Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
9
, No
.
5
,
Octo
ber
201
9
, pp.
4408
~
44
16
IS
S
N:
20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v
9
i
5
.
pp44
08
-
44
16
4408
Journ
al h
om
e
page
:
http:
//
ia
es
core
.c
om/
journa
ls
/i
ndex.
ph
p/IJECE
Identifyi
ng lea
rning style
th
ro
ugh
eye tr
ackin
g techn
ology in
adapti
ve learning
sy
s
tems
Inssaf El G
uabassi
1
, Z
akari
a
B
ou
s
alem
2
, Mo
hamm
ed
A
l Achh
ab
3
,
Is
mail
jell
ou
li
4
,
Badr Ed
dine
EL
Mohaji
r
5
1,4,5
Facul
t
y
of
Sc
ie
nc
es
,
Abde
lmale
k
Essaa
di
Univ
ersity
,
Moroco
2
Facul
t
y
of
Sci
e
nce
and Technol
ogie
s,
Hass
an
1
st
Univer
sit
y
,
Mor
occ
o
3
Nati
ona
l
School
of
Appli
ed
Sc
ience
s,
Abd
el
m
al
e
k
Essaa
di
Unive
rsit
y
,
Moroco
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
hist
or
y:
Re
cei
ved
J
ul
1
,
201
8
Re
vised
A
pr 18
, 2
01
9
Accepte
d
Apr
25
, 201
9
Le
arn
er
l
ea
rnin
g
s
t
y
l
e
rep
r
ese
nts
a
ke
y
prin
c
ipl
e
and
cor
e
v
al
ue
of
the
ada
pt
ive
l
ea
rn
in
g
s
y
stems
(ALS).
Mor
eove
r
,
und
ersta
nding individual
l
ea
rn
er
le
arn
ing
st
y
le
s
is
a
ver
y
good
condi
ti
on
for
havi
ng
the
b
est
servic
es
of
resourc
e
ad
aptat
ion.
How
eve
r
,
t
he
m
aj
ority
of
the
ALS,
whic
h
consider
le
arn
ing
st
y
le
s,
use
questi
onnair
es
in
orde
r
to
de
te
c
t
it,
where
as
thi
s
m
et
hod
has
a
var
ious
disadva
nta
g
e
s,
For
exa
m
ple
,
it
is
unsuita
ble
for
som
e
kinds
of
responde
nts,
tim
e
-
consum
ing
to
complet
e
,
it
m
a
y
be
m
isunderstood
b
y
responde
nt,
etc
.
In
th
e
pr
e
sent
pap
er,
w
e
propose
an
appr
oa
ch
for
aut
om
atic
al
l
y
detec
t
ing
l
ea
r
ning
st
y
le
s
in
ALS
base
d
on
e
y
e
tracki
n
g
te
chno
log
y
,
b
ecause
i
t
rep
rese
nt
s
one
of
the
m
ost
informat
ive
ch
ara
c
te
rist
ic
s
of
gaze
b
eha
vior
.
Th
e
expe
rimen
t
al
r
esult
s
show
e
d
a
h
igh
r
el
a
ti
on
ship
among
the
Fe
lde
r
-
Silv
e
rm
an
Le
a
rning
St
y
le
and
the
e
y
e
m
ovements
rec
ord
ed
whilst
l
ea
rning
.
Ke
yw
or
d
s
:
Ad
a
ptat
ion
Ad
a
ptive lea
rni
ng
Ey
e
tracki
ng
Learn
i
ng sty
le
d
et
ect
ion
Copyright
©
201
9
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
In
ssa
f El
Gua
ba
ssi,
Faculty
of S
ci
e
nces,
Abdelm
al
ek
Essaadi
Un
i
ver
sit
y,
P.O.BO
X 212
1, Tet
ua
n, 93
000,
M
orocc
o
.
Em
a
il
:
el
gu
aba
ssi@gm
ai
l.com
1.
INTROD
U
CTION
Trad
it
io
nal
ed
ucati
on
syst
e
m
s,
wh
ic
h
al
low
le
ar
ne
r
to
le
arn
in
dep
e
nd
e
ntly
without
at
te
nd
in
g
a
cl
assroom
to
m
eet
the
tutor
,
are
unable
to
dynam
ic
al
l
y
a
dap
t
to
t
he
le
arn
e
r’
s
nee
ds
.
S
ub
s
eq
ue
ntly
they
are
un
a
ble
to
inc
re
ase
the
outp
ut
of
the
le
a
rn
e
rs.
I
n
this
res
pect
,
rece
ntly
the
c
on
ce
pt
of
a
dapt
at
ion
has
bec
om
e
an
i
m
po
rtant issu
e o
f
resea
rch
i
n
le
arn
i
ng
area
; i
nd
eed, pro
vid
in
g
ada
ptivit
y i
n
le
arn
in
g
syst
e
m
s h
el
ps
learner
s t
o
m
ake th
e i
nten
ded lea
r
ning
outc
om
es v
ia
a
per
s
onal
iz
ed w
ay
.
This
pa
per
r
e
pr
ese
nts
a
c
on
ti
nu
at
ion
of
our
pr
e
vious
w
orks
car
ried
out
in
the
ada
pt
ive
le
a
rn
in
g
syst
e
m
s (
ALS)
[
1
-
4], in w
hic
h we
prov
i
de
d adaptivit
y i
n u
biquit
ous lear
ni
ng
syst
em
s b
a
sed on l
ear
ning
sty
les
of
Felde
r
-
Sil
ve
rm
an
and
le
a
rn
e
r
c
on
te
xt.
Howe
ver,
a
ne
w
a
dap
ta
ti
on
pro
blem
h
as
app
ea
re
d,
nam
el
y
the
autom
at
ic
detect
ion
of
le
ar
ner le
arn
i
ng
sty
le
s.
Learn
i
ng
sty
le
s
are
inc
reasing
ly
inco
rpo
r
at
ed
to
enh
a
nc
e
le
arn
in
g
outc
om
es
fo
r
le
arn
e
rs,
an
d
therefo
re
m
any
research
a
ppr
oach
e
s
are
done
in
edu
cat
io
n
area,
in
fact,
m
os
t
research
e
rs
ag
ree
that
le
arn
i
ng
sty
le
s
play
a
key
ro
le
in
t
his
area.
O
ne
of
the
m
os
t
widely
us
e
d
m
od
el
s
of
le
arn
i
ng
sty
le
s
is
the
Felder
-
Sil
ve
rm
an
Inde
x
of
Learn
i
ng
Sty
le
s
(ILS)
.
T
his
m
od
el
is
ex
trem
e
ly
si
m
ple
an
d,
furthe
r
m
or
e,
the
res
ults
are
easy
to
inter
pret
.
H
ow
e
v
er
,
a
t
the
sam
e
time
the
qu
est
io
nnai
res
i
n
gen
e
r
al
su
f
fer
f
ro
m
sever
al
disad
va
ntages,
bo
t
h
the
or
et
ic
al
and
pr
act
ic
a
l.
