
x
an
d
y
r
esp
ec
tiv
ely
.
Af
ter
ex
tr
ac
t
in
g
im
p
licit
s
o
cial
r
elatio
n
s
,
a
u
s
er
-
u
s
er
m
atr
i
x
is
cr
ea
ted
.
T
h
en
th
e
s
im
ilar
ity
b
etwe
en
ea
c
h
p
air
o
f
u
s
er
s
wh
o
h
av
e
e
x
p
lic
it
or
im
p
licit
s
o
cial
r
elatio
n
s
is
co
m
p
u
ted
u
s
in
g
th
e
Pear
s
o
n
co
r
r
elatio
n
co
ef
f
icien
t
as f
o
llo
ws
:
(
,
)
=
∑
(
,
ℎ
−
̅
̅
̅
)
(
,
ℎ
−
̅
̅
̅
)
ℎ
=
1
√
∑
(
,
ℎ
−
̅
̅
̅
)
2
ℎ
=
1
√
∑
(
,
ℎ
−
̅
̅
̅
)
2
ℎ
=
1
(
2
)
,
is
th
e
r
atin
g
v
alu
e
o
f
item
j
th
at
is
r
ated
b
y
u
s
er
u
.
̅
r
ef
er
s
to
th
e
av
er
ag
e
r
atin
g
o
f
t
h
e
u
s
er
x
,
an
d
n
im
p
lies
th
e
co
m
m
o
n
item
s
b
etwe
en
u
s
er
s
u
an
d
v
.
T
h
e
s
im
ilar
ity
r
an
g
es
b
etwe
en
[
-
1
,
1
]
;
1
m
ea
n
s
m
atch
in
g
b
etwe
en
u
s
er
s
an
d
-
1
in
d
icate
s
co
m
p
letely
d
is
s
im
ilar
u
s
er
s
.
T
h
e
f
o
u
r
s
o
u
r
ce
s
ar
e
in
co
r
p
o
r
at
ed
in
to
th
e
SVD++
alg
o
r
ith
m
.
T
h
e
SVD++
p
r
ed
ic
tio
n
f
o
r
m
u
la
d
e
p
en
d
s
o
n
th
e
b
aselin
e,
as f
o
llo
w
s
:
,
=
+
+
+
(
+
1
√
|
|
∑
∈
)
(
3
)
∑
∈
=
∑
(
,
>
0
)
∈
(
)
|
|
(
4
)
Her
e,
is
th
e
av
er
a
g
e
r
atin
g
o
f
th
e
en
tire
d
ataset.
an
d
ar
e
t
h
e
m
ea
n
v
alu
es
o
f
u
s
er
u
an
d
item
i
r
esp
ec
tiv
ely
,
th
at
ar
e
d
er
iv
ed
f
r
o
m
th
e
o
b
s
er
v
ed
d
ev
iatio
n
s
.
is
th
e
lef
t
o
r
th
o
g
o
n
al
v
al
u
es
o
f
th
e
u
s
er
-
item
m
atr
ix
,
wh
ich
im
p
lies
th
e
u
s
er
s
.
im
p
lies
th
e
r
ig
h
t
o
r
th
o
g
o
n
al
o
f
th
e
u
s
er
-
item
m
atr
ix
,
wh
ich
in
d
icate
s
th
e
item
s
.
|
N
u
|
is
th
e
n
u
m
b
er
o
f
item
s
r
ated
b
y
u
s
er
u
an
d
is
th
e
lef
t
o
r
th
o
g
o
n
al
v
alu
e
o
f
th
e
b
in
ar
y
m
atr
ix
ar
r
ay
.
(
,
>
0
)
eq
u
als o
n
e
if
,
h
as a
v
alu
e,
o
th
er
wis
e
ze
r
o
.
Fig
u
r
e
1
.
Ov
e
r
v
iew
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
T
h
e
s
to
ch
asti
c
g
r
ad
ien
t
d
escen
t
(
SGD
)
alg
o
r
ith
m
is
ad
o
p
te
d
to
o
p
tim
ize
an
d
r
e
d
u
ce
th
e
r
eg
u
lar
ized
s
q
u
ar
e
er
r
o
r
.
T
h
e
o
p
tim
izatio
n
tech
n
iq
u
e
is
p
er
f
o
r
m
e
d
to
r
eg
u
lar
ize
th
e
s
q
u
ar
e
e
r
r
o
r
v
ia
th
e
SGD
alg
o
r
ith
m
.
T
h
e
r
esu
lt
o
f
s
q
u
ar
e
er
r
o
r
s
h
o
u
ld
b
e
clo
s
e
to
a
p
ar
ticu
lar
th
r
es
h
o
ld
;
th
is
p
r
o
ce
s
s
is
ca
lled
r
eg
u
lar
izatio
n
.
I
n
itially
,
,
,
an
d
ar
e
co
m
p
u
ted
.
T
h
en
,
r
an
d
o
m
v
alu
es
ar
e
s
et
to
v
ec
to
r
s
p
u,
k
a
n
d
q
i,
k
,
wh
e
r
e
p
u
,
k
in
d
i
ca
tes
a
v
ec
to
r
o
f
u
s
er
s
,
u
r
ef
e
r
s
to
th
e
n
u
m
b
e
r
o
f
u
s
er
s
,
an
d
k
is
th
e
d
im
en
s
io
n
v
alu
e.
q
i,
k
im
p
lies
a
v
ec
to
r
o
f
item
s
an
d
i
s
tan
d
s
f
o
r
th
e
n
u
m
b
e
r
o
f
item
s
.
T
h
e
s
im
ilar
ity
o
f
ex
p
licit
an
d
im
p
lic
i
t
s
o
cial
r
elatio
n
s
is
u
tili
ze
d
to
en
h
an
ce
an
d
r
ed
u
ce
th
e
r
eg
u
lar
izatio
n
er
r
o
r
o
f
v
ec
t
o
r
an
d
th
e
v
alu
e
o
f
u
s
er
-
b
ias
(
)
.
As
s
h
o
wn
in
(
5
)
a
n
d
(
6
)
ar
e
u
s
ed
to
r
ed
u
ce
th
e
r
eg
u
lar
izatio
n
er
r
o
r
:
=
∑
(
,
−
−
−
−
(
+
1
√
|
|
∑
∈
)
)
2
(
,
)
∈
+
(
2
+
2
+
‖
‖
2
+
‖
‖
2
+
‖
‖
2
)
+
∑
∑
(
,
)
‖
−
‖
2
+
(
)
=
1
∑
∑
(
,
∗
)
‖
−
∗
‖
2
∗
∗
(
)
=
1
(
5
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6
9
3
0
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
,
Vo
l.
19
,
No
.
2
,
Ap
r
il 2
0
2
1
:
4
7
1
-
47
8
474
=
∑
(
,
)
∈
(
−
−
−
−
(
+
1
√
|
|
∑
∈
)
)
2
+
(
2
+
2
+
‖
‖
2
+
‖
‖
2
+
‖
‖
2
)
(
6
)
w
h
er
e
βe
an
d
βi
a
r
e
c
o
n
s
tan
t
v
alu
es
th
at
c
o
n
tr
o
l
th
e
im
p
ac
t
d
eg
r
ee
s
o
f
th
e
ex
p
licit
a
n
d
im
p
licit
r
elatio
n
s
h
ip
s
r
esp
ec
tiv
ely
;
F
(
i)
an
d
F
*
(
i)
r
e
s
p
ec
tiv
ely
d
en
o
te
th
e
n
u
m
b
er
s
o
f
ex
p
licit
f
r
ie
n
d
s
an
d
im
p
licit
f
r
ien
d
s
o
f
u
s
er
u
;
s
im(
x,
y)
r
ef
er
s
to
th
e
s
im
i
lar
ity
v
alu
e
b
etwe
en
u
s
er
s
x
an
d
y
;
K
is
th
e
n
u
m
b
er
o
f
r
atin
g
s
,
a
n
d
λ
is
th
e
co
n
s
tan
t
v
alu
e
th
at
d
eter
m
in
es th
e
d
eg
r
ee
o
f
th
e
co
n
s
tr
ain
t.
T
h
e
last
two
eq
u
atio
n
s
ar
e
u
tili
ze
d
to
f
i
n
d
th
e
b
est v
alu
e
f
o
r
th
e
v
ec
to
r
s
,
,
,
an
d
.
T
h
e
f
o
llo
w
in
g
eq
u
atio
n
s
s
h
o
w
th
e
u
p
d
a
tin
g
v
alu
es a
f
ter
ea
c
h
lo
o
p
:
=
+
(
−
)
(
7
)
=
+
(
∂
E
∂
(
+
1
√
|
|
∑
∈
)
−
)
(
8
)
=
+
(
(
√
|
|
)
−
)
(
9
)
=
+
(
−
)
(
1
0
)
=
+
(
−
(
1
1
)
w
h
er
e
is
th
e
lear
n
in
g
r
ate,
wh
ich
is
a
co
n
s
tan
t v
alu
e
th
at
co
n
tr
o
ls
th
e
r
eg
u
lar
izatio
n
er
r
o
r
d
u
r
in
g
th
e
tr
ain
in
g
s
tag
e.
Af
ter
o
p
tim
izin
g
th
e
v
al
u
es
o
f
th
e
v
ec
to
r
s
,
p
r
ed
ictio
n
v
alu
es
ar
e
co
m
p
u
ted
u
s
in
g
as
s
h
o
wn
in
(
3
)
.
