Why Does Collaborative Filtering Work?

odecrackΤεχνίτη Νοημοσύνη και Ρομποτική

29 Οκτ 2013 (πριν από 3 χρόνια και 9 μήνες)

73 εμφανίσεις

Online Supplement for

Why Does Collaborative Filtering Work?



Transaction
-
based Recommendation Model Validation
and Selection by Analyzing Bipartite Random Graphs


Zan Huang

Department of
Supply Chain and Information Systems, Pennsylvania State University
, University Park, PA, 16802,
USA, zanhuang@psu.edu

Daniel Dajun Zeng

Department of Management Information Systems, University of Arizona
;
Institute of Automation, Chinese
Academy of Sciences
, Tucson, AZ, 85721
, USA,
zeng@eller.arizona.edu


Network Topolog
ical Measures on “Small World,” “Clustering,” and “Scale
-
free” Phenomena

Three major concepts related to such topological features are: “small world,” “clustering,” and
“scale
-
free” phenomena
(Albert and Barabási 2002, Newman, et al. 2001)
.

Small Wor
ld
: The small world concept describes the fact that despite their often large size,
most networks exhibit a relatively short path between any two vertices. The distance between
two vertices is defined as the number of edges along the shortest path connecti
ng them. The
a
v
erage path length

(or typical/characteristic distance) measure
L
, defined as the average of the
path lengths of all connected vertex pairs, quantifies this property.

Clustering
: Many real
-
world networks show an inherent tendency to cluster.
A typical
exa
m
ple is social networks, in which cliques form, representing circles of friends or
acquaintances in which every member knows every other member. Such a tendency is quantified
by the clustering coefficient measure
(Newman, et al. 2001, Watts and Strogatz 1998)
. We adopt
the New
man d
e
finition:





(1)

where a triangle is a set of three vertices each of which is connected to both of the others, and a
connected triple is three vertices
x
-
y
-
z
, with both vertices
x

and
z

connected with
y

(note that
x
-
y
-
z

and
z
-
y
-
x

are considered the same connected triple). The factor 3 in the numerator accounts for
the fact that each triangle contributes to three connected triples of vertices. The clustering
coeff
i
cient
C

is strictly bounded between 0 and 1 and measures the exte
nt to which being a
neighbor is a transitive property. In our context, for example, a consumer graph represents
relationships b
e
tween consumers who purchase the same products. In a consumer graph with a
high clustering coefficient (close to 1) such a co
-
pu
rchase relationship tends to be transitive
under most cases, i.e., if consumers
a

and
b

purchase the same products and consumers
b

and
c

pu
r
chase the same products, then consumers
a

and
c

are highly likely to do so as well.

Scale
-
free
: The scale
-
free prope
rty is linked to the degree distribution of a graph. The
d
e
gree

of a vertex in a graph is the number of edges incident on that vertex. We define
p
(
k
)
, known as
the degree distribution of the graph, to be the probability that a vertex chosen uniformly at
ra
n
dom has degree
k

(i.e., the fraction of vertices that have degree
k
). Scale
-
free graphs refer to
graphs with power
-
law degree distributions as described by (2):






(2)

where
α

is a positive constant. Power
-
law degree distributions have been observed in a wide
range of networks, including many of the real networks mentioned previously.

Collaborative Filtering Algorithms

We first introduce a common notation for describing a coll
aborative filtering problem. The input
of the problem is an
M


N

interaction matrix

A

= (
a
ij
) associated with
M

consumers
C

= {
c
1
,
c
2
,…, c
M
} and
N

products
P

= {
p
1
,
p
2
, …,
p
N
}. We focus on recommendation that is based on
transactiona
l data. That is,
a
ij

can take the value of either 0 or 1 with 1 representing an observed
transaction between
c
i

and
p
j

(for example,
c
i

has purchased
p
j
) and 0 absence of transaction. We
consider the output of a collaborative filtering algorithm to be
pote
ntial scores

of products for
individual consumers that represent possibilities of future transactions. A ranked list of
K

pro
d
ucts with the highest potential scores for a target consumer serves as the recommendations.