For
exam
ple,
it
is
extrem
ely
i
m
po
ssible
t
o
record
c
ha
ng
ing
of
le
arn
in
g
sty
le
s
beca
us
e
it
is
im
po
ssible
to
r
epeate
dly
ask
le
arn
e
rs
to
com
pl
et
e
the
quest
ionnaire
for
ti
m
e
an
d
cost c
on
ce
rn. Fort
un
at
el
y, this
problem
can
be
so
l
ved u
si
ng
autom
at
ic
d
et
ect
ion
of lea
r
ning sty
le
s.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
Id
e
ntif
yi
ng
le
arnin
g
sty
le
th
r
ough eye tr
ack
ing
tec
hnol
og
y
in ada
ptive
learn
i
ng syste
ms
(
In
ss
af El
G
uaba
s
si
)
4409
Nowa
days,
it
is
po
ssible
to
detect
the
l
earn
e
r
le
ar
ning
sty
le
s
via
the
bio
m
et
ric
t
echnolo
gies.
Fo
r
exam
ple
ey
e
trackin
g,
in
fact,
this
te
ch
nolo
gy
has
bee
n
extensi
vely
use
d
in
a
la
r
ge
va
riet
y
of
a
pp
l
ic
at
ion
s
su
c
h
as
diag
nosti
cs
[5
,
6],
int
eracti
on
a
nd
a
ccessi
bili
ty
[7
-
11
]
an
d
analy
ti
cs
[12
-
15]
.
Thi
s
pap
e
r
inv
e
sti
gates
and
disc
us
ses
an
a
ppr
oach
to
pro
vid
e
ada
pt
at
ion
in
le
ar
ni
ng
syst
em
s
thro
ug
h
ey
e
trac
ki
ng
or
gaze
-
tra
ckin
g,
this
te
ch
no
l
og
y
al
lows
aut
oma
ti
cal
ly
un
de
rs
ta
nd
in
g
of
i
nd
i
vidual
le
ar
ner
le
arn
i
ng
prefe
re
nces
by
recor
din
g
of
the point
-
of
-
ga
ze (P
OG).
The
pap
e
r
is
structu
re
d
as
f
ollows.
A
fter
pr
ese
ntin
g
the
backg
rou
nd,
S
ect
ion
3
co
nc
entrates
on
pro
po
se
d
a
ppr
oach.
Se
ct
ion
4
descr
i
bes
our
resea
rc
h
m
et
hodo
l
og
y.
S
ect
ion
5
pr
e
se
nts
the
ex
pe
rim
ental
resu
lt
s.
Finall
y, Sect
io
n 6 dis
play
s the m
ai
n
con
cl
us
io
ns
a
nd
fu
t
ur
e
resea
r
ch.
2.
BACKG
ROU
ND
In
this
sect
io
n
we
will
discuss
the
three
m
ain
te
rm
s
to
achieve
our
a
ppr
oa
ch,
these
te
rm
s
are:
Firstl
y
the lea
r
ning
obj
ect
s (LO
),
sec
ondly t
he
le
a
rni
ng
sty
le
s an
d t
hir
dly t
he
ey
e t
rack
i
ng
.
2
.
1.
Le
ar
ning
o
b
ject
s
Learn
i
ng
obj
e
ct
(LO)
is
a
con
t
roversi
al
con
ce
pt,
bec
au
se
unfortu
nate
ly
the
sci
entifi
c
literat
ur
e
do
e
s
not
pro
vid
e
a
si
ng
le
,
c
oncrete,
a
nd
co
nse
nsual
de
finiti
on
for
it
,
A
le
arn
i
ng
obj
ect
c
an
be
de
fine
d
a
s
“
an
y
entit
y,
dig
it
al
or
non
-
dig
it
al,
tha
t
ma
y
be
us
e
d
for
le
arn
ing
,
e
du
c
atio
n
or
trainin
g
”
[16].
Mo
re
si
m
pl
y,
a
le
arn
in
g
obje
ct
is
“
an
y
dig
it
al
reso
ur
ce
th
at
can
be
reu
s
ed
to
sup
port
le
arnin
g”
[
17
]
.
Likewise,
a
le
a
rn
i
n
g
obj
ect
ca
n
be
de
fine
d
as
“
A
di
gital
sel
f
-
co
ntai
ned
and
reu
sable
entit
y,
wi
th
a
cl
e
ar
e
du
c
at
ional
pur
pose,
wi
th
at
le
as
t
three
interna
l
an
d
editable
co
m
pone
nts:
c
on
t
ent,
le
arnin
g
activ
it
ie
s
an
d
el
emen
ts
of
con
te
xt
.
Th
e
le
arnin
g
obje
ct
s
m
us
t
ha
ve
an
ext
er
na
l
structure
of
i
nfo
rm
atio
n
t
o
fa
ci
li
tate
their
id
entif
ic
ation
,
storage
and
ret
rie
val
:
the
meta
data
.”
[18].
In
s
hort,
we
pro
po
se
t
he
fo
ll
owin
g
ve
ry
global
an
d
work
i
ng
de
fini
ti
on
:
A
le
ar
ning
obje
ct
is a d
i
gital
learni
ng r
es
our
ce.
On
the
oth
e
r
hand,
the
le
ar
ning
obj
ect
s
ha
ve
the
f
ollowi
ng
key
cha
ra
ct
erist
ic
s
[1
9]:
Is
a
par
t
of
le
arn
in
g
m
at
er
ia
ls
-
essenti
al
ly
con
sist
of
sm
a
ll
er
un
it
s
of
le
arn
i
ng,
usual
ly
between
2
m
inu
te
s
and
15
m
inu
te
s.
I
s
sel
f
-
c
on
ta
ine
d
-
e
ver
y
le
ar
ning
obj
ect
ca
n
be
t
aken
in
dep
e
nd
ently
.
Is
Re
us
a
ble
(RL
O
=
re
us
a
ble
le
arn
in
g
obj
ect
)
-
each
le
ar
ning
obj
ect
ca
n
be
us
e
d
in
m
ulti
ple
le
arn
in
g
m
ater
ia
l.
Is
A
ggre
gated
-
L
Os
ca
n
be
gro
up
e
d
into
l
arg
e
r
co
ll
ect
io
ns
of
co
ntent.
Is
ta
gg
e
d
with
m
et
adata
-
ea
ch
LO
ha
s
des
cripti
ve
inf
or
m
at
ion
al
lowing.
T
he
m
edia
con
te
nt
of
a
le
ar
ning
obj
ect
ca
n
incl
ude:
vid
e
o,
te
xt,
aud
i
o,
im
age,
exam
ple,
def
in
it
ion
,
il
lustrati
on
,
exe
rcise,
diag
ram
, s
im
ulati
on
, as
sessm
ent, etc.
2
.
2
.
Le
ar
ning
styles
Learn
e
rs
hav
e
diff
e
re
nt
ways
of
le
ar
ning;
each
le
arn
e
r
ha
s
his
own
per
ce
pt
ion
le
vel,
his
own
way
of
bui
ldi
ng
an
d
retai
ning
knowle
dge.
That’
s
w
hy
a
sin
gl
e
sty
le
wo
n'
t
be
a
ppr
opria
te
for
al
l
le
arn
e
rs
.
Ther
e
f
or
e
it
is
necessa
ry
to
gra
sp
his
le
ar
ning
sty
le
first,
in
orde
r
to
co
rr
e
ct
ly
cho
ose
the
strat
egies
a
nd
ada
pt
the syst
e
m
to
it.
