IEEE TRANSACTIONS ON

EMERGING TOPICS
IN COMPUTING
Received 14 November 2013; revised 30 January 2014; accepted 25 February 2014. Date of publication 10 March 2014;
date of current version 30 October 2014.
Digital Object Identifier 10.1109/TETC.2014.2310485

ClubCF: A Clustering-Based Collaborative

Filtering Approach for Big Data Application
RONG HU1,2 , (Member, IEEE), WANCHUN DOU1 , (Member, IEEE), and
JIANXUN LIU2 , (Member, IEEE)
1 State Key Laboratory for Novel Software Technology, Department of Computer Science and Technology, Nanjing University,

Nanjing 210093, China

2 Key Laboratory of Knowledge Processing and Networked Manufacturing, Hunan University of Science and Technology,

Xiangtan 411201, China

CORRESPONDING AUTHOR: R. HU (ronghu@126.com)
This work was supported in part by the National Science Foundation of China under Grants 91318301 and 61321491, in part by the National
Key Technology Research and Development Program of the Ministry of Science and Technology under Grant 2011BAK21B06,
and in part by the Excellent Youth Found of Hunan Scientific Committee of China under Grant 11JJ1011.

ABSTRACT

Spurred by service computing and cloud computing, an increasing number of services are
emerging on the Internet. As a result, service-relevant data become too big to be effectively processed
by traditional approaches. In view of this challenge, a clustering-based collaborative filtering approach
is proposed in this paper, which aims at recruiting similar services in the same clusters to recommend
services collaboratively. Technically, this approach is enacted around two stages. In the first stage, the
available services are divided into small-scale clusters, in logic, for further processing. At the second stage,
a collaborative filtering algorithm is imposed on one of the clusters. Since the number of the services in
a cluster is much less than the total number of the services available on the web, it is expected to reduce
the online execution time of collaborative filtering. At last, several experiments are conducted to verify the
availability of the approach, on a real data set of 6225 mashup services collected from ProgrammableWeb.
INDEX TERMS

Big data application, cluster, collaborative filtering, mashup.

I. INTRODUCTION

Big data has emerged as a widely recognized trend, attracting attentions from government, industry and academia [1].
Generally speaking, Big Data concerns large-volume, complex, growing data sets with multiple, autonomous sources.
Big Data applications where data collection has grown
tremendously and is beyond the ability of commonly used
software tools to capture, manage, and process within a
tolerable elapsed time is on the rise [2]. The most fundamental challenge for the Big Data applications is to explore
the large volumes of data and extract useful information or
knowledge for future actions [3].
With the prevalence of service computing and cloud
computing, more and more services are deployed in cloud
infrastructures to provide rich functionalities [4]. Service
users have nowadays encounter unprecedented difficulties in
finding ideal ones from the overwhelming services. Recommender systems (RSs) are techniques and intelligent applications to assist users in a decision making process where

302

they want to choose some items among a potentially overwhelming set of alternative products or services. Collaborative filtering (CF) such as item- and user-based methods
are the dominant techniques applied in RSs [5]. The basic
assumption of user-based CF is that people who agree in
the past tend to agree again in the future. Different with
user-based CF, the item-based CF algorithm recommends a
user the items that are similar to what he/she has preferred
before [6]. Although traditional CF techniques are sound and
have been successfully applied in many e-commerce RSs,
they encounter two main challenges for big data application:
1) to make decision within acceptable time; and 2) to generate
ideal recommendations from so many services. Concretely,
as a critical step in traditional CF algorithms, to compute
similarity between every pair of users or services may take
too much time, even exceed the processing capability of
current RSs. Consequently, service recommendation based
on the similar users or similar services would either lose its
timeliness or couldnt be done at all. In addition, all services

2168-6750
2014 IEEE. Translations and content mining are permitted for academic research only.
Personal use is also permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