Fin
ally
,
th
e
ev
alu
atio
n
p
r
o
ce
s
s
is
p
er
f
o
r
m
ed
to
c
o
m
p
u
te
t
h
e
ac
cu
r
ac
y
o
f
th
e
p
r
ed
ictio
n
b
y
c
o
m
p
ar
i
n
g
th
e
r
esu
lts
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
with
th
e
test
in
g
p
ar
t
th
r
o
u
g
h
th
e
m
ea
n
a
b
s
o
lu
te
er
r
o
r
(
MA
E
)
an
d
r
o
o
t
m
ea
n
s
q
u
ar
e
er
r
o
r
(
R
MSE
)
.
3.
RE
SU
L
T
S
A
ND
AN
AL
Y
SI
S
I
n
th
is
s
ec
tio
n
,
th
e
d
ataset,
p
ar
am
eter
s
,
ev
alu
atio
n
m
etr
ic,
an
d
an
aly
s
is
o
f
th
e
r
esu
lts
ar
e
ex
p
lain
ed
.
I
n
s
u
b
s
ec
tio
n
3
.
1
,
th
e
en
v
ir
o
n
m
en
t
s
etu
p
in
clu
d
es
d
escr
ip
t
io
n
o
f
th
e
d
atasets
,
th
e
ev
a
lu
atio
n
m
etr
ics
ar
e
d
eter
m
in
d
ed
,
a
n
d
th
e
p
r
ev
i
o
u
s
s
tu
d
ies
th
at
ar
e
u
tili
ze
d
in
th
e
co
m
p
ar
is
o
n
ar
e
a
d
d
r
ess
ed
as
well.
Mo
r
eo
v
e
r
,
in
s
u
b
s
ec
tio
n
3
.
3
,
th
e
r
esu
lts
ar
e
ex
p
lain
ed
an
d
th
e
ad
v
a
n
tag
es
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
ar
e
h
i
g
h
lig
h
ted
.
Fin
ally
,
in
s
u
b
s
ec
tio
n
3
.
4
,
th
e
im
p
ac
t o
f
ex
p
lo
itin
g
s
o
cial
r
elatio
n
s
is
d
em
o
n
s
tr
ated
.
3
.
1
.
E
nv
iro
nm
ent
s
et
up
E
n
v
ir
o
n
m
en
t
s
etu
p
in
clu
d
es
d
a
taset
s
elec
tio
n
,
ev
alu
atio
n
m
etr
ics,
p
ar
am
eter
s
ettin
g
s
,
an
d
tr
ai
n
in
g
s
ize.
All
o
f
th
e
m
ar
e
ex
p
lain
ed
in
d
etail
.
T
h
r
ee
d
atasets
wer
e
u
tili
ze
d
in
th
is
s
tu
d
y
:
L
ast.Fm
,
C
iao
,
an
d
Fil
m
T
r
u
s
t.
L
ast.Fm
was
r
elea
s
ed
b
y
H
etR
ec
in
2
0
1
1
in
t
h
e
f
r
am
e
wo
r
k
o
f
t
h
e
s
ec
o
n
d
i
n
ter
n
at
io
n
al
wo
r
k
s
h
o
p
o
n
in
f
o
r
m
atio
n
[
2
3
]
.
T
h
is
d
ataset
co
n
tain
s
2
,
1
0
0
u
s
er
s
an
d
1
8
,
7
4
5
item
s
.
E
ac
h
u
s
er
r
ated
5
0
ite
m
s
.
T
h
e
d
ata
in
clu
d
e
lis
ten
in
g
co
u
n
ts
f
o
r
ea
c
h
item
.
As
s
h
o
wn
(
1
2
)
was
ad
o
p
ted
t
o
m
ap
th
e
co
u
n
t
in
g
v
alu
es
o
v
er
a
r
an
g
e
(
1
-
5
)
o
f
r
atin
g
v
alu
es,
wh
ich
was
p
r
o
p
o
s
ed
b
y
[
2
4
]
.
Mo
r
e
o
v
er
,
th
is
d
ataset
co
n
tain
s
a
s
o
cial
in
f
o
r
m
atio
n
f
ile
o
n
th
e
r
elatio
n
s
h
ip
s
b
etwe
en
th
e
u
s
er
s
.
r
=
{
⌊
l
og
10
l
⌋
+
1
,
if
⌊
l
og
10
l
⌋
+
1
≤
5
5
,
othe
r
w
ise
}
(
1
2
)
Her
e,
r
is
th
e
r
atin
g
in
teg
er
v
al
u
e,
l
th
e
lis
ten
in
g
co
u
n
t
,
an
d
⌊
x
⌋
th
e
o
p
er
atio
n
o
f
r
o
u
n
d
i
n
g
d
o
wn
to
war
d
s
ze
r
o
.
I
n
th
is
s
tu
d
y
,
to
c
o
m
p
ar
e
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
with
o
th
er
s
tu
d
ies,
u
s
er
s
with
f
ewe
r
t
h
a
n
f
iv
e
r
elatio
n
s
h
ip
s
wer
e
r
em
o
v
ed
.
T
h
e
r
ef
o
r
e
,
1
,
1
2
3
u
s
er
s
r
em
ain
e
d
.
C
iao
was
th
e
s
ec
o
n
d
d
ataset
g
en
er
ate
d
b
y
[
2
5
]
.
T
h
is
d
ataset
en
ab
les
u
s
er
s
to
cr
ea
te
th
eir
r
elatio
n
s
h
ip
s
.
I
t
is
a
p
r
o
d
u
ct
r
e
v
iew
web
s
ite
o
n
wh
ich
u
s
er
s
ca
n
r
ate
an
d
r
ev
iew
p
r
o
d
u
cts.
T
h
is
d
ataset
co
n
s
is
ts
o
f
7
,
3
7
5
u
s
er
s
an
d
9
9
,
7
4
6
ite
m
s
.
I
n
m
atr
ix
f
ac
to
r
izatio
n
m
e
th
o
d
s
,
wh
en
a
u
s
er
o
r
an
item
v
ec
to
r
is
n
o
t
r
ated
,
th
e
en
tire
v
ec
to
r
v
alu
es
ar
e
ze
r
o
.
T
h
er
ef
o
r
e,
it
b
ec
o
m
es
im
p
o
s
s
ib
le
to
f
in
d
th
e
b
est
lo
w
-
r
an
k
lin
ea
r
r
e
p
r
esen
t
atio
n
o
f
th
e
u
s
er
-
item
m
atr
i
x
[
2
6
]
.
T
h
u
s
,
in
th
is
s
tu
d
y
,
ev
e
r
y
u
s
er
with
at
leas
t
one
r
elatio
n
s
h
i
p
,
an
d
wh
o
h
ad
a
r
atin
g
g
r
ea
ter
th
an
o
r
e
q
u
al
to
1
,
was
s
elec
ted
.
Similar
ly
,
ev
er
y
item
r
ated
a
t
g
r
ea
ter
th
an
1
was
s
elec
ted
.
Su
b
s
eq
u
en
tly
,
6
,
7
6
7
u
s
er
s
with
2
2
,
2
2
9
item
s
r
em
ain
ed
.
Fil
m
T
r
u
s
t,
p
r
o
d
u
ce
d
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
A
d
o
p
tin
g
ex
p
licit a
n
d
imp
licit so
cia
l rela
tio
n
s
b
y
S
V
D++
fo
r
r
ec
o
mme
n
d
a
tio
n
.
.
.
(
Mo
h
s
in
Ha
s
a
n
Hu
s
s
ein
)
475
by
[
2
7
]
,
was
th
e
last
d
ataset
u
s
ed
in
th
is
s
tu
d
y
.
T
h
is
d
ataset,
a
ls
o
co
n
tain
in
g
s
o
cial
in
f
o
r
m
ati
o
n
,
in
v
o
lv
es
1
,
5
0
8
u
s
er
s
an
d
2
,
0
7
1
item
s
.
Acc
u
r
a
cy
m
etr
ics
wer
e
ad
o
p
ted
to
e
v
a
lu
ate
p
r
ed
ictio
n
a
cc
u
r
ac
y
in
th
i
s
s
tu
d
y
;
th
ese
b
ein
g
m
ea
n
ab
s
o
lu
te
er
r
o
r
(
MA
E
)
a
n
d
r
o
o
t
m
ea
n
a
b
s
o
lu
te
er
r
o
r
(
R
MSE
)
[
2
8
,
2
9
]
.
T
h
e
f
o
llo
win
g
f
o
r
m
u
las
co
m
p
u
te
th
ese
m
etr
ics.
=
1
∑
|
,
−
̅
,
|
=
1
(
1
3
)
=
√
1
∑
(
,
−
̅
,
)
2
=
1
(
1
4
)
w
h
er
e
N
is
th
e
to
tal
n
u
m
b
er
o
f
th
e
p
r
ed
ictio
n
,
,
an
d
̅
,
ar
e
th
e
ac
tu
al
r
atin
g
an
d
th
e
r
atin
g
v
alu
e
co
m
p
u
t
e
d
b
y
th
e
m
eth
o
d
o
f
item
i
an
d
b
y
u
s
er
u
r
esp
ec
tiv
ely
.
Z
er
o
v
al
u
e
im
p
lies
an
o
p
tim
al
r
esu
lt.