In
our
a
ppr
oa
ch
we
c
hose th
e Felder
-
Sil
ve
r
m
an
le
arn
in
g
s
ty
le
, b
ecause it
’s
I
nd
e
x of
Le
a
rn
i
ng
Sty
le
(I
LS)
of
fer
s
a
pract
ic
al
and
c
oncrete
appr
oach
t
o
determ
ine
the
do
m
inant
le
arn
i
ng
sty
le
s
of
eac
h
le
arn
er
,
pl
us
the
res
ults
of
ILS
can
be
easi
ly
con
ne
ct
ed
to
en
vi
ronm
ents.
Ac
cordin
g
to
F
el
der
-
Sil
ver
m
an
Lea
rn
i
ng
Sty
le
Mod
el
(FSLSM
)
[2
0],
le
ar
ner
s
are
char
act
er
iz
ed
by
their
pr
e
fer
e
nces
in
fo
ur
dim
ension
s as
sh
ow
n
in
T
a
ble 1
:
Table
1.
FSLS
M Lea
r
ning st
yl
e d
i
m
ension
s
Di
m
en
sio
n
s
Sty
le
Key
wo
rds
Grou
p
s
Proces
sin
g
Activ
e
Pref
ers to tr
y
an
d
exp
erience new con
cept
s
Ref
lectiv
e
Likes
to th
in
k
about th
in
g
s b
efore
tak
in
g
action
Perceive
Sen
sin
g
Likes
to u
se
m
e
th
o
d
s
estab
lish
ed
practically an
d
car
ef
u
lly
Intu
itiv
e
Tend
s to
work
qu
ick
ly
and
be in
n
o
v
ativ
e and
us
u
ally
ca
n
m
an
ip
u
late abs
tr
act and
m
ath
e
m
ati
cal
co
n
cept
s
Receiv
in
g
Visu
al
Better assi
m
ilatio
n
of
new in
f
o
r
m
atio
n
thro
u
g
h
grap
h
ics
,
d
e
m
o
n
stratio
n
s, d
iag
ra
m
s,
g
raph
s, et
c.
Verbal
Better assi
m
ilatio
n
th
rou
g
h
vo
calized an
d
wr
itten
word
s
Un
d
erstand
in
g
Seq
u
en
tial
Pref
ers to ass
i
m
ilat
e new kn
o
wled
g
e linearly
an
d
log
ically
Glo
b
al
Pref
ers a
s
y
ste
m
ati
c app
roach
2
.
3
.
Eye
t
r
acki
ng
In
t
he
sim
plest
te
r
m
s,
Ey
e
tracki
ng
or
ga
ze
-
tracki
ng
is
a
te
chnolo
gy
of
rec
ordin
g,
stud
y,
a
nd
m
easur
em
ent
of
ei
the
r
the
c
oor
din
at
es
of
a
hu
m
an
gaze
point.
I
n
ge
ner
a
l,
gaze
re
fer
s
t
o
wh
e
re
y
our
e
ye
s
are
fo
c
us
e
d
a
nd
t
he
refor
e
know
e
xactl
y
your
rea
l
pr
e
fer
e
nc
es.
The
dev
ic
e
us
e
d
to
determ
ine
the
di
recti
on
of
t
he
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
5
,
Oct
ober
201
9
:
4
4
0
8
-
4
4
1
6
4410
po
i
nt g
aze
is c
al
le
d
ey
e
-
track
er.
T
he
Ey
e t
ra
ckin
g
te
ch
nolo
gy h
as
b
ee
n us
ed
in
m
any sci
entifi
c stu
dies
su
c
h
as
neur
os
ci
ence
stud
ie
s,
m
ark
e
ti
ng
an
d
e
-
c
om
m
erce
research,
co
gn
it
ive
sci
ence,
ps
yc
ho
l
og
y,
ca
r
dri
ving,
m
edical
r
esear
ch,
etc.
An
e
xam
ple
of
hig
h
pe
rfo
rm
i
ng
ey
e
trackin
g
syst
e
m
is
dep
ic
te
d
on
Fig
ure
1.
Thr
e
e
vital
par
ts
of
this
syst
e
m
are
[21
]
:
1.
Custom
-
desig
ne
d
se
nsors
:
It
consi
sts
of
c
ust
om
design
e
d
i
nfrar
e
d
pro
j
ect
or
s
,
c
us
tom
iz
e
d
im
age
sens
or
s,
op
ti
cs a
nd c
us
tom
p
ro
ces
sin
g wit
h
em
bed
de
d al
gorithm
s
2.
Adva
nced al
go
rithm
s : It inter
pr
et
s t
he
im
age stream
g
ener
at
ed by t
he se
nso
rs
3.
User
-
ori
ente
d app
li
cat
io
ns
:
An lay
er to act
ivate
the m
any d
iffe
re
nt w
ay
s
the tec
hnol
ogy ca
n be
us
e
d
Figure
1. H
ow
the
ey
e trac
ker
works [
21]
3.
PROP
OSE
D APP
ROAC
H
Each
le
ar
ner
con
cei
ves
an
d
per
cei
ve
s
le
arn
i
ng
diff
e
re
ntly
.
They
hav
e
diff
e
ren
t
stre
ng
t
hs
a
nd
weaknesse
s.
T
her
e
are
se
veral
le
arn
ing
sty
le
s,
and
the
refor
e
we
m
us
t
c
reate
a
cou
rse
that
adap
ts
to
eac
h
le
arn
er
.
In
th
is
way,
the
ind
iv
idu
al
pr
e
fe
ren
c
es
of
the
le
arners
are
res
pected,
an
d
co
ns
e
quently
they
can
ha
ve
bette
r
res
ults.
Learn
i
ng
sty
le
identific
at
io
n
is
tradit
ion
al
ly
done
by
usi
ng
a
quest
io
nn
ai
r
e.
Be
side
the
us
e
of
the
qu
e
sti
onna
ires,
they
are
a
us
efu
l
m
eth
od
to
in
vestigat
e.
H
ow
e
ve
r
,
they
hav
e
a
nu
m
ber
of
m
ajo
r
disad
va
ntages a
s w
el
l:
1.
Ma
y be
un
s
uitable
for s
om
e
kinds
of r
es
po
nd
e
nts
2.
Ther
e
is the
d
a
ng
e
r of q
uestio
nn
ai
re f
at
i
gu
e
3.
Ma
y be ig
nore
certai
n qu
est
i
ons
by r
es
po
nd
e
nt
4.
Ma
y be m
isun
der
st
ood by
res
pondent
5.
Ti
m
e
-
con
s
um
i
ng to
c
om
plete
6.
Etc
.
On
the
oth
e
r
ha
nd,
syst
em
c
annot
re
co
rd
c
hangin
g
of
le
arn
i
ng
sty
le
s
be
cause
it
is
i
m
po
ssible
to
rep
eat
e
dly
ask
le
arn
ers
to
c
om
plete
the
qu
est
io
nn
ai
re
.
Re
cent
stud
ie
s
dem
on
strat
e
t
hat
it
is
po
ssi
ble
to
identify
the
m
os
t
prefe
rr
e
d
l
earn
i
ng
sty
le
s,
thou
gh
bio
m
et
ric
te
chnolo
gi
es
in
par
ti
cula
r
m
ou
se
m
ov
em
ents,
acce
le
ro
m
et
er an
d ey
e trac
king,
et
c.