VOLUME 2, NO. 3, SEPTEMBER 2014

IEEE TRANSACTIONS ON

Hu et al.: ClubCF: Clustering-Based Collaborative Filtering Approach

are considered when computing services rating similarities

in traditional CF algorithms while most of them are different
to the target service. The ratings of these dissimilar ones may
affect the accuracy of predicted rating.
A nave solution is to decrease the number of services
that need to be processed in real time. Clustering are such
techniques that can reduce the data size by a large factor by
grouping similar services together. Therefore, we propose a
Clustering-based Collaborative Filtering approach (ClubCF),
which consists of two stages: clustering and collaborative
filtering. Clustering is a preprocessing step to separate big
data into manageable parts [7]. A cluster contains some
similar services just like a club contains some like-minded
users. This is another reason besides abbreviation that we
call this approach ClubCF. Since the number of services in
a cluster is much less than the total number of services, the
computation time of CF algorithm can be reduced significantly. Besides, since the ratings of similar services within a
cluster are more relevant than that of dissimilar services [8],
the recommendation accuracy based on users ratings may be
enhanced.
The rest of this paper is organized as follows. In Section II,
A service BigTable is designed for storage requirement
of ClubCF. It recruits BigTable and is capable of storing
service-relevant big data in distribute and scalable manner.
In Section III, ClubCF approach is described in detail step
by step. First, characteristic similarities between services
are computed by weighted sum of description similarities
and functionality similarities. Then, services are merged into
clusters according to their characteristic similarities. Next, an
item-based CF algorithm is applied within the cluster that the
target service belongs to. In Section IV, several experiments
are conducted on a real dataset extracted from Programmable
Web (http://www.programmableweb.com). Related work is
analyzed in Section V. At last, we draw some conclusions and
present some future work in Section VI.
II. PRELIMINARY KNOWLEDGE

To measure the similarity between Web services, Liu et al. [9]

investigated the metadata from the WSDL (Web Service
Description Language) files and defined a Web service
as = hN , M , D, Oi, where N is the name that specifies a Web
service, M is the set of messages exchanged by the operation
invocation, D is the set of data types, and O is the set of operations provided by the Web service. From the definition, three
types of metadata from WSDL can be identified for similarity
matching: the plain textual descriptions, the operation that
captures the purposed functionality and the data type relate to
the semantic meanings. For evaluating reputation of service,
Li et al. [10] defined a Web service as WS(id, d, t, sg, rs, dor)
where id is its identity, d is its text description, t is its
classification, sg denotes the level of its transaction volume,
rs is its review set, and dor is its reputation degree. In the
SOA Solution Stack (S3) [11] proposed by IBM, a service is
defined as an abstract specification of one or more businessaligned IT functions. This specification provides consumers
VOLUME 2, NO. 3, SEPTEMBER 2014

EMERGING TOPICS
IN COMPUTING
with sufficient information to be able to invoke the business
functions exposed by a service provider.
Although the definitions of service are distinct and
application-specific, they have common elements which
mainly include service descriptions and service functionalities. In addition, rating is an important user activity that
reflects their opinions on services. Especially in application
of service recommendation, service rating is an important
element. As more and more services are emerging on the
Internet, such huge volume of service-relevant elements are
generated and distributed across the network, which cannot
be effectively accessed by traditional database management
system. To address this problem, Bigtable is used to store
services in this paper. Bigtable [12] is a distributed storage system of Google for managing structured data that
is designed to scale to a very large size across thousands
of commodity servers. A Bigtable is a sparse, distributed,
persistent multi-dimensional sorted map. The map is indexed
by a row key, column key, and a timestamp; each value in
the map is an uninterpreted array of bytes. Column keys are
grouped into sets called column families, which form the
basic unit of access control. A column key is named using
the following syntax: family:qualifier, where family refers
to column family and qualifier refers to column key. Each
cell in a Bigtable can contain multiple versions of the same
data which are indexed by timestamp. Different versions of a
cell are stored in decreasing timestamp order, so that the most
recent versions can be read first.
In this paper, all services are stored in a Bigtable which is
called service Bigtable. The corresponding elements will be
drawn from service Bigtable during the process of ClubCF.
Formally, service Bigtable is defined as follow.
Definition 1: A service Bigtable is defined as a table
expressed in the format of < Service_ID > < Timestamp >
{< Description >: [< d1 >, < d2 >, . . .];
< Functionality > [:< f1 >, < f2 >, . . .];
< Rating >: [< u1 >, < u2 >, . . .]}
The elements in the expression are specified as follows:
1. Service_ID is the row key for uniquely identifying a
service.
2. Timestamp is used to identify time when the record is
written in service Bigtable.
3. Description, Functionality and Rating are three column
families.
4. The identifier of a description word, e.g. d1 and d2 , is used
as a qualifier of Description.
5. The identifier of a functionality e.g. f1 and f2 is used as a
qualifier of Functionality.
6. The identifier of a user, e.g. u1 and u2 is used as a qualifier
of Rating.
A slice of service Bigtable is illustrated in Table I. The
row key is s1 . The Description column family contains the
words for describing s1 , e.g., driving. The Functionality
column family contains the service functionalities,
303