I
n
co
n
tr
ast,
th
e
h
i
g
h
v
alu
e
im
p
lies
lo
wer
ac
cu
r
ac
y
.
Ou
r
p
r
o
p
o
s
ed
m
eth
o
d
is
co
m
p
ar
ed
with
o
th
er
m
atr
i
x
f
a
cto
r
izati
o
n
s
tu
d
ies
in
clu
d
in
g
:
−
SVD
p
r
o
p
o
s
ed
b
y
[
3
0
]
:
T
h
is
m
eth
o
d
m
ak
es
p
r
e
d
ictio
n
s
b
y
ex
p
lo
itin
g
in
f
o
r
m
atio
n
o
n
l
y
an
d
it
is
u
s
ed
to
co
m
p
ar
e
e
v
alu
atio
n
s
f
o
r
ev
er
y
d
ataset.
−
Pro
b
ab
ilis
tic
m
atr
ix
f
ac
t
o
r
izat
io
n
(
PMF)
p
r
o
p
o
s
ed
b
y
[
8
]
:
T
h
is
m
eth
o
d
u
tili
ze
s
o
n
ly
th
e
u
s
er
-
i
tem
m
atr
ix
f
o
r
p
r
ed
ictio
n
a
n
d
it is
u
s
ed
f
o
r
r
esu
lts
with
ev
er
y
d
ataset.
−
SVD++
p
r
o
p
o
s
ed
b
y
[
9
]
:
T
h
is
m
eth
o
d
is
ag
ain
o
n
ly
u
s
ed
f
o
r
r
atin
g
in
f
o
r
m
atio
n
,
wh
ic
h
is
ap
p
lied
to
all
d
atasets
.
−
So
cial
r
ec
o
m
m
en
d
atio
n
(
So
R
eg
)
p
r
o
p
o
s
ed
b
y
[
1
7
]
:
T
h
e
a
u
th
o
r
h
er
e
u
s
es
ex
p
licit
s
o
cial
in
f
o
r
m
atio
n
in
ad
d
itio
n
to
t
h
e
r
atin
g
i
n
f
o
r
m
at
io
n
.
−
E
I
SR
p
r
o
p
o
s
ed
b
y
[
2
0
]
:
T
h
is
s
t
u
d
y
is
clo
s
e
to
o
u
r
wo
r
k
s
in
ce
ex
p
licit
an
d
im
p
licit
r
elatio
n
s
h
ip
s
ar
e
ad
o
p
ted
.
T
h
e
co
m
p
a
r
is
o
n
is
ac
h
iev
ed
f
o
r
L
ast.Fm
an
d
C
iao
.
−
User
E
m
b
ed
d
in
g
UE
-
SVD++
in
[
2
2
]
.
T
h
is
is
c
o
n
s
tr
u
cted
an
d
u
tili
ze
d
f
o
r
b
o
th
ex
p
licit
a
n
d
im
p
licit
r
atin
g
in
f
o
r
m
atio
n
with
th
e
u
s
er
em
b
ed
d
in
g
m
atr
ix
b
y
th
e
p
r
o
p
o
s
ed
u
s
er
-
wis
e
m
u
tu
al
in
f
o
r
m
atio
n
.
I
t
was
u
s
ed
f
o
r
Fil
m
T
r
u
s
t.
Fo
r
im
p
ar
tial
co
m
p
ar
is
o
n
,
e
v
er
y
m
eth
o
d
is
im
p
lem
en
ted
with
th
e
s
am
e
p
ar
am
eter
v
alu
es.
T
h
e
p
ar
am
eter
s
ettin
g
s
u
s
ed
in
th
is
s
tu
d
y
ar
e
d
em
o
n
s
tr
ated
as f
o
llo
ws:
−
d
: th
e
d
im
en
s
io
n
s
o
f
th
e
laten
t
f
ea
tu
r
es e
q
u
allin
g
5
an
d
1
0
.
−
β_
e
an
d
β_
i
ar
e
th
e
weig
h
ts
o
f
ex
p
licit
an
d
im
p
licit
s
o
cial
r
elatio
n
s
h
ip
s
r
esp
ec
tiv
ely
.
T
h
e
v
alu
es
o
f
b
o
th
weig
h
ts
ar
e
0
.
0
5
f
o
r
L
ast.Fm
,
0
.
0
0
1
f
o
r
C
iao
an
d
0
.
0
0
5
f
o
r
F
ilm
T
r
u
s
t.
−
λ
is
th
e
r
eg
u
lar
izatio
n
p
a
r
am
et
er
s
et
at
0
.
0
0
1
f
o
r
L
ast.Fm
an
d
C
iao
,
an
d
at
0
.
0
0
0
1
f
o
r
Fil
m
T
r
u
s
t.
−
α
is
th
e
lear
n
in
g
r
ate
s
et
at
0
.
0
5
f
o
r
L
ast.Fm
,
0
.
0
0
6
f
o
r
C
iao
an
d
0
.
0
1
5
f
o
r
Fil
m
T
r
u
s
t
.
3
.
2
.
Dis
cus
s
io
ns
T
h
e
aim
o
f
th
is
s
tu
d
y
is
to
s
h
o
w
th
e
im
p
ac
t o
f
i
n
teg
r
atin
g
s
o
cial
r
elatio
n
s
(
ex
p
licit
a
n
d
im
p
licit)
with
th
e
r
atin
g
in
f
o
r
m
atio
n
(
e
x
p
licit
r
atin
g
an
d
im
p
licit
f
ee
d
b
ac
k
in
f
o
r
m
atio
n
)
in
to
th
e
SVD++
m
eth
o
d
to
allev
iate
th
e
s
p
ar
city
is
s
u
e.
T
o
im
p
lem
en
t
th
e
p
r
o
p
o
s
ed
m
eth
o
d
,
th
e
d
atasets
ar
e
d
iv
id
ed
in
to
two
p
ar
ts
:
tr
ain
in
g
an
d
test
in
g
.
T
h
e
tr
ain
in
g
p
ar
t
is
8
0
%
o
f
th
e
en
tire
u
s
er
s
,
wh
ich
a
r
e
s
elec
ted
r
an
d
o
m
ly
.
T
h
e
o
th
er
p
ar
t
(
2
0
%)
is
f
o
r
test
in
g
.
T
h
e
ex
p
er
im
e
n
ts
ar
e
im
p
lem
en
ted
1
0
tim
es
an
d
th
e
m
ea
n
s
o
f
th
e
o
u
tco
m
es
f
o
r
th
e
MA
E
an
d
R
MSE
v
alu
es
ar
e
co
m
p
u
ted
.
T
ab
les
1
to
3
s
h
o
w
th
e
r
esu
lts
o
f
MA
E
an
d
R
MSE
wh
en
d=
5
an
d
d=
1
0
an
d
th
e
y
ar
e
b
en
ch
m
ar
k
ed
with
o
th
er
s
tu
d
i
es
ap
p
lied
to
th
e
L
ast.Fm
,
C
i
ao
,
an
d
Fil
m
T
r
u
s
t
d
atasets
.
T
h
e
b
est
r
esu
lts
ar
e
p
r
esen
ted
in
b
o
ld
f
ac
e.
As
s
h
o
wn
in
T
ab
les
1
to
3
,
s
o
ci
al
r
elatio
n
s
h
av
e
a
s
ig
n
if
ican
t
i
m
p
ac
t
o
n
en
h
an
ci
n
g
p
r
e
d
ictio
n
ac
cu
r
ac
y
.
I
n
th
e
Fil
m
T
r
u
s
t
d
ataset
(
T
ab
l
e
1
)
,
th
e
clo
s
est
r
esu
lts
to
o
u
r
s
tu
d
y
a
r
e
UE
-
SVD++
,
wh
er
e
th
e
p
r
o
p
o
s
ed
m
eth
o
d
o
u
tp
er
f
o
r
m
ed
UE
-
SVD++
with
an
im
p
r
o
v
em
en
t
o
f
ab
o
u
t
0
.
0
2
%
f
o
r
MA
E
an
d
0
.
0
0
0
8
%
f
o
r
R
MSE
wh
en
d
=5
.
W
h
er
ea
s
wh
en
d
=1
0
,
th
e
im
p
r
o
v
e
m
en
t
is
ab
o
u
t
0
.
0
3
%
f
o
r
MA
E
an
d
0
.
0
1
%
f
o
r
R
MSE
.
Mo
r
eo
v
er
,
th
e
im
p
r
o
v
em
e
n
ts
in
u
s
in
g
s
o
cial
r
elatio
n
s
ar
e
p
er
s
is
ten
t,
esp
ec
ially
b
y
ad
d
in
g
im
p
licit
s
o
cial
r
elatio
n
s
th
at
f
u
r
th
er
b
o
o
s
t
th
e
p
r
ed
ictio
n
p
r
o
ce
s
s
.
T
h
er
ef
o
r
e,
E
I
SR
an
d
th
e
p
r
o
p
o
s
ed
m
eth
o
d
r
ea
c
h
ed
p
ea
k
r
esu
lts
f
o
r
b
o
th
m
etr
ics
(
R
MSE
an
d
MA
E
)
b
y
u
s
in
g
th
is
s
o
u
r
ce
.