In
t
his
prese
nt
work
we
pro
pose
a
n
a
ppr
oac
h
to
pro
vid
e
a
dap
ta
ti
on
i
n
le
arn
i
ng
syst
e
m
s
thr
ough
ey
e
trackin
g
te
c
hnology,
this
la
t
te
r
al
lows
aut
om
atical
ly
det
ect
the
prefe
r
red
le
arn
i
ng
st
yl
e
of
eac
h
le
arn
e
r.
Our
pa
per
will
present
t
he
res
ults
of
t
he
bas
el
ine
stu
dy
co
nducte
d
t
o
ide
nt
ify
le
arn
e
r
le
a
r
ning
sty
le
base
d
on
the
Felde
r
-
Sil
ver
m
an
Lear
ni
ng
Sty
le
Mod
e
l
(F
SLSM
)
[
20]
.
This
pap
e
r
c
ons
ist
s
of
t
hr
ee
m
ai
n
pro
cesses
:
The
le
ar
ning
obj
ect
s
a
re
fir
stl
y
descr
ibe
d.
Nex
t,
t
he
le
ar
ning
sty
le
s
are
detect
ed.
Fi
na
ll
y,
the
adap
ta
ti
on
is
pro
vid
e
d.
M
ore detai
ls o
n p
rocesses wil
l
be give
n belo
w
in
the Resea
rch
Me
thodo
l
og
y s
ect
ion
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
Id
e
ntif
yi
ng
le
arnin
g
sty
le
th
r
ough eye tr
ack
ing
tec
hnol
og
y
in ada
ptive
learn
i
ng syste
ms
(
In
ss
af El
G
uaba
s
si
)
4411
4.
RESEA
R
CH
METHO
DOL
OGY
This
sect
io
n
de
scribes
t
he
th
re
e
phases
of
res
earch
m
et
hodolog
y
wh
ic
h
inc
lud
e:
descr
i
bi
ng
le
arn
i
ng
obj
ect
s,
d
et
ect
ing l
ear
ning sty
le
s,
an
d pro
vidi
ng
a
da
ptivit
y.
4
.
1.
Des
cri
bin
g
le
ar
ning
obj
ects
This
phase
c
onsist
s
essenti
al
l
y
to
giv
e
the
use
r
acce
ss
to
a
le
arn
in
g
ob
j
ect
,
base
d
on
an
i
nd
ic
at
io
n
of
their
co
ntent
a
nd
/
or
thei
r
na
ture
(
f
or
m
at
,
t
ype),
a
nd
co
nse
qu
e
ntly
the
m
ai
n
pur
pose
is
to
ena
ble
le
arners
t
o
seek
a
nd
us
e
l
earn
i
ng
ob
j
ect
s,
a
nd
to
e
na
bl
e
them
to
co
m
pi
le
le
arn
in
g
obj
ec
ts
f
or
ea
ch
i
ndivi
dual
le
arn
e
r.
To
de
scri
be
ou
r
le
arn
i
ng
obj
e
ct
s,
we
use
d
t
he
internati
onal
sta
nd
a
rd
LO
M
[22],
w
hich
proposes
a
m
et
adata
m
od
el
descr
i
ption
asso
ci
at
ed
with
pe
da
gogi
cal
obj
ect
s
w
hateve
r
they
a
re
dig
it
al
or
not.
In
m
or
e
ge
ner
al
te
rm
s,
The
IEEE’
s
Lear
ning
O
bject
Me
ta
data
(LO
M),
use
d
to
desc
ribe
a
le
arn
in
g
ob
je
ct
us
ed
to
cre
at
e
of
well
structu
red
descr
ipti
on
of
le
arn
in
g
res
ources
,
to
sup
port
the
re
us
a
bili
ty
of
le
ar
ning
obj
ect
s,
to
fa
ci
li
ta
te
their
inte
ropera
bili
ty
,
and
to
ai
d
disc
overa
bi
li
t
y,
usual
ly
in
the
c
on
te
xt
of
a
da
ptive
le
a
rn
i
ng
syst
em
s
(A
L
S).
A
r
ep
rese
ntati
on
of the
hiera
rc
hy of elem
ents in
the
LOM
da
ta
m
od
el
is s
hown in F
ig
ur
e
2.
Figure
2.
The
le
arn
i
ng ob
j
ect
m
et
adata sche
m
e
The
nee
d
for
adap
ta
ti
on
of
l
earn
i
ng
ob
j
ect
s
to
t
he
nee
ds
of
t
he
le
ar
ner,
we
a
dd
e
d
an
el
em
ent
to
te
chn
ic
al
cat
eg
or
y
of
the
LO
M
data
m
od
el
.
Of
c
ourse,
this
el
e
m
ent
is
the
le
arn
in
g
sty
le
s
accor
ding
to
F
el
der
-
Sil
ver
m
an
Lea
rn
i
ng
Sty
le
Mod
el
(F
S
LSM)
[20],
bu
t
our
fo
c
us
will
be
especial
ly
on
the
Vis
ual
/
Verbal
dim
ension
.T
he
le
arn
in
g
obj
ec
ts
in
ou
r
syst
e
m
are
la
beled
as
rep
r
esente
d
in
T
able
2.
T
his
descr
ipti
on
is
based
on the t
heoreti
cal
d
escri
ptio
ns ab
out l
eanin
g st
yl
es’
cha
ract
erist
ic
s o
f
FSL
SM.
Table
2
. L
abel
s of
our
le
ar
ning
obj
ect
s
Visu
al
Verbal
I
m
ag
e
Diag
ra
m
Vid
eo
Bo
ard
An
i
m
atio
n
Graph
Si
m
u
latio
n
Slid
esh
o
w
Hy
p
ertext
Co
n
f
erence vid
eo
Au
d
io
Text
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
5
,
Oct
ober
201
9
:
4
4
0
8
-
4
4
1
6
4412
Af
te
r
seei
ng
our
le
ar
ning
obj
e
ct
desi
gn,
i
n
t
he
ne
xt
sect
io
n
w
e
will
prese
nt
how
we
c
an
autom
at
ic
ally
e
stim
at
e lea
rn
ers'
le
arn
ing st
yl
es
?
4
.
2.
De
tectin
g l
earnin
g
s
ty
le
s
It
is
ve
ry
im
p
or
ta
nt
to
intr
oduce
the
ada
pt
at
ion
t
o
the
onli
ne
le
ar
ni
ng
syst
e
m
by
de
ve
lop
in
g
the
m
echan
ism
to
identify
the
le
arn
i
ng
sty
le
s
to
c
om
ply
with
the
re
quirem
e
nts
of
F
SLSM
[19].
Howe
ve
r
the
pro
blem
a
ti
c are:
a.
It
is
i
m
po
ssibl
e
to
re
peatedly
ask
eac
h
l
ear
ne
r
to
c
om
plete
the
quest
io
nnai
re
for
ti
m
e
a
nd
c
os
t
c
on
ce
r
n,
wh
e
reas
the
s
yst
e
m
has
to
face
the
chall
eng
e
of
recor
ding
the
cha
ngin
g
of
le
arn
i
ng
sty
le
durin
g
the
le
arn
in
g.
b.
The q
uestionna
ires h
a
ve
a
nu
m
ber
o
f
m
ajor disad
va
ntages
i
n gen
e
ral, as
s
how
n pr
e
v
io
usl
y.