A naïve recommendation algorithm makes

recommendation simply based on popularity of
the products, i.e., recommending to each consumer the most popular products that are not
purchased previously by this consumer. We refer to this algorithm as the
top
-
K most popular

algorithm. This naive algorit
hm has been used as a comparison benchmark in many previous
reco
m
mendation algorithm evaluation studies. Many would not consider this algorithm as a
recommendation algorithm as the recommendations are not customized at all for individual
customers. Neverth
e
less, in some situations this naïve algorithm was reported to have achieved
comparable or better performance than other more complex collaborative filtering algorithms
(Huang, et al. 2007)
. Another baseline benchmark algorithm often used in recommendation
algorithm evalu
a
tion studies is the
random

recommendation, which randomly selects
K

products
not app
earing the customer’s transaction history as the recommendation.

One basic collaborative filtering algorithm is the well
-
tested
user
-
based
neighborhood
algorithm using statistical correlation
(Brees
e, et al. 1998)
. To predict the potential interests of a
given consumer, this algorithm first identifies a set of similar consumers based on correlation
coe
f
ficients or similarity measures using the past transactions, and then makes a prediction based
on
the behavior of these similar consumers. The fundamental assumption is that consumers who
have previously bought a large set of the same products will continue to buy the same set of new
products in the future. Formally, the algorithm first computes a cons
umer similarity matrix
WC

=
(
wc
st
),
s
,
t

= 1, 2, …,
M.
The similarity score
wc
st

is calculated based on the row vectors
of
A

using
a vector

similarity function
(
such as in
(Breese, et al. 1998)
)
. A

high similarity score
wc
st

indicates that consumers
s

and
t

may have similar preferences since they have previously
pu
r
chased a large set of common products.
WC∙A

gives potential scores of the products for each
consumer.

The
item
-
based

algorithm
(Deshpande and Karypis 2
004)

is different from the user
-
based
algorithm only in that product similarities are computed instead of consumer similarities. The
a
s
sumption here is that products that have been bought by the same set of consumers will
continue to be co
-
purchased by ot
her consumers. The user
-
based and item
-
based algorithms are
the mostly commonly used CF algorithms. Formally, this algorithm first computes a product
similarity m
a
trix

WP

= (
wp
st
),
s
,
t

= 1, 2, …,
N.
Here, t
he similarity score
wp
st

is calculated
based on c
olumn vectors of
A
. A high similarity score
wp
st

indicates that products
s

and
t

are
similar in the sense that they have been co
-
purchased by many consumers.
A∙WP

gives the
potential scores of the products for each consumer.

Under the graphical representat
ion, both the user
-
based and item
-
based algorithms rely on
the paths of length 3 (involving 4 nodes, which we refer to as 4
-
node paths) to make
recomme
n
dations: “target consumer


purchased product


similar consumer


unpurchased
product” or “target consu
mer


purchased product


other consumer


similar product as the
purchased ones.” Specifically, the “target consumer


purchased product


similar consumer”
and “pu
r
chased product


other consumer


similar product” parts are the foundation for the
constr
uction of consumer and product similarity matrices,
WC

and
WP
. The more such length
-
2
paths between two consumers (products) the more similar they are. The concatenation of “


unpu
r
chased product” and “target consumer

” to the length
-
2 paths corresponds t
o the matrix
multiplication of
WC∙A

and
A∙WP

in the user
-
based and item
-
based algorithms that generate
recommendations.