Ad
d
itio
n
ally
,
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
ac
h
iev
ed
t
h
e
b
est
r
esu
lts
an
d
ex
ce
ed
ed
all
th
e
af
o
r
em
en
tio
n
ed
s
tu
d
ies
wh
er
e
th
e
d
if
f
er
e
n
c
e
in
T
ab
le
2
b
etwe
en
t
h
e
p
r
o
p
o
s
ed
m
eth
o
d
an
d
th
e
n
ea
r
est
s
tu
d
y
(
E
I
SR
)
is
0
.
0
0
1
7
f
o
r
MA
E
an
d
0
.
0
0
4
7
f
o
r
R
MSE
wh
en
d
=5
.
T
h
e
s
am
e
p
r
ef
e
r
en
ce
ca
n
b
e
s
ee
n
in
T
ab
le
3
,
wh
er
e
th
e
p
r
o
p
o
s
ed
m
eth
o
d
also
o
u
tp
e
r
f
o
r
m
ed
o
th
er
s
tu
d
ies
f
o
r
b
o
t
h
m
etr
ic
s
.
T
o
s
u
m
u
p
,
th
e
ef
f
ec
tiv
en
ess
o
f
th
is
s
tu
d
y
m
itig
ates
th
e
s
p
ar
s
ity
r
atio
b
y
ex
p
lo
itin
g
f
o
u
r
s
o
u
r
c
es
o
f
in
f
o
r
m
atio
n
,
n
am
ely
,
ex
p
licit
s
o
cial
r
elatio
n
s
,
im
p
li
cit
s
o
cial
r
elatio
n
s
,
ex
p
licit
u
s
er
s
’
r
atin
g
s
,
an
d
im
p
licit
f
ee
d
b
ac
k
in
f
o
r
m
atio
n
.
T
h
er
ef
o
r
e,
th
e
r
e
s
u
lts
o
f
t
h
e
ta
b
les s
h
o
w
th
at
th
e
p
r
o
p
o
s
ed
m
eth
o
d
o
u
tp
er
f
o
r
m
th
e
p
r
ev
i
o
u
s
s
tu
d
ies in
ter
m
s
o
f
ac
cu
r
ac
y
an
d
ac
h
iev
e
th
e
o
b
je
ctiv
e
o
f
th
is
s
tu
d
y
b
y
f
u
r
th
er
im
p
r
o
v
i
n
g
p
r
ed
ictio
n
ac
c
u
r
ac
y
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6
9
3
0
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
,
Vo
l.
19
,
No
.
2
,
Ap
r
il 2
0
2
1
:
4
7
1
-
47
8
476
T
ab
le
1
.
Per
f
o
r
m
an
ce
co
m
p
a
r
is
o
n
s
f
o
r
th
e
Fil
m
T
r
u
s
t d
ataset
(
8
0
% tr
ain
in
g
)
.
T
h
e
b
est r
esu
lts
ar
e
p
r
esen
ted
in
b
o
ld
f
ac
e
A
l
l
u
sers
M
e
t
r
i
c
s
d
=
5
d
=
1
0
P
M
F
M
A
E
0
.
7
1
4
0
.
7
3
5
R
M
S
E
0
.
9
4
9
0
.
9
6
8
S
V
D
M
A
E
0
.
7
0
9
0
.
7
0
9
R
M
S
E
0
.
9
2
5
0
.
9
5
4
S
V
D
+
+
M
A
E
0
.
6
1
3
0
.
6
1
1
R
M
S
E
0
.
8
0
4
0
.
8
0
2
S
o
R
e
g
M
A
E
0
.
6
7
4
0
.
6
6
8
R
M
S
E
0
.
8
7
8
0
.
8
7
5
U
E
S
V
D
+
+
M
A
E
0
.
6
2
0
3
0
.
6
2
8
0
R
M
S
E
0
.
8
0
2
6
0
.
8
1
2
0
P
r
o
p
o
se
d
m
e
t
h
o
d
M
A
E
0
.
6
0
7
2
0
.
6
0
9
6
R
M
S
E
0
.
8
0
2
0
0
.
8
0
1
1
T
ab
le
2
.
Per
f
o
r
m
an
ce
co
m
p
a
r
is
o
n
s
f
o
r
th
e
C
iao
d
ataset
(
8
0
% tr
ain
in
g
)
.
T
h
e
b
est r
esu
lts
ar
e
p
r
esen
ted
in
b
o
ld
f
ac
e
A
l
l
u
sers
M
e
t
r
i
c
s
d
=
5
d
=
1
0
P
M
F
M
A
E
0
.
9
2
0
1
.
0
7
8
R
M
S
E
1
.
2
0
6
0
.
8
2
2
S
V
D
M
A
E
0
.
7
8
4
0
.
7
8
8
R
M
S
E
1
.
0
3
3
1
.
0
3
7
S
V
D
+
+
M
A
E
0
.
7
5
2
0
.
7
4
8
R
M
S
E
1
.
0
1
3
1
.
0
0
1
S
o
R
e
g
M
A
E
0
.
8
9
9
0
.
8
1
5
R
M
S
E
1
.
1
8
3
1
.
0
7
6
EI
S
R
M
A
E
0
.
7
2
8
8
0
.
7
2
8
5
R
M
S
E
0
.
9
6
1
2
0
.
9
6
0
6
P
r
o
p
o
se
d
m
e
t
h
o
d
M
A
E
0
.
7
2
7
1
0
.
7
2
6
7
R
M
S
E
0
.
9
5
6
5
0
.
9
5
5
7
T
ab
le
3
.
Per
f
o
r
m
an
ce
co
m
p
a
r
is
o
n
s
f
o
r
th
e
L
ast.Fm
d
ataset
(
8
0
% tr
ain
in
g
)
.
T
h
e
b
est r
esu
lts
ar
e
p
r
esen
ted
in
b
o
ld
f
ac
e
A
l
l
u
sers
M
e
t
r
i
c
s
d
=
5
d
=
1
0
P
M
F
M
A
E
0
.
4
2
7
0
.
4
2
6
R
M
S
E
0
.
5
3
7
0
.
5
3
5
S
V
D
M
A
E
0
.
4
1
4
0
.
4
1
4
R
M
S
E
0
.
5
3
4
0
.
5
3
3
S
V
D
+
+
M
A
E
0
.
4
1
1
1
0
.
4
1
0
3
R
M
S
E
0
.
5
3
0
7
0
.
5
3
0
1
EI
S
R
M
A
E
0
.
4
0
6
5
0
.
4
0
6
0
R
M
S
E
0
.
5
2
6
3
0
.
5
2
5
8
P
r
o
p
o
se
d
m
e
t
h
o
d
M
A
E
0
.
4
0
1
2
0
.
4
0
0
6
R
M
S
E
0
.
5
1
8
7
0
.
5
1
7
7
I
n
T
ab
les
4
an
d
5
,
m
u
ltip
le
d
i
m
en
s
io
n
v
alu
es
ar
e
ap
p
lied
to
ch
ec
k
th
e
p
er
f
o
r
m
an
ce
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
in
d
if
f
er
en
t
d
v
al
u
es
with
th
e
ex
p
er
im
en
ts
b
ein
g
e
x
ec
u
ted
wh
en
d
=
1
0
,
3
0
,
5
0
,
7
0
,
an
d
9
0
.
Fo
r
ea
ch
d
im
en
s
io
n
v
alu
e,
MA
E
an
d
R
MSE
ar
e
ca
lcu
lated
.
I
n
T
ab
le
4
,
th
e
r
esu
lts
s
h
o
w
th
at
th
e
ac
cu
r
ac
y
is
s
ig
n
if
ican
tly
im
p
r
o
v
e
d
wh
e
n
th
e
d
im
e
n
s
io
n
v
alu
e
in
cr
ea
s
es,
wh
er
e
in
cr
ea
s
in
g
th
e
d
im
en
s
io
n
v
alu
e
m
ea
n
s
m
o
r
e
in
f
o
r
m
atio
n
ad
d
ed
to
t
h
e
v
ec
to
r
s
p
an
d
q
to
co
m
p
u
te
th
e
p
r
ed
ictio
n
.
T
h
e
b
est r
esu
lts
ar
e
ac
h
iev
ed
wh
en
d
eq
u
als 7
0
,
wh
er
e
th
e
r
esu
lt
o
f
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
is
0
.
5
1
6
6
f
o
r
R
MSE
an
d
0
.
3
9
8
9
f
o
r
MA
E
.
Ho
wev
er
,
th
e
ac
cu
r
ac
y
s
lig
h
tly
d
eter
io
r
ates
to
0
.
5
1
7
4
an
d
0
.
3
9
9
4
f
o
r
R
MSE
an
d
MA
E
r
esp
ec
tiv
ely
wh
en
d
is
9
0
.
T
h
e
s
a
m
e
s
ce
n
ar
io
ca
n
b
e
s
ee
n
in
o
th
er
s
tu
d
ies
(
SVD,
S
VD+
+,
E
-
SVD++
)
s
o
th
at
f
o
r
v
ar
io
u
s
d
im
en
s
io
n
v
alu
es,
o
u
r
p
r
o
p
o
s
ed
m
eth
o
d
o
u
tp
er
f
o
r
m
s
o
th
er
s
tu
d
ies.
T
a
b
le
5
s
h
o
ws
m
u
ltip
le
v
alu
es
o
f
d
f
o
r
th
e
C
iao
d
ataset.
T
h
e
s
am
e
s
ch
em
e
ca
n
b
e
s
ee
n
as in
T
ab
le
4
,
wh
e
r
e
R
MSE
an
d
MA
E
a
r
e
r
e
d
u
ce
d
(
im
p
r
o
v
in
g
ac
cu
r
ac
y
)
wh
en
d
in
c
r
ea
s
es.