In
ver
y
ge
ner
a
l
te
r
m
s,
an
autom
atic
m
e
tho
d
f
or
trac
king
le
arn
i
ng
sty
le
s
is
the
essenti
al
par
t
of
a
n
eff
ic
ie
nt m
assive
ALS.
In
t
he
prese
nt
stud
y,
we
us
ed
an
ey
e
gaze
r
ecorder
t
o
rec
ord
t
he
tim
e
that
the
par
ti
ci
pa
nt
s
gazed
a
t
te
xt
-
base
d
or
gr
a
ph
ic
-
ba
s
e
d
le
arn
i
ng
obj
ect
s.
T
he
r
efore,
we
c
om
par
ed
the
resu
lt
s
wit
h
the
resu
lt
s
obta
ine
d
th
r
ough
the
qu
e
sti
onnaire
of
t
he
FS
LSM
.
I
n
this
fi
rst
stud
y,
we
f
oc
us
e
d
es
pecial
ly
on
t
he
Visu
al
/
Verbal
d
im
ension
.
The
m
ajo
r
ste
ps
to
ac
hieve
the
com
par
iso
n
to
fin
d
the
r
el
at
ion
sh
i
p
am
ong
the
Felde
r
-
Sil
ver
m
an
Learn
i
ng Sty
le
and the
eye
tra
ckin
g
a
re:
a.
The
le
ar
ner
c
om
plete
the
Felder
-
Sil
ver
m
an
I
nd
e
x
of
Le
arn
i
ng
Sty
le
s
(I
LS
)
quest
io
nnai
re
to
deter
m
ine
wh
et
her he/s
he
is v
is
ual or
ve
rb
al
b.
The
syst
em
pr
opos
es
to
t
he
le
arn
e
r
a
le
arn
i
ng
obj
ect
w
hi
ch
is
disp
la
ye
d
in
te
xt
an
d
gr
a
phic
form
.
Fo
r
that
pur
pose,
t
he
scree
n
is
div
ide
d
int
o
two
disti
nct
pa
rts.
O
ne
half
of
t
he
scree
n
offe
re
d
te
xt
ual
represe
ntati
on
of
le
a
rn
i
ng
ob
je
ct
wh
il
st
the
seco
nd
half
of
the
sc
reen
off
ered
grap
hical
re
prese
ntati
on
of
the sam
e lea
rn
ing o
bject
.
c.
Ey
e
trackin
g
t
echnolo
gy
is
use
d
in
t
his
ste
p
to
determ
ine
the
point
-
of
-
ga
ze
(POG
)
of
the
le
arn
e
r
on
the
screen
.
d.
Gen
e
rati
on
of
t
he results
of ey
e tracki
ng expe
rim
ent
e.
Com
par
ison
of
the
res
ults
of
the
Felde
r
-
Sil
ve
rm
an
Index
of
Lea
rn
i
ng
Sty
le
s
(ILS)
quest
ionnaire
a
nd
th
e
resu
lt
s
obta
ine
d via ey
e trac
kin
g t
ech
no
l
og
y.
The
res
ults
of
this
exp
e
rim
ent
will
be
pr
esented
an
d
disc
usse
d
in
Re
su
lt
s
and
Disc
us
si
on
sect
io
n.
In
t
he
ne
xt
sect
ion
, w
e w
il
l
see
how
to provi
de
ada
ptati
on
i
n
the o
nline
le
arn
i
ng
syst
em
by
the
use
of
le
arn
i
ng
obj
ect
s a
nd lea
rn
e
r
le
ar
ning st
yl
es
?
4.
3
.
Pr
ov
idi
n
g ad
ap
ti
vit
y
Af
te
r
desc
ribi
ng
t
he
le
ar
ni
ng
obj
e
ct
s
an
d
detect
in
g
th
e
le
arn
e
r
le
ar
ning
sty
le
s,
we
are
now
com
m
it
te
d
to
pr
ovidi
ng
le
ar
ne
rs
with
an
ada
pted
co
urse
accor
ding
to
his/her
I
ndivid
ual
le
arn
in
g
prefe
r
ences
.
The
a
da
ptati
on
ca
n
be
e
xpla
ined
as:
T
he
le
arn
er
prof
il
e
is
re
pr
es
ente
d
as
vecto
r,
a
nd
al
s
o
eac
h
le
arn
i
ng
obj
ect
s
i
n
the
co
ur
se
is
r
ep
resen
te
d
as
a
vecto
r,
a
nd
th
en
the
le
ar
ner
interest
pr
e
di
ct
ion
is
ac
hie
ved
by
cal
culat
ing
sim
il
arit
y value b
e
tween t
he
le
ar
ne
r profile
vecto
r
a
nd the lea
rn
i
ng ob
j
ect
s
vector.
Ther
e
ar
e
m
any
si
m
il
arity
m
easur
e
s
avail
able,
but
the
m
os
t
widely
-
us
e
d
m
easur
e
in
the
fiel
d
of
inf
or
m
at
ion
r
es
earch
is
the
cos
ine sim
i
la
rity
.
Give
n
tw
o
vec
tors,
V
LO
=<L
O
LL
,
L
O
P
,
LO
Lg
,
LO
LS
>
desc
r
ibing
the
le
arni
ng
obj
ect
s
an
d
the
ve
ct
or
V
LP
=<LP
LL
,
L
P
P
, LP
Lg
, LP
LS
>
descr
i
bing th
e lea
rn
e
r profil
e,
Wh
e
re:
LL
: R
epr
ese
nt
s the le
vel (
e
.g.
Begin
ner,
In
te
rm
ediat
e, o
r
E
xp
e
rt).
P
: R
ep
resen
t
s the
plate
form
(
e.
g. Com
pu
te
r,
m
ob
il
e, table
ts, etc
.)
.
Lg
: R
e
pr
ese
nt
s the lan
guage
(
e.
g. A
rab
ic
,
F
ren
c
h,
En
glish
,
etc.).
LS
:
Re
prese
nt
s the lear
ning
sty
le
s (
e.g.
Visu
al
or
Verbal)
4
1
2
4
1
2
4
1
4
1
)
,
c
os
(
i
i
i
i
i
i
i
i
LP
LO
LP
LO
LP
LO
V
V
Wh
e
n
th
e
res
ul
ti
ng
sc
or
e
is
1
the
le
arn
i
ng
ob
j
ect
LOi
a
nd
the
le
ar
ning
pr
ofi
le
LPi
are
ide
ntica
l,
an
d
0
if
the
re
is
no
thing
i
n
com
m
on
betwee
n
the
m
.
The
qu
est
io
n
no
w
bein
g
as
ked
is
w
hat
are
the
ne
xt
ste
ps
after
cal
culat
ing
sim
il
arit
y
value
?
Of
co
urse,
t
here
are
m
any
ste
ps
al
ong
t
he
road
to
en
surin
g
adap
ta
ti
on
i
n
l
earn
i
ng
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
Id
e
ntif
yi
ng
le
arnin
g
sty
le
th
r
ough eye tr
ack
ing
tec
hnol
og
y
in ada
ptive
learn
i
ng syste
ms
(
In
ss
af El
G
uaba
s
si
)
4413
syst
e
m
s
fo
r
ex
a
m
ple:
Ar
ra
nge
m
ent
of
the
diff
e
re
nt
sel
ect
ed
le
arn
i
ng
ob
j
ect
s,
but
we
will
exp
la
in
th
e
m
in
oth
e
r
pa
per
be
cause
we
f
oc
use
d
in
this
w
ork
on
ide
ntifyi
ng
le
ar
ning
sty
le
s
autom
a
ti
c
a
lly.