IEEE TRANSACTIONS ON

EMERGING TOPICS
IN COMPUTING

Hu et al.: ClubCF: Clustering-Based Collaborative Filtering Approach

TABLE 1. A slice of a service bigtable.

e.g., Google maps. And the Rating column family contains

the ratings given by some users at different time, e.g., 4 is
a rating that u1 gave to s1 at timestamp t6 .
III. A CLUSTERING-BASED COLLABORATIVE
FILTERING APPROACH (ClubCF) FOR BIG DATA
APPLICATION

According to Definition 1, a service could be expressed

as a triple, s = (D, F, R), where D is a set of words for
describing s, F is a set of functionalities of s, R is a set of
ratings some users gave to s. Five kinds of service similarities
are computed based on D, F and R during the process of
ClubCF, which are defined as follow.
Definition 2: Suppose st = hDt , Ft , Rt i and sj =
hDj , Fj , Rj i are two services. The similarity between st and sj
is considered in five dimensions which are description similarity D_sim(st , sj ), functionality similarity F_sim(st , sj ),
characteristic similarity C_sim(st , sj ), rating similarity
0
R_sim(st , sj ) and enhanced rating similarity R_sim (st , sj ),
respectively.
With this assumption, a ClubCF approach for Big Data
application is presented, which aims at recommending
services from overwhelming candidates within an acceptable
time. Technically, ClubCF focuses on two interdependable
stages, i.e., clustering stage and collaborative filtering stage.
In the first stage, services are clustered according to their
characteristic similarities. In the second stage, a collaborative
filtering algorithm is applied within a cluster that a target
service belongs to.
Concretely, Fig. 1 depicts the specification of the ClubCF
approach step by step.
A. DEPLOYMENT OF CLUSTERING STAGE

Step 1.1: Stem Words

Different developers may use different-form words to
describe similar services. Using these words directly may
influence the measurement of description similarity. Therefore, description words should be uniformed before further
usage. In fact, morphological similar words are clubbed
together under the assumption that they are also semantically similar. For example, map, maps, and mapping are forms of the equivalent lexeme, with map as
the morphological root form. To transform variant word
forms to their common root called stem, various kinds
of stemming algorithms, such as Lovins stemmer, Daw304

FIGURE 1. Specification of the ClubCF Approach.

son Stemmer, Paice/Husk Stemmer, and Porter Stemmer,

have been proposed [13]. Among them, Porter Stemmer
(http://tartarus.org/martin/PorterStemmer/) is one of the most
widely used stemming algorithms. It applies cascaded rewrite
rules that can be run very quickly and do not require the use
of a lexicon [14].
In ClubCF approach, the words in Dt are gotten from
service Bigtable where row key = st and column family = Description. The words in Dj are gotten from service Bigtable where row key = sj and column family =
Description. Then these words are stemmed by Porter
0
0
Stemmer and put into Dt and Dj , respectively.
Step 1.2: Compute Description Similarity And
Functionality Similarity
Description similarity and functionality similarity are both
computed by Jaccard similarity coefficient (JSC) which is a
statistical measure of similarity between samples sets [15].
For two sets, JSC is defined as the cardinality of their
intersection divided by the cardinality of their union. Concretely, description similarity between st and sj is computed
by formula (1):
 0 T 0
D
Dj 
t
(1)
D_sim(st , sj ) =  0 S 0 
 Dt D 
j