Ho
wev
e
r
,
th
e
d
if
f
er
en
ce
h
er
e
is
in
th
e
p
ea
k
v
alu
es
o
f
R
MSE
an
d
MA
E
wh
e
n
d
=
5
0
.
T
h
e
r
esu
lts
d
ec
lin
e
s
lig
h
tly
a
f
ter
5
0
(
wh
en
d
=7
0
)
,
R
MSE
an
d
MA
E
ar
e
0
.
9
5
7
4
an
d
0
.
7
2
7
1
r
esp
ec
tiv
ely
.
W
h
en
d
=5
0
,
th
e
r
esu
lts
ar
e
0
.
9
5
4
2
f
o
r
R
MSE
an
d
0
.
7
2
6
2
f
o
r
MA
E
.
I
n
s
u
m
m
a
r
y
,
f
o
r
all
d
atasets
an
d
m
etr
ics
t
h
at
ar
e
u
s
ed
in
th
is
s
tu
d
y
,
th
e
p
r
o
p
o
s
ed
m
eth
o
d
ac
co
m
p
lis
h
ed
th
e
b
est
r
esu
lts
in
ter
m
s
o
f
ac
cu
r
ac
y
f
o
r
all
d
im
en
s
io
n
v
alu
es.
C
o
n
s
eq
u
e
n
tly
,
u
s
in
g
s
o
cial
in
f
o
r
m
atio
n
f
o
r
b
o
th
s
o
u
r
ce
s
(
ex
p
licit
an
d
im
p
licit)
with
im
p
licit
f
ee
d
b
ac
k
h
a
s
s
ig
n
if
ican
t
im
p
r
o
v
e
m
en
t,
wh
er
e
th
e
r
esu
lts
r
ev
ea
l th
e
o
u
tp
er
f
o
r
m
an
ce
o
f
t
h
e
p
r
o
p
o
s
ed
m
eth
o
d
in
all
d
im
en
s
io
n
v
alu
es.
T
ab
le
4
.
Per
f
o
r
m
an
ce
co
m
p
a
r
is
o
n
s
f
o
r
m
u
ltid
im
e
n
s
io
n
al
v
al
u
es f
o
r
L
ast.Fm
(
8
0
% tr
ain
in
g
)
M
a
t
r
i
c
e
s
M
e
t
h
o
d
s
S
V
D
S
V
D
+
+
E
-
S
V
D
+
+
P
r
o
p
o
se
d
m
e
t
h
o
d
d
=
1
0
R
M
S
E
0
.
5
3
3
0
.
5
3
2
1
0
.
5
2
5
7
0
.
5
1
7
7
M
A
E
0
.
4
1
4
0
0
.
4
1
2
3
0
.
4
0
3
3
0
.
4
0
0
6
d
=
3
0
R
M
S
E
0
.
5
3
2
8
0
.
5
3
1
6
0
.
5
2
4
8
0
.
5
1
7
5
M
A
E
0
.
4
1
3
8
0
.
4
1
1
8
0
.
4
0
2
5
0
.
3
9
9
8
d
=
5
0
R
M
S
E
0
.
5
3
2
5
0
.
5
2
9
3
0
.
5
2
2
4
0
.
5
1
6
9
M
A
E
0
.
4
1
3
4
0
.
4
0
9
2
0
.
4
0
1
2
0
.
3
9
9
2
d
=
7
0
R
M
S
E
0
.
5
3
1
5
0
.
5
2
8
9
0
.
5
2
2
1
0
.
5
1
6
6
M
A
E
0
.
4
1
2
8
0
.
4
0
8
7
0
.
3
9
9
4
0
.
3
9
8
9
d
=
9
0
R
M
S
E
0
.
5
3
2
8
0
.
5
2
9
5
0
.
5
2
4
2
0
.
5
1
7
4
M
A
E
0
.
4
1
3
5
0
.
4
0
6
9
0
.
4
0
1
2
0
.
3
9
9
4
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
A
d
o
p
tin
g
ex
p
licit a
n
d
imp
licit so
cia
l rela
tio
n
s
b
y
S
V
D++
fo
r
r
ec
o
mme
n
d
a
tio
n
.
.
.
(
Mo
h
s
in
Ha
s
a
n
Hu
s
s
ein
)
477
T
ab
le
5
.
Per
f
o
r
m
an
ce
co
m
p
a
r
is
o
n
s
f
o
r
m
u
ltid
im
e
n
s
io
n
al
v
al
u
es f
o
r
th
e
C
iao
d
ataset
(
8
0
%
tr
ain
in
g
)
M
a
t
r
i
c
e
s
M
e
t
h
o
d
s
S
V
D
S
V
D
+
+
E
-
S
V
D
+
+
P
r
o
p
o
se
d
m
e
t
h
o
d
d
=
1
0
R
M
S
E
1
.
0
3
7
1
.
0
0
1
0
.
9
5
7
1
0
.
9
5
5
7
M
A
E
0
.
7
8
8
0
.
7
4
8
8
0
.
7
2
9
2
0
.
7
2
6
7
d
=
3
0
R
M
S
E
1
.
0
3
2
0
.
9
9
7
2
0
.
9
5
6
4
0
.
9
5
5
1
M
A
E
0
.
7
8
7
4
0
.
7
4
8
6
0
.
7
2
8
8
0
.
7
2
6
5
d
=
5
0
R
M
S
E
1
.
0
2
1
0
.
9
9
6
0
.
9
5
5
8
0
.
9
5
4
2
M
A
E
0
.
7
8
6
9
0
.
7
4
8
1
0
.
7
2
8
6
0
.
7
2
6
2
d
=
7
0
R
M
S
E
1
.
0
1
6
0
.
9
9
5
2
0
.
9
5
7
7
0
.
9
5
6
1
M
A
E
0
.
7
8
4
0
.
7
4
8
3
0
.
7
2
8
9
0
.
7
2
6
6
d
=
9
0
R
M
S
E
1
.
0
2
4
0
.
9
9
6
0
.
9
5
8
5
0
.
9
5
7
4
M
A
E
0
.
7
8
7
0
.
7
4
8
5
0
.
7
2
9
4
0
.
7
2
7
1
3.
3
.
I
m
pa
c
t
o
f
cha
ng
ing
bet
a
v
a
lues
Fig
u
r
es 2
,
3
an
d
4
s
h
o
w
th
e
im
p
ac
t o
f
th
e
b
eta
p
ar
am
eter
in
d
if
f
er
en
t v
alu
es f
o
r
Fil
m
T
r
u
s
t,
C
iao
,
an
d
L
ast.Fm
d
atasets
.
T
h
e
b
eta
p
ar
am
eter
is
u
s
ed
to
co
n
tr
o
l
th
e
d
eg
r
ee
o
f
ex
p
licit
an
d
im
p
licit
s
o
cial
r
elatio
n
s
.
T
h
e
b
eta
v
alu
e
is
m
u
ltip
lied
b
y
th
e
s
im
ilar
i
ty
v
alu
e
f
o
r
ea
ch
p
air
o
f
ex
p
licit
o
r
im
p
licit
s
o
cial
r
e
latio
n
s
.
As
s
ee
n
in
th
ese
f
ig
u
r
es,
th
e
R
MSE
v
alu
e
ch
an
g
es
wh
e
n
th
e
b
eta
v
al
u
e
d
o
es.
Fo
r
ex
am
p
le,
Fig
u
r
e
2
s
h
o
ws
th
e
R
MSE
r
esu
lts
f
o
r
d
if
f
er
e
n
t
v
alu
es.
I
t c
an
b
e
s
ee
n
th
at
th
e
R
MSE
ac
h
iev
ed
th
e
wo
r
s
t
r
esu
lt
(
0
.
8
0
2
8
)
wh
en
b
eta
eq
u
als
0
.
0
0
0
1
.
Ho
we
v
er
,
t
h
e
r
esu
lts
im
p
r
o
v
e
d
wh
e
n
b
eta
i
n
cr
ea
s
ed
.
T
h
e
b
est
r
esu
lt
(
0
.
8
0
1
1
)
is
r
e
g
is
ter
ed
wh
en
b
eta
eq
u
als
0
.
0
5
.
Af
ter
war
d
,
th
e
R
MSE
r
esu
lt
ag
ain
s
tar
ts
to
g
r
a
d
u
ally
in
c
r
ea
s
e
(
b
e
less
ac
cu
r
ate)
wh
en
th
e
b
eta
v
alu
e
in
cr
ea
s
es.
T
h
e
s
am
e
th
i
n
g
ca
n
b
e
s
ee
n
in
Fig
u
r
es
3
a
n
d
4
wh
er
e
b
eta
v
alu
es
af
f
ec
t
th
e
R
MSE
r
esu
lt.
Hen
ce
,
ex
p
licit
an
d
im
p
licit
s
o
cial
r
elatio
n
s
h
av
e
a
s
ig
n
if
ican
t
r
o
le
to
en
h
an
ce
th
e
p
r
e
d
ictio
n
.
T
h
is
r
o
le
is
s
h
o
wn
b
y
th
e
b
eta
v
alu
e.
T
h
u
s
,
th
e
b
eta
v
alu
e
ca
n
en
h
an
ce
p
r
ed
ict
io
n
ac
cu
r
ac
y
f
o
r
a
s
p
ec
if
ic
v
a
lu
e.