The
resu
l
ts
and
discuss
i
on
sect
ion
ans
we
rs
one
m
a
in
qu
est
i
on
:
Ca
n
we
under
sta
nd
le
ar
ne
rs'
le
arn
in
g
sty
le
s
by
ey
e
trackin
g
te
chnolo
gy in
adap
ti
ve
lear
nin
g sy
ste
m
s (
ALS
)
?
5.
E
X
PERI
MEN
T
5.1.
P
art
ic
ipa
nt
s
Mult
iple
par
ti
c
ipants
to
ok
part
in
this
ex
pe
rim
ent.
H
ow
e
ve
r,
du
e
to
pro
ble
m
relat
ed
to
cal
ibrati
on,
we
go
t
se
ve
n
va
li
d
data
(2
m
a
le
and
5
fem
ale)
.
T
heir
ave
ra
ge
a
ge
was
30.
5
ye
ars
ol
d
a
nd
m
os
t
of
them
wer
e
from
gr
ad
uate
stud
ie
s.
Thei
r
var
ie
d
f
ro
m
physi
cs,
m
edici
n
e,
an
d
c
om
pu
te
r
sci
ence
.
T
w
o
of
t
hem
had
norm
al
visio
n
a
nd the
oth
e
r had
corre
ct
ed
to
norm
al
v
isi
on (
t
hey
w
ere
wear
in
g gl
asses).
5.2.
Pr
ocedur
e
The
w
ho
le
e
xperim
ent
la
sted
ap
prox
im
at
e
ly
20
-
30
m
inu
te
s,
dep
e
ndin
g
on
pa
rtic
ipan
t’s
read
i
ng
sp
ee
d,
rea
ding
com
pr
ehe
ns
io
n
and
cal
ibr
at
io
n
proces
s.
E
xperim
en
t
pr
ocedur
e
is
sho
wn
i
n
F
ig
ur
e
3
.
T
he
first
sect
ion
of
the
exp
e
rim
ent
star
te
d
with
the
I
LS
qu
e
sti
onnai
re.
The
n,
afte
r
finish
i
ng
this
pa
rt,
the
par
ti
ci
pan
t
i
s
aske
d
to
cal
ib
r
at
e
the
ey
e
-
tra
cker.
A
fter
wards,
he
is
pray
e
d
to
le
ar
n
the
NLP
c
o
ncep
t,
on
ce
t
he
cal
ibr
at
ion
is
qu
al
ifie
d.
I
n
these
m
o
m
ents
,
the
gaze
tra
ckin
g
syst
e
m
is
reco
r
ded
t
he
par
ti
ci
pa
nt’s
ey
e
m
ov
em
ents
.
Finall
y, the
res
ults are
ge
ner
at
ed
a
nd s
howe
d.
Figure
3
.
Expe
rim
ent p
ro
ce
du
re
5
.
3
.
Equi
pme
nt
an
d
setti
n
g
We
rec
orde
d
ey
e
m
ov
em
ent
s
from
par
ti
cip
ants
us
in
g
a
GazeRe
co
rd
e
r
[23]
gaze
trac
king
syst
em
,
wh
ic
h
is c
har
a
ct
erized
by:
1.
Au
t
om
atical
ly
record
s
us
in
g o
rd
i
nar
y
Web
Ca
m
2.
No h
a
r
dw
a
re
r
equ
i
red
-
only
w
ebcam
an
d c
om
pu
te
r
3.
No d
i
rect p
hys
ic
al
co
ntact
with
us
ers
for
gaz
e tracki
ng
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
5
,
Oct
ober
201
9
:
4
4
0
8
-
4
4
1
6
4414
4.
Ey
e tracki
ng
in
real t
i
m
e
5.
Gaze trac
king
us
a
bili
ty
test
in
g
6.
Creat
e ex
per
im
ents
with a si
ngle
cl
ic
k
7.
Heatm
ap
su
r
fa
ce is
ge
ner
at
e
d dynam
ic
al
ly
with a
dap
ti
ve
t
i
m
e w
indow
GazeRe
co
rd
e
r
has
a
nu
m
ber
of
diff
e
re
nt
c
onfig
ur
at
io
n
pa
ram
et
ers
that
con
t
ro
l
ey
e
tracki
ng,
as
ind
ic
at
ed
in
Ta
ble
3.
Table
3.
C
onfi
gurati
on
par
am
et
ers
Measu
re
Valu
es
Ti
m
e
wind
o
w
5
0
0
m
s
Gaze
Plo
t
Size
12%
Ad
ap
tiv
e exten
d
tim
e
win
d
o
w du
ring
static scr
een
Yes
Exten
d
ti
m
e
win
d
o
w when
scre
en
app
eara
n
ce
ch
an
g
e less th
an
5%
Sh
o
w
m
o
u
se
curso
r
No
Sh
o
w f
ace
No
(FOV
)
ca
m
f
ield
o
f
view v
ercic
al
4
2
deg
W
eb
Ca
m
lo
catio
n
Res
o
lu
tio
n
6
4
0
*
4
8
0
Calib
ration
T
y
p
e
5
po
in
ts
Pu
p
il dis
tan
s
5.2
m
5
.
4
.
In
terf
ace
d
escri
pt
io
n
The
ex
per
im
ental
validat
ion
of
ou
r
ap
proac
h
is
done
by
presenti
ng
the
c
on
ce
pt
of
Ne
uro
-
li
nguisti
c
pro
gr
am
m
ing
(
NLP)
to
the
pa
rtic
ipants.
O
f
course,
they
ha
ve
no
pri
or
in
form
ation
or
knowle
dge
ab
out
this
con
ce
pt.
The
s
creen
is
div
ide
d
int
o
t
wo
dis
t
inct
pa
rts
as
s
how
n
in
F
i
gur
e
4
.
On
e
hal
f
of
the
scree
n
offer
e
d
te
xtu
al
r
e
prese
ntati
on
of
NLP w
hilst
the sec
ond half
of the
screen
off
e
re
d her
grap
hical
re
pr
ese
ntati
on.
Figure
4
.
Interf
ace re
pr
ese
ntati
on
5
.
5
.
Res
ults
GazeRe
co
rd
e
r
[23]
gaze
trac
ki
ng
syst
e
m
us
es
an
ordi
nar
y
Web
Ca
m
to
track
an
d
rec
ord
wh
at
pa
rt
of
the
scree
n
the
par
ti
ci
pa
nt
w
as
lookin
g
at
,
and
the
n
ge
ner
at
e
d
a
sm
oo
th
heatm
ap
wh
ic
h
is
a
gr
a
ph
ic
a
l
represe
ntati
on
of
this
data.
Th
ere
a
re
se
ver
al
diff
e
re
nt
ad
v
a
nt
ages
of
heatm
ap.
