Recommender Systems
-2-
Feedback Detection
Non-Intrusive
Intrusive
user diharapakan
meniliai item atau
page -pate
-3-
Content/Profile-based
Red
Mars
Found
ation
Jurassic
Park
Lost
World
2001
Machine
Learning
User
Profile
Neuromancer
2010
Difference
Engine
Collaborative Filtering
User
Database
A
B
C
:
Z
9
3
:
5
A
B
C 9
: :
Z 10
A
B
C
:
Z
5
3
A
B
C 8
: :
Z
:
7
Correlation
Match
Active
User
A 9
B 3
C
. .
Z 5
A 6
B 4
C
: :
Z
A
B
C
:
Z
9
3
:
5
A 10
B 4
C 8
. .
Z 1
A 10
B 4
C 8
. .
Z 1
Extract
- 4 -C
Recommendations
Collaborative Filtering
User
Database
A 9
B 3
C
: :
Z 5
A
B
C 9
: :
Z 10
sis
y
l
ana o the
n
tio ilar t ers
a
l
re
r
sim clust
o
C
s
i
n
e
o
r
i
e
t
H ocia
Ass lysis!
Ana
Active
User
A 5
B 3
C
: :
Z 7
A
B
C 8
: :
Z
Correlation
Match
A
B
C
.
Z
9
3
.
5
A 6
B 4
C
: :
Z
A
B
C
.
Z
10
4
8
.
1
A 9 A 10
B 3 B 4
C
C 8
: : . .
Z 5 Z 1
Extract
Recommendations
C
-5-
Item-User Matrix
-6-
-7-
Item2
Item3
Item4
Item5
Alice
User1
User2
User3
User4
-8-
Item2
Item3
Item4
Item5
Alice
User1
User2
User3
User4
-9-
ca ,u
covar(ra , ru )
ra ru
rx
i 1
x ,i
i 1
a ,i
ra )(ru ,i ru )
m
(r
rx
2
(
r
r
)
x ,i x
i 1
Rating Prediction
Prediksi penilaian, pa,i, untuk setiap i, untuk aktif
user , a, dengan menggunakan k user lain
(neighbor) dengan u {1,2,k}.
ca ,u
k
pa ,i ra
ri,j adalah user I penilaian terhadap item j
w
u 1
a ,u
covar(ra , ru )
ra ru
(ru ,i ru )
| w
u 1
a ,u
Significance Weighting
Significance weights, sa,u, didasarkan
pada jumlah co-rated items, m.
wa ,u sa ,u ca ,u
s a ,u
1 if m 50
m
if m 50
50
ca ,u
covar(ra , ru )
ra ru
Item1
Item2
Item3
Item4
Item5
Alice
User1
sim = 0,85
User2
sim = 0,00
User3
sim = 0,70
User4
sim = -0,79
- 15 -
Pearson correlation
Melihat berbagai perilakuk user
- 16 -
Model-based approaches
Didasarkan off-line pre-processing atau tahap "model-learning"
Pada saat dijalankan, hanya model yang sudah buat digunakan
untuk prediksi
models diupdate /dilatih ulang secara periodik
Banyak teknik yang digunakanlarge variety of techniques used
Membangun dan update model komputasinya sangat mahal
item-based CF adalah contoh pendekatan berbasis model
- 17 -
Contoh
Perhatikan items yang sama dengan Item5
Tentukan utuk user Alice's terhadapa item ini untuk prediksi
Item5
Item1
Item2
Item3
Item4
Item5
Alice
User1
User2
User3
User4
1
- 18 -
- 19 -
Making predictions
Fungsi prediksi
Neighborhood dibatasi
Tidak semua neighbor dihutung prediksinya
- 20 -
Probabilistic models
clustering models, Bayesian networks, probabilistic Latent
Semantic Analysis
Costs of pre-processing
Usually not discussed
Incremental updates possible?
- 21 -
- 22 -
Matrix factorization
M U V T
- 23 -
ty
et
Pr
an
om
W
ve
Lo
Mar
0.70
y
0.06
r
Prediction:
ui ru
Sue 0.31 0.93
Vk T
y
ra
0.23
tP
Ea
-0.44
s
in
Tw
0.47
0.30
rd
Ha
Bob
Dim
2
e
Di
Alic
e
Dim
1
or
at
in
Uk
M k U k k Vk
rm
Te
SVD:
Dim
1
Dim
2
0.58
0.26 0.18
0.66
0.36
k Dim Dim
1
2
= 3 + 0.84 = 3.84
Dim
2
3.23
- 24 -
Sue
Terminator
Twins
Bob
Mary
- 25 -
- 26 -
Ite
m2
Ite
m3
Ite
m4
Ite
m5
User
1
0
2
support (2/4), confidence (2/2) (without Alice)
User
0
0
3
Simplest approach
Alice
transform 5-point ratings into binary
ratings (1 = above user average) User
Item1 Item5
Probabilistic methods
- 28 -
Item2
Item3
Item4
Item5
Alice
User1
User2
User3
User4
More to consider
- 30 -
2008:
- 31 -
(
y
y
)
y that
has to be determined such
is minimal.
A huge variety of different learning
strategies have been applied trying to
estimate f(x)
Non parametric neighborhood models
MF models, SVMs, Neural Networks, Bayesian Networks,
- 32 -