W
h
en
th
e
b
eta
v
alu
e
is
g
r
ea
ter
o
r
s
m
aller
th
a
n
a
th
r
esh
o
l
d
,
th
e
ac
c
u
r
ac
y
is
n
eg
ativ
ely
af
f
ec
ted
.
Fig
u
r
e
2
.
I
m
p
ac
t o
f
b
eta
v
al
u
e
s
o
n
R
MSE
r
esu
lt f
o
r
Fil
m
T
r
u
s
t
Fig
u
r
e
3
.
I
m
p
ac
t o
f
b
eta
v
al
u
e
s
o
n
R
MSE
r
esu
lt
f
o
r
C
iao
Fig
u
r
e
4
.
I
m
p
ac
t o
f
b
eta
v
al
u
e
s
o
n
R
MSE
r
esu
lt f
o
r
L
ast.Fm
4.
CO
NCLU
SI
O
N
I
n
t
h
i
s
p
a
p
e
r
,
s
o
ci
a
l
r
el
a
t
i
o
n
s
w
e
r
e
e
x
p
l
o
i
t
e
d
i
n
t
w
o
c
h
a
n
n
el
s
(
e
x
p
l
i
ci
t
a
n
d
i
m
p
li
c
it
r
e
l
at
io
n
s
h
i
p
s
)
t
o
i
m
p
r
o
v
e
r
e
c
o
m
m
e
n
d
a
t
i
o
n
p
e
r
f
o
r
m
a
n
c
e
.
I
m
p
l
i
c
i
t
r
el
a
t
i
o
n
s
h
i
p
s
w
e
r
e
e
x
t
r
a
c
t
e
d
b
y
a
p
p
l
y
i
n
g
t
h
e
r
e
s
o
u
r
c
e
a
l
l
o
c
at
i
o
n
i
n
d
e
x
(
R
A
I
)
w
h
i
c
h
p
r
e
d
i
c
ts
h
i
d
d
e
n
i
n
f
o
r
m
a
t
i
o
n
i
n
t
h
e
s
o
c
ia
l
g
r
a
p
h
.
F
o
u
r
s
o
u
r
c
es
o
f
i
n
f
o
r
m
a
t
io
n
,
n
a
m
e
l
y
,
e
x
p
l
ic
it
s
o
c
i
a
l
r
el
a
t
i
o
n
s
,
i
m
p
l
i
ci
t
s
o
ci
a
l
r
e
l
a
t
i
o
n
s
,
e
x
p
li
c
i
t
r
at
i
n
g
,
a
n
d
i
m
p
l
i
c
it
r
a
t
i
n
g
f
e
e
d
b
a
c
k
a
r
e
i
n
c
o
r
p
o
r
a
t
e
d
i
n
t
o
t
h
e
S
V
D
++
m
e
t
h
o
d
t
o
a
ll
e
v
i
a
t
e
t
h
e
s
p
a
r
s
i
t
y
p
r
o
b
l
e
m
.
S
o
c
ia
l
r
el
a
t
io
n
s
(
i
m
p
l
i
ci
t
a
n
d
e
x
p
l
i
ci
t
r
el
a
t
io
n
s
h
i
p
s
)
a
r
e
u
t
il
i
z
ed
t
o
h
e
l
p
a
c
t
i
v
e
u
s
e
r
s
t
o
f
i
n
d
t
h
eir
p
r
e
f
e
r
e
n
c
e
i
t
e
m
s
b
y
c
o
m
p
u
t
i
n
g
s
i
m
i
l
a
r
i
t
ie
s
b
e
tw
e
e
n
a
c
ti
v
e
u
s
e
r
s
a
n
d
t
h
e
i
r
s
o
ci
a
l
r
e
l
a
t
i
o
n
s
.
T
h
r
e
e
r
e
al
-
w
o
r
l
d
d
a
t
as
e
ts
w
e
r
e
u
s
e
d
,
n
a
m
el
y
L
as
t
.
F
m
,
F
i
l
m
T
r
u
s
t
a
n
d
C
ia
o
.
T
h
e
r
es
u
l
ts
o
f
t
h
e
e
x
p
e
r
i
m
e
n
t
r
e
v
e
a
l
e
d
t
h
a
t
s
o
ci
a
l
r
el
a
ti
o
n
s
h
a
v
e
a
n
o
b
v
i
o
u
s
i
m
p
a
ct
t
o
b
o
o
s
t
t
h
e
p
r
e
d
i
c
ti
o
n
a
c
c
u
r
a
c
y
.
M
o
r
e
o
v
e
r
,
t
h
e
r
e
s
u
l
ts
r
e
v
e
a
l
e
d
t
h
a
t
o
u
r
p
r
o
p
o
s
e
d
m
e
th
o
d
i
s
s
u
p
e
r
i
o
r
t
o
o
t
h
e
r
s
t
u
d
i
e
s
s
u
c
h
a
s
S
V
D
,
S
V
D+
+
,
P
MF
,
E
U
-
S
V
D+
+
,
S
o
cR
e
g
,
a
n
d
E
I
SR
i
n
t
e
r
m
s
o
f
ac
c
u
r
a
c
y
.
T
h
e
p
r
o
p
o
s
e
d
m
et
h
o
d
s
i
g
n
i
f
i
ca
n
t
l
y
i
m
p
r
o
v
e
d
t
h
e
p
r
e
d
i
c
ti
o
n
b
y
e
x
p
l
o
i
t
i
n
g
u
s
e
r
s
’
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6
9
3
0
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
,
Vo
l.
19
,
No
.
2
,
Ap
r
il 2
0
2
1
:
4
7
1
-
47
8
478
r
e
l
a
t
i
o
n
s
h
i
p
s
as
it
c
o
n
s
i
d
e
r
s
t
h
a
t
a
l
l
f
r
i
e
n
d
s
h
a
v
e
s
i
m
i
la
r
t
as
t
es
.
A
c
c
o
r
d
i
n
g
l
y
,
c
l
as
s
i
f
y
i
n
g
f
r
i
e
n
d
s
i
n
t
o
v
a
r
i
o
u
s
g
r
o
u
p
s
a
n
d
c
o
m
p
u
t
i
n
g
t
h
e
s
i
m
i
l
a
r
i
t
y
o
f
e
a
c
h
g
r
o
u
p
s
e
p
a
r
a
t
e
l
y
m
a
y
e
n
h
a
n
c
e
p
r
e
d
i
c
t
i
o
n
s
a
n
d
a
c
h
i
e
v
e
m
o
r
e
p
r
e
c
i
s
e
r
e
s
u
l
t
s
.
E
v
e
n
t
u
a
l
l
y
,
t
h
e
p
r
o
p
o
s
e
d
m
et
h
o
d
a
d
o
p
t
e
d
t
h
e
p
e
a
r
s
o
n
c
o
r
r
e
l
at
i
o
n
t
o
c
o
m
p
u
t
e
t
h
e
s
i
m
il
a
r
i
t
y
.
D
i
f
f
e
r
e
n
t
s
i
m
il
a
r
i
ty
m
e
a
s
u
r
e
s
c
a
n
b
e
e
m
p
l
o
y
e
d
t
o
f
u
r
t
h
e
r
i
m
p
r
o
v
e
t
h
e
r
e
s
u
l
ts
.
RE
F
E
R
E
NC
E
S
[1
]
S
.
P
o
t
lu
ri
a
n
d
K.
S
.
Ra
o
,
“
Im
p
r
o
v
e
d
q
u
a
li
ty
o
f
se
rv
ice
-
b
a
se
d
c
lo
u
d
se
rv
ice
ra
n
k
i
n
g
a
n
d
re
c
o
m
m
e
n
d
a
ti
o
n
m
o
d
e
l,
”
T
EL
KOM
NIKA
T
e
lec
o
mm
u
n
ic
a
ti
o
n
C
o
mp
u
ti
n
g
El
e
c
tro
n
ics
a
n
d
Co
n
tro
l
,
v
o
l
.
1
8
,
n
o
.
3
,
p
p
.
1
2
5
2
-
1
2
5
8
,
J
u
n
.
2
0
2
0
.
[2
]
M
.
Ku
n
a
v
e
r
a
n
d
T.
P
o
ž
rl,
“
Kn
o
wle
d
g
e
-
Ba
se
d
S
y
ste
m
s
Div
e
rsity
in
re
c
o
m
m
e
n
d
e
r
sy
ste
m
s
–
A
su
rv
e
y
,
”
v
o
l.
1
2
3
,
p
p
.
1
5
4
-
1
6
2
,
2
0
1
7
,
d
o
i:
1
0
.
1
0
1
6
/
j
.
k
n
o
sy
s.
2
0
1
7
.
0
2
.
0
0
9
.
[3
]
I.
P
ra
se
ty
a
n
i
n
g
r
u
m
,
e
t
a
l.
,
“
Ap
p
li
c
a
ti
o
n
o
f
re
c
o
m
m
e
n
d
a
ti
o
n
s
y
st
e
m
with
AH
P
m
e
th
o
d
a
n
d
se
n
ti
m
e
n
t
a
n
a
ly
sis,”
T
EL
KOM
NIKA
T
e
lec
o
mm
u
n
ic
a
ti
o
n
C
o
mp
u
ti
n
g
El
e
c
tro
n
ics
a
n
d
Co
n
tro
l
,
v
o
l
.
1
8
,
n
o
.