Many recent CF algorithms explore data patterns beyond 4
-
node paths
(Aggarwal, et al.
1999, Huang, et al. 2004, Huang, et al. 2005, Huang, et al. 2007, Mirza, et al. 2003)
. The graph
-
based algo
rithms explicitly explore longer paths to exploit the transitive consumer
-
product
associations. The fundamental a
s
sumption is that the behavior of the transitive neighbors
(neighbors of the neighbors) is also informative in predicting the behavior of the c
onsumer. We
use the
spreading activation

algorithm in
(Huang, et al. 2004)

in this study.
This algorithm starts
with graph
-
based representation of the interaction matrix. Both the consumers and products are
represented as nodes
each with an act
i
vation level

j
, j
= 1,

,
N
.
To generate

r
ecommendations
for consumer
c
, the corresponding node is set to have activation level 1 (

c
= 1). Activation
levels of all other nodes are set
at

0. After initialization the algorithm repeatedly performs the
fo
l
lowing
activation procedure:

j
(
t

+ 1) =
,
where
f
s

is the continuous

SIGMOID

transformation function

or other normaliz
a
tion functions
;

t
ij

equals


if
i

and
j

correspond to an
observed transaction and 0 othe
r
wise (
0 <


< 1
). The algorithm stops
when

activation levels of
all n
odes converge. The final activation l
e
vels

j

of the product nodes

give the potential scores of
all products for consumer
c
.

In essence this algorithm achieves efficient exploration of the
co
n
nectedness

of a consumer
-
product pair within the consumer
-
produc
t graph context. The
connectedness concept corresponds to the nu
m
ber of paths between the pair and their lengths and
serves as the predictor of occu
r
rence of future interaction.


Extension for Rating
-
based Recommendation

In this paper, we have focused on t
ransaction
-
based recommendation where the input data is of
unary nature with only positive observations (e.g., the presence of a sales transaction indicates
positive utility of the product to the customer while absence of such a sales transaction may
r
e
vea
l that the utility is either negative or unknown). Transaction
-
based recommendation has
wide applications as no explicit feedback from the customers is needed. Any sales operation that
keeps track of the sales transaction data can apply transaction
-
based r
ecommendation alg
o
rithms
to see if future sales are predictable and to develop actionable strategies to take advantage of the
predictions. On the other hand, rating
-
based recommendation such as the Netlifx movie
recommendation represents a major portion of

the existing recommender system research
lit
e
rature. The specific graph topological measures and model selection and validation
framework presented in this paper are designed specifically for the transaction
-
based
recomme
n
dation task. As the input unary i
nteraction data for transaction
-
based recommendation
is naturally represented by an undirected unweighted bipartite graph, the recommendation task in
this co
n
text can be viewed as a task for predicting the occurrence of a future link
in
the graph.
The foll
ow
-
up graph topological measures and the notion of randomness of a graph developed in
this paper are all based on this fundamental representation. Therefore the framework presented
in this paper only applies for transaction
-
based recommendation algorithm
selection and
validation.

Although it is beyond the main focus of this paper, we provide some insights here on how to
extend our general framework to deal with rating
-
based recommendation algorithms. For the
rating
-
based recommendation tasks, we can still

employ a bipartite graph to represent the input
data. The difference is that the edges in the graph are now labeled by the specific value of the
ra
t
ing which carries information about positive and negative utility. The recommendation task is
to predict th
e label of an unobserved edge. The topological measures on such a weighted
bipartite graph should be defined differently to capture the data patterns exploited by specific
collabor
a
tive filtering algorithms. For example, for the transaction
-
based recommend
ation case,
we are interested in whether a four
-
node path
c
1

p
1

c
2

p
2

tends to form a four
-
node cycle
(measured by the 4
-
node clustering coefficient). For the rating
-
based recommendation case, we
may assign the edge value by normalized rating values,
, where
r
ij

is the rating
customer
c
i

gives product
p
j
,

is the mean rating for customer
c
i
, and
s
i

is the standard deviation
of ratings of customer
c
i
. Within this graph, we are interested in the relationship betwe
en the
products of edge va
l
ues along the path
c
1

p
1

c
2

p
2
,
r
11
’r
21
’r
22

, and the value of edge
c
1

p
2
,
r
12

,
for every 4
-
node cycle in the graph. Using the product is important here because the meaningful
sign of preference is preserved. For example, a posi
tive
r
11
’r
21
’r
22
’ may

be result of
c
1

and
c
2

both liking
p
1

and
c
2

liking
p
2

or
c
1

and
c
2

both disliking
p
1

and
c
2

liking
p
2
. Both situations may
indicate
c
1

is likely to like
p
2

(positive
r
12

) if collaborative filtering works. Note that the
standard user
-
based neighbo
r
hood CF algorithm (
3
) may be viewed as aggregating all edge value
products of 4
-
node path connecting
c
i

and
p
j

to pr
e
dict the edge value of
c
i

p
j

(4
).