VOLUME 2, NO. 3, SEPTEMBER 2014

IEEE TRANSACTIONS ON

Hu et al.: ClubCF: Clustering-Based Collaborative Filtering Approach

larger
 S
 It
Tcan be inferred from this formula that the
D0 D0  is, the more similar the two services are. D0 D0 
t
t
j
j
is the scaling factor which ensures that description similarity
is between 0 and 1.
The functionalities in Ft are gotten from service
Bigtable where row key = st and column family =
Functionality. The functionalities in Fj are gotten from
service Bigtable where row key = sj and column family =
Functionality. Then, functionality similarity between st
and sj is computed using JSC as follow:
 T 
Ft Fj 
F_sim(st , sj ) =  S 
(2)
Ft Fj 
Step 1.3: Compute Characteristic Similarity
Characteristic similarity between st and sj is computed
by weighted sum of description similarity and functionality
similarity, which is computed as follow:
C_sim(st , sj ) = D_sim(st , sj ) + F_sim(st , sj ) (3)
In this formula, [0, 1] is the weight of description
similarity, [0, 1] is the weight of functionality similarity
and + = 1. The weights express relative importance
between these two.
Provided the number of services in the recommender
system is n, characteristic similarities of every pair of services
are calculated and form a n n characteristic similarity
matrix D. An entry dt,j in D represents the characteristic
similarity between st and sj .
Step 1.4: Cluster Services
Clustering is a critical step in our approach. Clustering
methods partition a set of objects into clusters such that
objects in the same cluster are more similar to each other
than objects in different clusters according to some defined
criteria.
Generally, cluster analysis algorithms have been utilized
where the huge data are stored [16]. Clustering algorithms can
be either hierarchical or partitional. Some standard partitional
approaches (e.g., K means) suffer from several limitations:
1) results depend strongly on the choice of number of clusters K , and the correct value of K is initially unknown; 2) cluster size is not monitored during execution of the K -means
algorithm, some clusters may become empty (collapse),
and this will cause premature termination of the algorithm;
3) algorithms converge to a local minimum [17]. Hierarchical
clustering methods can be further classified into agglomerative or divisive, depending on whether the clustering
hierarchy is formed in a bottom-up or top-down fashion. Many current state-of-the-art clustering systems exploit
agglomerative hierarchical clustering (AHC) as their clustering strategy, due to its simple processing structure and
acceptable level of performance. Furthermore, it does not
require the number of clusters as input. Therefore, we use an
AHC algorithm [18], [19] for service clustering as follow.
Assume there are n services. Each service is initialized to
be a cluster of its own. At each reduction step, the two most
VOLUME 2, NO. 3, SEPTEMBER 2014

EMERGING TOPICS
IN COMPUTING
similar clusters are merged until only K (K < n) clusters
remains.
Algorithm 1 AHC algorithm for service clustering
Input: A set of services S = {s1 , . . . , sn },
 
a characteristic similarity matrix D = di,j nn ,
the number of required clusters K .
Output: Dendrogramk for k = 1 to |S|.
1. Ci = {si } , i;
2. dCi ,Cj = di,j , ij;
3. for k = |S| down to K
4.
Dendrogramk = {C1 , . . . , Ck };
5.
lm = dCi ,Cj ;
6.
Cl = Join (Cl , Cm );
7.
for each Ch S
8.
if Ch 6 = Cl and Ch 6 = Cm


9.
dCl ,Ch = Average dCl ,Ch , dCm ,Ch ;
10:
end if
11: end for
12. S = S {Cm } ;
13. end for

B. DEPLOYMENT OF COLLABORATIVE FILTERING STAGE

Up to now, item-based collaborative filtering algorithms have

been widely used in many real world applications such as at
Amazon.com. It can be divided into three main steps, i.e.,
compute rating similarities, select neighbors and recommend
services.
Step 2.1: Compute Rating Similarity
Rating similarity computation between items is a timeconsuming but critical step in item-based CF algorithms.
Common rating similarity measures include the Pearson correlation coefficient (PCC) [20] and the cosine similarity
between ratings vectors. The basic intuition behind PCC measure is to give a high similarity score for two items that tend to
be rated the same by many users. PCC which is the preferred
choice in most major systems was found to perform better
than cosine vector similarity [21]. Therefore, PCC is applied
to compute rating similarity between each pair of services in
ClubCF. Provided that service st and sj are both belong to
the same cluster, PCC-based rating similarity [22] between
st and sj is computed by formula (4):


R_sim st , sj




P
T
ui Ut Uj rui ,st r st rui ,sj r sj
= qP

2 qP

2
T
T
r

r
st
ui Ut Uj ui ,st
ui Ut Uj rui ,sj r sj
(4)
Here, Ut is a set of users who rated st while Uj is a set
of users who rated sj , ui is a user who both rated st and sj ,
rui ,st is the rating of st given by ui which is gotten from service
Bigtable where row key = st and column key = Rating :
ui , rui ,sj is the rating of sj given by ui which is gotten from
305