I
nd
ee
d
it
is
sim
ple,
easy
to
us
e
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
Id
e
ntif
yi
ng
le
arnin
g
sty
le
th
r
ough eye tr
ack
ing
tec
hnol
og
y
in ada
ptive
learn
i
ng syste
ms
(
In
ss
af El
G
uaba
s
si
)
4415
and
the
re
su
lt
s
are
easy
to
in
te
rp
ret.
In
ou
r
case,
the
heatm
aps
ind
ic
at
e
wh
e
re
the
le
ar
ner
ha
d
vie
we
d
the
screen
?
He
h
a
d l
ooke
d
at
te
xt
and
/
or grap
hic
s
?
The
heatm
aps
f
or
each
us
er
in
dicat
e
that
there
i
s
a
si
gnific
ant
di
ff
e
re
nce
betwee
n
vi
su
al
an
d
ve
rbal
le
arn
er
s.
Indee
d
the
res
ults
dem
on
strat
ed
t
hat
the
verbal
le
arn
e
rs
acc
ordi
ng
t
o
I
LS
qu
e
sti
onnaire
of
t
he
Felder
-
Sil
ve
rm
an,
s
pe
nt
a
m
ajorit
y
of
their
tim
e
loo
king
at
the
te
xtu
al
par
t
of
t
he
sc
r
een
as
il
lu
strat
ed
i
n
F
igure
5(a)
.
More
ov
e
r
the
visu
al
le
arn
e
rs
sp
en
d
m
or
e
t
i
m
e
loo
ki
ng
at
the
gr
a
ph
ic
al
par
t
of
the
sc
reen
as
sh
ow
n
in
Fi
gur
e 5
(
b)
.
(a)
(b)
Figure
5
.
V
e
rbal
and
vis
ual
le
arn
e
r heat
m
ap
6.
CONCL
US
I
O
N
AND
F
UT
U
RE W
ORK
The
le
ng
t
h
of
a
fixati
on
is
one
of
t
he
m
os
t
im
po
rtant
ch
ara
ct
erist
ic
s
to
under
sta
nd
hu
m
an
be
hav
i
or,
in
fact,
ey
e
m
ov
em
ent
recor
ding
du
rin
g
re
adin
g
is
ve
ry
im
po
rtant
inf
orm
at
ion
in
ada
pt
ive
le
arn
in
g
s
yst
e
m
s
(A
L
S)
.
An
a
ppr
oac
h
for
I
de
ntifyi
ng
le
ar
ni
ng
sty
le
through
ey
e
track
ing
te
ch
no
l
ogy
in
ALS
ha
s
been
pr
ese
nted
w
hich
ad
dr
e
sses
the
nee
d
to
(1)
Descr
i
be
le
ar
ning
ob
j
ect
s
in
ada
pted
way
in
differe
nt
ki
nd
of
courses
,
(
2)
Prov
i
de
a
n
aut
om
at
ed
m
et
ho
d
for
de
te
ct
ing
le
arn
in
g
sty
le
s
base
d
on
ey
e
trackin
g
te
c
hnol
og
y,
and
(3)
Pro
vide
an
ALS
us
in
g
le
arn
i
ng
sty
le
s
and
le
ar
ning
obj
ect
s.
W
e
a
r
e
inv
est
igati
ng
sever
al
exte
nsi
on
s
to
this w
ork:
a.
Im
ple
m
enting
the
pro
po
se
d
a
ppr
oac
h on m
ult
iple com
pu
ti
ng
p
la
tf
or
m
s (
cr
oss
-
platf
orm
)
b.
In
cl
ud
i
ng o
t
hers b
i
om
et
ric te
c
hnologies s
uc
h as
m
ou
se m
ove
m
ents
c.
Com
pleti
ng
th
e Pro
vid
in
g A
da
ptivit
y su
bs
ec
ti
on
REFERE
NCE
S
[
1
]
El
Guab
assi,
I
.
,
Al
Achha
b,
M.,
Je
ll
ouli,
I
.
,
El
Mohaji
r
,
B.
E
.
,
“
Rec
om
m
ende
r
s
y
s
te
m
for
ubiqu
it
o
us
le
arn
ing
base
d
on
decision
tr
ee,
”
In
In
formation Sci
en
ce
and
Tec
hnology
(
CiSt
)
,
2016
4th
I
EE
E
I
nte
rnational
Col
loqui
um
on
,
IE
E
E
,
Oct
2016
,
pp.
53
5
-
540.
[
2
]
El
Guaba
ss
i
,
I.,
Al
Achha
b,
M.,
Jell
oul
i
,
I.,
Mohaji
r,
B.
E
.
E.
,
“
Conte
x
t
-
Aw
are
R
ec
om
me
nder
S
y
stems
for
Le
arn
ing,”
Int
ernati
onal Journal
of
In
formation
S
ci
en
ce and
Tech
nology
,
vol
.
1(1
)
,
pp
.
17
-
25
,
201
8
.
[
3
]
Guaba
ss
i,
I.
E
.
,
Achha
b,
M
.
A.,
Jell
ou
li
,
I
.
,
Mohaji
r,
B
.
E.
E
.
,
“
Towa
r
ds
ada
pti
ve
ub
iqui
tous
learni
n
g
s
y
stems
,”
In
te
rn
ati
onal Journal of
Knowledge
an
d
Learning
,
vo
l.
11(1),
pp
.
3
-
23
,
2016
.
[
4
]
El
Guaba
ss
i,
I.,
Bousale
m
,
Z
.
,
Al
Achha
b,
M.
,
JE
LL
OU
LI,
I
.
,
Mohaji
r,
B.
E.
E.
,
“
Personali
z
e
d
ada
pt
ive
content
s
y
stem for cont
e
xt
-
awa
re
ubiqu
itous
le
arn
ing
,”
Pr
oce
dia
Compute
r Sc
ie
n
ce
,
vol
.
1
27,
pp
.
444
-
453
,
2018
.
[
5
]
Alamudun,
F.,
Yoon,
H.
J.,
Hu
dson,
K.
B.
,
Mo
rin‐Duc
ote,
G.,
Ham
m
ond,
T.
,
Toura
ss
i
,
G.
D.
,
“
Frac
ta
l
an
aly
si
s
of
visual
sea
rch
a
ct
ivit
y
for
m
ass
det
ec
t
ion
dur
ing
m
amm
ogra
phic
scre
eni
ng
,”
Me
dic
a
l
phy
sics
,
vol.
44(3)
,
pp.
832
-
846
,
20
17
.
[
6
]
Holm
qvist
K.,
N
y
ström
M.,
Anderss
on
R.
,
Dewhurs
t
R.
,
J
a
rodz
ka
H.,
Va
n
de
W
ei
j
er
J.
,
“
E
y
e
tra
ck
ing
:
A c
om
pre
hens
iv
e
guid
e to
m
et
ho
ds a
nd
m
ea
sures
,”
OU
P
Ox
ford
,
2011
.
[
7
]
Duchows
ki
A.
T
.
,
“
E
y
e tra
ck
ing m
et
hodolog
y
,”
Theory
and
pra
ct
i
ce
,
vol
.
328
,
200
7
.
[
8
]
Vijay
R
aj
ann
a,
“
Gaz
e
T
y
p
ing
T
hroug
h
Foot
-
Opera
t
ed
W
ea
r
able
Device,”
In
Pr
oce
ed
ings
of
the
18th
In
te
rnatio
nal
ACM
SIGACCE
SS
Confe
ren
ce
o
n
Computers
and
Acce
ss
ibi
l
it
y
(
ASSE
TS
’16)
,
ACM
,
New
York,
NY
,
US
A
,
2016
,
pp.