3
,
1
3
4
3
-
1
3
5
3
,
Ju
n
.
2
0
2
0
.
[4
]
H.
-
F
.
Wan
g
a
n
d
C.
-
T.
W
u
,
“
A
stra
teg
y
-
o
rien
ted
o
p
e
ra
ti
o
n
m
o
d
u
le
fo
r
re
c
o
m
m
e
n
d
e
r
sy
ste
m
s
in
E
-
c
o
m
m
e
rc
e
,
”
Co
mp
u
t
.
Op
e
r.
Res
.
,
v
o
l.
3
9
,
n
o
.
8
,
p
p
.
1
8
3
7
–
1
8
4
9
,
A
u
g
.
2
0
1
2
.
[5
]
T
.
J
i
u
l
e
,
e
t
a
l
.
,
“
C
o
l
l
a
b
o
r
a
t
i
v
e
f
i
lt
e
r
i
n
g
r
e
c
o
m
m
e
n
d
a
t
i
o
n
m
o
d
e
l
c
o
n
s
i
d
e
r
i
n
g
i
n
t
e
g
r
a
t
i
o
n
o
f
u
s
e
r
ra
t
in
g
a
n
d
a
t
t
r
i
b
u
t
e
s
i
m
i
l
a
r
i
t
y
,
”
T
E
L
K
O
M
N
I
K
A
T
e
le
c
o
m
m
u
n
i
c
a
t
i
o
n
C
o
m
p
u
t
i
n
g
E
l
e
c
t
r
o
n
i
c
s
a
n
d
C
o
n
t
r
o
l
,
v
o
l
.
1
4
,
n
o
.
2
,
p
p
.
2
0
8
-
2
1
6
,
2
0
1
6
.
[6
]
G
.
Li
n
d
e
n
,
B.
S
m
it
h
,
a
n
d
J.
Yo
r
k
,
“
Am
a
z
o
n
.
c
o
m
re
c
o
m
m
e
n
d
a
ti
o
n
s:
Item
-
to
-
it
e
m
c
o
ll
a
b
o
ra
t
iv
e
f
il
terin
g
,
”
IEE
E
In
ter
n
e
t
Co
m
p
u
t
.
,
v
o
l.
7
,
n
o
.
1
,
p
p
.
7
6
-
8
0
,
2
0
0
3
.
[7
]
B.
S
a
rwa
r,
G
.
Ka
ry
p
is,
J.
Ko
n
sta
n
,
a
n
d
J
.
Ried
l,
“
Item
-
b
a
se
d
c
o
l
l
a
b
o
ra
ti
v
e
fil
teri
n
g
re
c
o
m
m
e
n
d
a
ti
o
n
a
lg
o
rit
h
m
s,”
Pro
c
e
e
d
in
g
s
o
f
t
h
e
1
0
t
h
in
ter
n
a
ti
o
n
a
l
c
o
n
fer
e
n
c
e
o
n
W
o
rl
d
W
id
e
W
e
b
,
2
0
0
1
,
p
p
.
2
8
5
-
2
9
5
.
[8
]
A.
M
n
i
h
,
e
t
a
l.
,
“
P
ro
b
a
b
il
isti
c
m
a
tri
x
fa
c
to
riza
ti
o
n
,
”
A
d
v
a
n
c
e
s in
n
e
u
ra
l
i
n
f
o
rm
a
ti
o
n
p
ro
c
e
ss
in
g
sy
ste
ms
,
2
0
0
8
.
[9
]
Y.
Hu
,
Y.
Ko
re
n
,
a
n
d
C.
Vo
li
n
sk
y
,
“
C
o
ll
a
b
o
ra
ti
v
e
fi
lt
e
rin
g
f
o
r
imp
li
c
it
fe
e
d
b
a
c
k
d
a
tas
e
ts,”
2
0
0
8
Ei
g
h
t
h
IE
EE
In
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
D
a
ta
M
in
in
g
,
2
0
0
8
,
p
p
.
2
6
3
-
2
7
2
.
[1
0
]
A.
He
rn
a
n
d
o
,
J.
Bo
b
a
d
il
la,
a
n
d
F
.
Orte
g
a
,
“
A
n
o
n
n
e
g
a
t
iv
e
m
a
tri
x
fa
c
to
riza
ti
o
n
f
o
r
c
o
l
lab
o
ra
ti
v
e
fil
terin
g
re
c
o
m
m
e
n
d
e
r
sy
ste
m
s
b
a
se
d
o
n
a
Ba
y
e
sia
n
p
ro
b
a
b
il
ist
ic
m
o
d
e
l,
”
K
n
o
wled
g
e
-
Ba
se
d
S
y
st.
,
v
o
l
.
9
7
,
p
p
.
1
8
8
-
2
0
2
,
2
0
1
6
.
[1
1
]
Y.
Ca
o
,
W.
Li
,
a
n
d
D.
Z
h
e
n
g
,
“
An
imp
ro
v
e
d
n
e
i
g
h
b
o
r
h
o
o
d
-
a
wa
re
u
n
ifi
e
d
p
ro
b
a
b
il
isti
c
m
a
tri
x
fa
c
to
riza
ti
o
n
re
c
o
m
m
e
n
d
a
ti
o
n
,
”
W
ire
l.
Per
s.
C
o
mm
u
n
.
,
v
o
l.
1
0
2
,
n
o
.
4
,
p
p
.
3
1
2
1
-
3
1
4
0
,
Ja
n
.
2
0
1
8
.
[1
2
]
H
.
P
a
rv
in
,
e
t
a
l
.
,
“
A
s
c
a
l
a
b
l
e
a
n
d
ro
b
u
st
t
ru
s
t
-
b
a
s
e
d
n
o
n
n
e
g
a
t
iv
e
m
a
t
r
ix
f
a
c
to
r
i
z
a
t
io
n
r
e
c
o
m
m
e
n
d
e
r
u
s
in
g
t
h
e
a
l
t
e
r
n
a
ti
n
g
d
i
r
e
c
t
io
n
m
e
th
o
d
,
”
Kn
o
wl
e
d
g
e
-
Ba
s
e
d
S
y
s
t
.
,
v
o
l
.
1
6
6
,
p
p
.
9
2
-
1
0
7
,
2
0
1
9
.
[
1
3
]
K
.
S
u
r
e
n
d
ro
,
e
t
a
l.
,
“
In
t
e
g
r
a
t
e
d
S
o
c
i
a
l
M
e
d
i
a
Kn
o
wl
e
d
g
e
C
a
p
tu
r
e
i
n
M
e
d
i
c
a
l
D
o
m
a
in
o
f
I
n
d
o
n
e
si
a
,
”
T
E
L
K
OM
N
IK
A
T
e
l
e
c
o
m
m
u
n
i
c
a
ti
o
n
Co
mp
u
t
in
g
E
l
e
c
t
ro
n
i
c
s
a
n
d
C
o
n
t
r
o
l
,
v
o
l.
1
6
,
n
o
.
4
,
p
p
.
1
8
4
6
-
1
8
5
6
,
2
0
1
8
.
[1
4
]
R
.
K
u
m
a
r
,
e
t
a
l
.
,
“
S
o
c
i
a
l
P
o
p
u
l
a
r
i
t
y
b
a
s
e
d
S
V
D
+
+
R
e
c
o
m
m
e
n
d
e
r
S
y
s
t
e
m
,
”
I
n
t
.
J
.
C
o
m
p
u
t
.
A
p
p
l
.
,
v
o
l
.
8
7
,
n
o
.
1
4
,
2
0
1
4
.
[1
5
]
A
.
M
.
A
.
A
l
-
S
a
b
a
a
w
i
,
e
t
a
l
.
,
“
E
x
p
l
o
i
t
i
n
g
i
m
p
l
i
c
i
t
s
o
c
i
a
l
r
e
l
a
t
i
o
n
s
h
i
p
s
v
i
a
d
i
m
e
n
s
i
o
n
r
e
d
u
c
t
i
o
n
t
o
i
m
p
r
o
v
e
r
e
c
o
m
m
e
n
d
a
t
i
o
n
s
y
s
t
e
m
p
e
r
f
o
r
m
a
n
c
e
,
”
P
L
o
S
O
n
e
,
v
o
l
.
1
5
,
n
o
.
4
,
2
0
2
0
,
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s
:
/
/
d
o
i
.
o
r
g
/
1
0
.
1
3
7
1
/
j
o
u
r
n
a
l
.
p
o
n
e
.
0
2
3
1
4
5
7
.
[1
6
]
A.
An
a
n
d
h
a
n
,
L
.
S
h
u
i
b
,
M
.
A.
I
sm
a
il
,
a
n
d
G
.
M
u
jt
a
b
a
,
“
S
o
c
ial
m
e
d
ia
re
c
o
m
m
e
n
d
e
r
sy
ste
m
s:
re
v
iew
a
n
d
o
p
e
n
re
se
a
rc
h
issu
e
s
,
”
IEE
E
Acc
e
ss
,
v
o
l.
6
,
p
p
.
1
5
6
0
8
-
1
5
6
2
8
,
F
e
b
.
2
0
1
8
,
d
o
i:
1
0
.
1
1
0
9
/ACCES
S
.
2
0
1
8
.
2
8
1
0
0
6
2
.
[1
7
]
H.
M
a
,
D.
Zh
o
u
,
C.
L
iu
,
M
.
R
.
L
y
u
,
a
n
d
I.