IEEE TRANSACTIONS ON

EMERGING TOPICS
IN COMPUTING

Hu et al.: ClubCF: Clustering-Based Collaborative Filtering Approach

service Bigtable where row key = sj and column key =

Rating : ui , r st is the average rating of st , and r sj is the
average rating of sj . It should be noted that if the denominator
of formula (4) is zero, we make 0, in order to avoid division
by 0.
Although PCC can provide accurate similarity computation, it may overestimate the rating similarities when there
are a small amount of co-rated services. To address this
problem, the enhanced rating similarity [23] between st and sj
is computed by formula (5):
R_sim0 (st , sj ) =

2 |Ut Uj |
R_sim(st , sj )
|Ut | + |Uj |

(5)

In this formula, |Ut Uj | is the number of users

who rated both service st and sj , |Ut | and |Uj | are the
number of users who rated service st and sj , respectively. When the number of co-rated services is small,
2|Ut Uj |
will decrease the ratfor example, the weight |Ut |+|U
j|
ing similarity estimation between these two users. Since
2|Ut Uj |
is between the interval of [0,1] and
the value of |Ut |+|U
j|
the value of R_sim(st , sj ) is in the interval of [1, 1],
0
the value of R_sim (st , sj ) is also in the interval
of [1, 1].
Step 2.2: Select Neighbors
Based on the enhanced rating similarities between services,
the neighbors of a target service st are determined according
to constraint formula (6):


0
N (st ) = sj |R_sim (st , sj ) > , st 6 = sj
(6)
0

Here, R_sim (st , sj ) is the enhanced rating similarity between

service st and sj computed by formula (5), is a rating
similarity threshold. The bigger value of is, the chosen
number of neighbors will relatively less but they may be more
similar to the target service, thus the coverage of collaborative
filtering will decrease but the accuracy may increase. On the
contrary, the smaller value of is, the more neighbors are
chosen but some of them may be only slightly similar to
the target service, thus the coverage of collaborative filtering
will increase but the accuracy would decrease. Therefore,
a suitable should be set for the tradeoff between accuracy and coverage. While is assigned, sj will be selected
as a neighbor of st and put into the neighbor set N (st ) if
0
R_sim (st , sj ) > .
Step 2.3: Compute Predicted Rating
For an active user ua for whom predictions are being made,
whether a target service st is worth recommending depends on
its predicted rating. If N (st ) 6 = 8, similar to the computation
formula proposed by Wu et al. [24], the predicted rating
P(ua st ) in an item-based CF is computed as follow:




P
0
sj N (st ) rua ,sj r sj R_sim st , sj
Pua ,st = r st +
(7)


P
0
sj N (st ) R_sim st , sj
Here, r st is the average rating of st , N (st ) is the neighbor set
of st , sj N (st ) denotes sj is a neighbor of the target service
st , rua ,sj is the rating that an active user ua gave to sj , r sj is
306

the average rating of sj , and R_sim (st , sj ) is the enhanced

rating similarity between service st and sj computed using
formula (5).
If the predicted rating of a service exceeds a recommending
threshold, it will be a recommendable service for the active
user. A service is generally rated on a five-point scale from 1
(very dissatisfied) to 5 (very satisfied). Therefore, we set the
recommending threshold to 2.5 which is the median value of
the max rating. All recommendable services are ranked in
non-ascending order according to their predicted ratings so
that users may discover valuable services quickly.
C. TIME COMPLEXITY ANALYSIS

The time complexity of ClubCF can be divided into two parts:

1) the offline cluster building; and 2) the online collaborative
filtering.
There are two main computationally expensive steps
in the AHC algorithm. The first step is the computation of
the pairwise similarity between all the services. Provided the
number of services in the recommender system is n, the complexity of this step is generally O(n2 ) . The second step is the
repeated selection of the pair of most similar clusters or the
pair of clusters that best optimizes the criterion functionality.
A naive way of performing this step is to re-compute the gains
achieved by merging each pair of clusters after each level of
the agglomeration, and select the most promising pair. During
the lth agglomeration step, this will require O((n l)2 ) time,
leading to an overall complexity of O(n3 ) . Fortunately, if the
priority queue is implemented using a binary heap, the total
complexity of delete and insert operations is O((n l) log
(n l)). The overall complexity over the n 1 agglomeration
steps is O(n2 logn) [25].
Suppose there are m users and n services. The relationship
between users and services is denoted by a m n matrix.
Each entry ri,j represents the rating of the user ui on the
service sj . Then the time complexity of PCC-based item
similarity measures is O(nm2 ) [26]. Fortunately, the number
of service in a cluster is much less than the whole number of
services. Suppose that the number of services in a cluster Ck
is nk and the number of users who rated at least one service in
Ck is mk , then the time complexity of similarity computation
is O(nk m2k ). If the number of the target services neighbors
reaches to the max value, the worst-case time complexity of
item-based prediction is O(nk ). Since nk  n and mk  m,
the cost of computation of ClubCF may decrease significantly. Through the analysis above, it can be inferred that
ClubCF may meet the demand of real-time recommendation
to some extent.
IV. EXPERIMENTS AND EVALUATION
A. EXPERIMENTAL BACKGROUND