,
=

(
3
)

where
P
i,j

denotes the set of products both customers
c
i

and
c
j

have rated and
denotes cu
s
tomer
c
’s overall average rating, and
C

denotes the set of neighbors considered for target customer
c
.


=

(
4
)

where
Z

is a normalizing constant. An example measure can be defined based on |
r
12




r
11
’r
21
’r
22

| or
r
12

/
r
11
’r
21
’r
22

to reveal how one edge correlates
with
the product of three other
edges within a 4
-
node cycle. Similarly oth
er weighted bipartite graph topological measures can
be defined for the recommendation algorithm selection and validation purpose. Significant
further r
e
search efforts are needed to design these measures and evaluate their quality. With
these measures, sim
ilar strategy can be adopted to generate random weighted bipartite graphs to
compare with the actual graph observed to perform hypothesis testing. We note that there are
considerable recent efforts (e.g.,
(Antoniou and Tsompa 2008, Barrat, et al. 2004)
) to generalize
graph top
o
logical measures for weighed graphs (mainly unipartite weighted graphs), which may
serve as the foundation for developing speci
a
lized bipartite weighted graph measures for our
purpos
e.

References

Aggarwal, C. C., J. L. Wolf, K.
-
L. Wu and P. S. Yu. 1999. Horting hatches an egg: A new graph
-
theoretic approach to collaborative filtering,
Proceedings of the Fifth ACM SIGKDD
Conference on Knowledge Discovery and Data Mi
ning (KDD'99)
, San Diego, CA 201
-
212.

Albert, R. and A.
-
L. Barabási. 2002. Statistical mechanics of complex networks,
Reviews of
Modern Physics
,
74

47
-
97.

Antoniou, I. E. and E. T. Tsompa. 2008. Statistical analysis of weighted networks,
Discrete
Dynamics
in Nature and Society
,
2008

Article ID 375452.

Barrat, A., M. Barthelemy, R. Pastor
-
Satorras and A. Vespignani. 2004. The architecture of
complex weighted networks,
Proceedings of National Academy of Science
,
101
(11) 3747
-
3752.

Breese, J. S., D. Heckerman
and C. Kadie. 1998. Empirical analysis of predictive algorithms for
collaborative filtering,
Proceedings of the Fourteenth Conference on Uncertainty in Artificial
Intelligence
, Madison, WI 43
-
52.

Deshpande, M. and G. Karypis. 2004. Item
-
based top
-
N re
commendation algorithms,
ACM
Transactions on Information Systems
,
22
(1) 143
-
177.

Huang, Z., H. Chen and D. Zeng. 2004. Applying associative retrieval techniques to alleviate the
sparsity problem in collaborative filtering,
ACM Transactions on Information S
ystems
(TOIS)
,
22
(1) 116
-
142.

Huang, Z., X. Li and H. Chen. 2005. Link prediction approach to collaborative filtering,
Proceedings of the 5th ACM/IEEE
-
CS joint conference on Digital libraries
, Denver, CO 141
-
142.

Huang, Z., D. Zeng and H. Chen. 2007. A com
parative study of recommendation algorithms for
e
-
commerce applications,
IEEE Intelligent Systems
,
22
(5) 68
-
78.

Mirza, B. J., B. J. Keller and N. Ramakrishnan. 2003. Studying Recommendation Algorithms by
Graph Analysis,
Journal of Intelligent Information S
ystems
,
20
(2) 131
-
160.

Newman, M. E. J., S. H. Strogatz and D. J. Watts. 2001. Random graphs with arbitrary degree
distributions and their applications,
Phys. Rev.
,
E 64

026118.

Watts, D. J. and S. H. Strogatz. 1998. Collective dynamics of small
-
world netw
orks,
Nature
,
393

440
-
442.