Kin
g
,
“
Re
c
o
m
m
e
n
d
e
r
sy
ste
m
s
with
so
c
ial
re
g
u
lariz
a
ti
o
n
,
”
Pro
c
e
e
d
in
g
s
o
f
th
e
fo
u
rth
ACM
in
ter
n
a
ti
o
n
a
l
c
o
n
fer
e
n
c
e
o
n
W
e
b
se
a
rc
h
a
n
d
d
a
ta
min
in
g
,
2
0
1
1
,
p
p
.
2
8
7
-
2
9
6
.
[1
8
]
M
.
Hu
ss
e
in
,
H.
Na
ji
,
a
n
d
W
.
Bh
a
y
a
,
“
In
flu
e
n
ti
a
l
n
o
d
e
s
b
a
se
d
a
ll
e
v
iatio
n
o
f
u
se
r
c
o
ld
-
st
a
rt
p
ro
b
lem
in
re
c
o
m
m
e
n
d
a
ti
o
n
s
y
ste
m
,
”
Res
.
J
.
Ap
p
l
.
S
c
i.
,
v
o
l.
1
1
,
p
p
.
1
1
0
7
-
1
1
1
4
,
Ja
n
.
2
0
1
6
,
d
o
i:
1
0
.
3
9
2
3
/r
jas
c
i.
2
0
1
6
.
1
1
0
7
.
1
1
1
4
.
[1
9
]
Y.
Zh
a
n
g
,
W.
C
h
e
n
,
a
n
d
Z.
Yi
n
,
“
Co
ll
a
b
o
ra
ti
v
e
fil
teri
n
g
wit
h
so
c
ial
re
g
u
lariz
a
ti
o
n
fo
r
TV
p
ro
g
ra
m
re
c
o
m
m
e
n
d
a
ti
o
n
,
”
Kn
o
wled
g
e
-
Ba
se
d
S
y
st.
,
v
o
l.
5
4
,
p
p
.
3
1
0
-
3
1
7
,
2
0
1
3
.
[2
0
]
W.
Re
a
fe
e
,
N.
S
a
l
im,
a
n
d
A.
Kh
a
n
,
“
T
h
e
p
o
we
r
o
f
imp
li
c
it
s
o
c
ial
r
e
latio
n
in
ra
ti
n
g
p
r
e
d
icti
o
n
o
f
so
c
i
a
l
re
c
o
m
m
e
n
d
e
r
sy
ste
m
s,”
PL
o
S
O
n
e
,
v
o
l.
1
1
,
n
o
.
5
,
2
0
1
6
.
[2
1
]
H.
M
a
,
“
An
e
x
p
e
rime
n
tal
stu
d
y
o
n
imp
li
c
i
t
so
c
ial
re
c
o
m
m
e
n
d
a
ti
o
n
,
”
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
3
6
th
i
n
t
e
rn
a
ti
o
n
a
l
ACM
S
IGIR
c
o
n
fer
e
n
c
e
o
n
Res
e
a
rc
h
a
n
d
d
e
v
e
lo
p
me
n
t
i
n
i
n
fo
rm
a
t
io
n
re
triev
a
l
,
2
0
1
3
,
p
p
.
7
3
-
82
.
[2
2
]
W.
S
h
i,
L.
Wan
g
,
a
n
d
J.
Qin
,
“
Us
e
r
e
m
b
e
d
d
in
g
f
o
r
ra
ti
n
g
p
re
d
i
c
ti
o
n
i
n
S
VD
+
+
-
b
a
se
d
c
o
ll
a
b
o
ra
ti
v
e
fil
teri
n
g
,
”
S
y
mm
e
try
(Ba
se
l).
,
v
o
l.
1
2
,
n
o
.
1
,
p
.
1
2
1
,
2
0
2
0
.
[2
3
]
I.
Ca
n
ta
d
o
r,
e
t
a
l.
,
“
S
e
c
o
n
d
wo
rk
sh
o
p
o
n
in
f
o
rm
a
ti
o
n
h
e
tero
g
e
n
e
it
y
a
n
d
fu
si
o
n
i
n
re
c
o
m
m
e
n
d
e
r
sy
ste
m
s
(He
tRec
2
0
1
1
),
”
Pro
c
e
e
d
in
g
s o
f
t
h
e
fi
ft
h
AC
M
c
o
n
fer
e
n
c
e
o
n
Rec
o
mm
e
n
d
e
r sy
ste
ms
,
2011
,
p
p
.
3
8
7
-
388
.
[2
4
]
J.
Li
u
,
C.
W
u
,
a
n
d
W.
Li
u
,
“
Ba
y
e
sia
n
p
ro
b
a
b
il
isti
c
m
a
tri
x
fa
c
to
riz
a
ti
o
n
wit
h
so
c
ial
re
latio
n
s
a
n
d
it
e
m
c
o
n
ten
ts
fo
r
re
c
o
m
m
e
n
d
a
ti
o
n
,
”
De
c
is.
S
u
p
p
o
rt
S
y
st.
,
v
o
l
.
5
5
,
n
o
.
3
,
p
p
.
8
3
8
-
8
5
0
,
Ju
n
.
2
0
1
3
.
[2
5
]
J.
Tan
g
,
H.
G
a
o
,
H.
Li
u
,
a
n
d
A.
D
a
s
S
a
rm
a
,
“
e
Tru
st:
Un
d
e
rsta
n
d
i
n
g
tru
st
e
v
o
lu
ti
o
n
i
n
a
n
o
n
l
in
e
wo
rl
d
,
”
Pro
c
e
e
d
in
g
s
o
f
t
h
e
1
8
th
ACM
S
IGKD
D i
n
ter
n
a
ti
o
n
a
l
c
o
n
fer
e
n
c
e
o
n
K
n
o
wle
d
g
e
d
isc
o
v
e
ry
a
n
d
d
a
t
a
min
i
n
g
,
2
0
1
2
,
p
p
.
2
5
3
-
2
6
1
.
[2
6
]
R.
Ch
e
n
,
e
t
a
l
.
,
“
A
su
r
v
e
y
o
f
c
o
ll
a
b
o
ra
ti
v
e
fil
teri
n
g
-
b
a
se
d
re
c
o
m
m
e
n
d
e
r
s
y
ste
m
s:
F
ro
m
trad
i
ti
o
n
a
l
m
e
th
o
d
s
t
o
h
y
b
ri
d
m
e
th
o
d
s
b
a
se
d
o
n
so
c
ial
n
e
two
rk
s,”
IEE
E
Acc
e
ss
,
v
o
l
.
6
,
p
p
.
6
4
3
0
1
-
6
4
3
2
0
,
2
0
1
8
.
[2
7
]
G
.
G
u
o
,
J.
Z
h
a
n
g
,
a
n
d
N.
Y
o
r
k
e
-
S
m
it
h
,
“
A
n
o
v
e
l
b
a
y
e
sia
n
si
m
il
a
rit
y
m
e
a
su
re
fo
r
re
c
o
m
m
e
n
d
e
r
sy
ste
m
s,”
T
we
n
ty
-
T
h
ir
d
In
ter
n
a
ti
o
n
a
l
J
o
in
t
Co
n
fer
e
n
c
e
o
n
Arti
fi
c
i
a
l
I
n
telli
g
e
n
c
e
,
Ja
n
.
2
0
1
3
.
[2
8
]
M
.
Ja
li
li
,
S
.
Ah
m
a
d
ian
,
M
.
Iz
a
d
i,
P
.
M
o
ra
d
i
,
a
n
d
M
.
S
a
leh
i
,
“
Ev
a
lu
a
ti
n
g
c
o
ll
a
b
o
ra
ti
v
e
fi
lt
e
rin
g
re
c
o
m
m
e
n
d
e
r
a
lg
o
rit
h
m
s: a su
rv
e
y
,
”
IEE
E
a
c
c
e
ss
,
v
o
l
.
6
,
p
p
.
7
4
0
0
3
-
7
4
0
2
4
,
2
0
1
8
.
[2
9
]
A.
Da
’u
,
N.
S
a
li
m
,
I.
Ra
b
iu
,
a
n
d
A.
Os
m
a
n
,
“
Re
c
o
m
m
e
n
d
a
ti
o
n
sy
s
tem
e
x
p
l
o
it
i
n
g
a
sp
e
c
t
-
b
a
se
d
o
p
in
i
o
n
m
in
i
n
g
wit
h
d
e
e
p
lea
rn
in
g
m
e
th
o
d
,
”
I
n
f.
S
c
i.
(
Ny
).
,
v
o
l.
5
1
2
,
p
p
.
1
2
7
9
-
1
2
9
2
,
F
e
b
.
2
0
2
0
.
[3
0
]
B.
S
a
rwa
r,
G
.
Ka
ry
p
is,
J.
Ko
n
sta
n
,
a
n
d
J.
Ried
l,
“
Ap
p
l
ica
ti
o
n
o
f
d
i
m
e
n
sio
n
a
li
t
y
re
d
u
c
ti
o
n
i
n
re
c
o
m
m
e
n
d
e
r
sy
ste
m
-
a
c
a
se
stu
d
y
,
”
M
in
n
e
so
ta
U
n
iv
M
in
n
e
a
p
o
li
s De
p
t
o
f
C
o
mp
u
ter
S
c
ien
c
e
,
2
0
0
0
.
Evaluation Warning : The document was created with Spire.PDF for Python.