345
–
346
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
5
,
Oct
ober
201
9
:
4
4
0
8
-
4
4
1
6
4416
[
9
]
Raj
ann
a
V.,
a
nd
Hammond
T.
,
“
GA
WS
CHI:
gaz
e
-
aug
m
ent
ed,
wea
r
a
ble
-
suppleme
nt
e
d
computer
-
hum
an
int
er
ac
t
ion,”
In
Proceedi
ngs
of
the
Nint
h
Bi
enni
a
l
ACM
S
ymposium
on
Ey
e
Tr
ack
ing
Re
search
&
Appl
ic
a
ti
ons
,
ACM
,
Mar. 2016
,
pp.
233
-
236
.
[
1
0
]
Raj
ann
a
V.,
Ma
l
la
A.
H.
,
Bhag
at
R.
A.,
H
amm
on
d
T.
,
“
D
y
Gaz
ePass:
A
gaz
e
g
est
ure
-
base
d
d
y
n
a
m
ic
aut
hen
ti
c
ati
on
s
y
stem
to
count
e
r
shoulder
surfin
g
and
vid
eo
an
aly
sis
a
tt
a
cks
,”
In
Id
ent
ity,
Sec
uri
t
y,
and
Be
ha
vi
or
Anal
ysis
(
ISBA
)
,
2018
IEEE
4
th
I
nte
rnational
Co
nfe
renc
e
,
on
I
EEE
,
Jan
2018
,
pp.
1
-
8
.
[
1
1
]
Raj
anna
V
.
,
Po
lsley
S.
,
T
aele
P.,
Ham
m
ond
T
,
“
A
gaze
g
esture
-
base
d
user
aut
hen
ti
c
at
ion
sy
stem
to
counte
r
shoulder
-
surfing
at
tacks
,”
In
Pr
oce
ed
ings
of
the
2017
CHI
Con
fe
renc
e
Exten
d
e
d
Abstract
s
on
Hum
an
Fac
tors
in
Computing
Syst
e
ms
,
ACM
,
2017,
pp.
1978
-
1986.
[
1
2
]
Alamudun
F.
T.
,
Ham
m
ond
T.,
Yoon
H.
J.,
Tou
rassi
G.
D.,
“
Ge
om
et
r
y
and
Gest
ure
-
Based
Fea
tu
res
from
Sacc
ad
ic
E
y
e
-
Movem
ent
as
a
B
iom
et
ric
in
Radi
o
log
y
,”
In
Inte
rnat
ional
Confe
ren
ce
on
Augme
nt
ed
Co
gnit
ion
,
Springe
r,
Cham
,
Jul
2017
,
pp.
123
-
138
.
[
1
3
]
Bec
ker
W
.
,
and
Jürgens
R.
,
“
An
ana
l
y
sis
of
the
sac
c
adi
c
s
y
st
em
b
y
m
e
ans
of
dou
ble
step
stim
uli,
”
V
ision
researc
h
,
vol.
19(9)
,
pp
.
9
67
-
983
,
1979
.
[
1
4
]
Kaul
P.,
R
aj
ann
a
V.,
Ham
m
ond
T.,
“
Expl
oring
users'
per
ceive
d
ac
t
ivi
t
ie
s
in
a
sketc
h
-
base
d
i
nte
lligent
tut
or
i
ng
s
y
stem
through
e
y
e
m
ovement
dat
a
,”
In
Proce
edi
ngs
of
the
ACM
Symposium
on
Appl
ie
d
Pe
rce
p
ti
on
,
AC
M
,
Jul
2016
,
pp
.
13
4
-
134.
[
1
5
]
Ra
y
n
er
K
.
,
Car
l
son
M.,
Fraz
ie
r
L.
,
“
The
intera
ct
ion
of
s
y
nt
ax
and
sem
ant
ic
s
during
sente
nc
e
proc
essing:
E
ye
m
ovements
in
t
he
ana
l
y
sis
o
f
sem
ant
ic
a
lly
bias
ed
sente
nce
s,
”
J
ournal
of
ve
rbal
le
arning
and
ve
rbal
behavior
,
vol.
22(3)
,
pp
.
3
58
-
374
,
1983
.
[
1
6
]
Le
arn
ing
Techn
olog
y
S
ta
nd
ard
s
Comm
it
te
e
,
“
Draft
Standa
rd
for
Le
arn
ing
O
bje
c
t
Meta
da
ta,
”
IEE
E
Standa
rd
1484.
12.
1
,
New
Y
ork:
Inst
it
ut
e
of
Elec
tri
cal and
E
le
c
tronic
s E
ng
in
ee
rs
,
2002
.
[
1
7
]
W
il
e
y
Dav
id
A
,
"Connec
ti
ng
L
ea
rning
Obje
ct
s
to
Instruc
ti
ona
l
Design
The
or
y
:
A
Defi
nit
ion,
A
Meta
phor,
an
d
A T
axonom
y
"
,
i
n
Wiley
,
Davi
d
A.
(
DO
C)
,
The
Instructi
onal
Us
e
of
Learning
Obj
ec
ts:
Onl
ine V
er
sion
,
2002
.
[
1
8
]
Reha
k,
Daniel
R.
;
Mason,
Robin
,
"Eng
agi
ng
w
it
h
th
e
L
ea
rn
ing
Objec
t
E
conomy
",
in
Lit
tl
e
john
,
Allison,
Re
usin
g
Online
Re
sour
ces
:
A
Sustainab
le
Approach
to
E
-
L
earning
,
London
:
Kogan
Pag
e, p
p.
22
-
30
,
2003
.
[
1
9
]
Bec
k,
Rob
ert
J.
,
"W
hat
Are
Le
ar
ning
Objects
?
,
"
Learning
Objec
t
s
,
Cent
er
for I
nt
ernati
onal
Edu
c
ati
on
,
Univ
ersity of
W
isconsin
-
Mi
lwauke
e
,
r
et
ri
eve
d
apr
29
2008
.
[
2
0
]
Felde
r,
R.
M.
,
Silve
rm
an,
L
.
K
,
“
Le
arn
ing
and
t
ea
ch
ing
s
t
y
le
s
i
n
engi
nee
r
ing
e
duca
t
ion.
Engi
n
ee
ring
edu
cation
,
”
Journal
of Engin
ee
ring
Edu
ca
t
ion
,
vol
.
78(7)
,
pp
.
674
-
681
,
1988
.
[
2
1
]
Babe
nko
A.
,
Kot
enko
A.
,
“
The Ey
e
Tr
ibe,”
Docto
ral
d
isserta
t
ion,
Sum
y
St
ate
Uni
ver
sit
y
,
2016
.
[
2
2
]
IEE
E
Learni
ng
Te
chno
log
y
Stan
dar
ds Com
m
it
tee,
I
EEE
LOM w
orking
dra
f
t
6
.
1.
,
2001.
[
2
3
]
Deja
,
Sz
y
m
on
,
Gaz
eRecorde
r
.
Avail
ab
le
:
htt
ps:
//
ww
w.
f
aceboo
k.
com /
gazereco
rde
r/
,
2016
.
Evaluation Warning : The document was created with Spire.PDF for Python.