To verify ClubCF, a mashup dataset is used in the experiments. Mashup is an ad hoc composition technology of
Web applications that allows users to draw upon content
retrieved from external data sources to create value-added
VOLUME 2, NO. 3, SEPTEMBER 2014

IEEE TRANSACTIONS ON

Hu et al.: ClubCF: Clustering-Based Collaborative Filtering Approach

services [27]. Compared to traditional developer-centric

composition technologies, e.g., BPEL (Business Process
Execution Language) and WSCI (Web Service Choreography Interface), mashup provides a flexible and easy-ofuse way for service composition on web [28]. Recently,
mashup has become one of the hottest buzzwords in
the area of web applications, and many companies and
institutions provide various mashup solutions or re-label
existing integration solutions as mashup tools. For example, HousingMaps (http://www.housingmaps.com) combines
property listings from Craigslist (http://www.craigslist.org/)
with map data from Google Maps (http://maps.google.com/)
in order to assist people moving from one city to another
and searching for housing offers. More interesting mashup
services include Zillow (http://www.zillow.com/) and
SkiBonk (http://www.skibonk.com/).
Manual mashup development requires programming skills
and remains an intricate and time consuming task, which
prevents the average user from programming own mashup
services. To enable even inexperienced end users to mashup
their web services, many mashup-specific development tools
and frameworks have emerged [29]. The representative
approaches of end user mashup tools include Google Mashup
Editor, Yahoo Pipes, Microsoft Popfly, Intel Mash Maker,
and IBMs QEDWiki [30]. These tools speed up the overall
mashup development process, resulting in an explosion in
the amount of mashup services available on the Internet.
Meanwhile, a large number of mashup services are similar
to each other, in their components and in the logic [31]. Over
mashup-oriented big data, ClubCF is a suitable approach for
recommending ideal mashup services for users.
The data for our experiments was collected from
ProgrammableWeb, a popular online community built around
user-generated mashup services. It provides the most characteristic collection [32]. The extracted data was used to
produce datasets for the population of mashup services.
The dataset included mashup service name, tags, and APIs
used. As of Dec 2012, 6,225 mashup services and related
information are crawled from this site, which are labeled
with 20,936 tags among which 1,822 tags are different. And,
15,450 APIs are used by these mashup services among which
1,499 APIs are different in name. The tags are stemmed using
Porter Stemmer algorithm and 1,608 different stems of tags
are obtained.
Since there are very few ratings available by now, we
generate pseudorandom integers in the range 0 to 5 as the
ratings of mashup services. Assume there are 500 users that
have rated some mashup services published on the website.
Then the user-item matrix consists of 500 rows and 6,225
columns. In total, 50,000 non-zero ratings are generated.
The sparsity level of the matrix is 98.39% (sparsity level =
1-50000/500 6226 = 0.9839).
We add an empirical evaluation based on a well known
statistical test, namely the l-fold cross validation [33]. The
ratings records are split into l mutually exclusive subsets
(the folds) of equal size. During each step, it is tested on fold
VOLUME 2, NO. 3, SEPTEMBER 2014

EMERGING TOPICS
IN COMPUTING
TABLE 2. The experimental environments.

and trained on the rest. The cross-validation process is then

repeated l times, with each of the l subsets used exactly once
as the validation data. In this paper, 5-fold cross validation
is applied (i.e., l = 5). In order to distribute test data and
training data over all clusters, 20% services of each cluster
was included in test data and 80% of it was included in
training data for each data split.
B. EXPERIMENTAL ENVIRONMENTS

The experiments are conducted in a hybrid network

environment that consists of three local hosts. The
mashup services recruited in the experiment are distributed
among the three hosts. As listed in Table II, the cluster (HDFS: http://114.212.190.91:50070, and JobTracker:
http://114.212.190.91:50030, Campus Network) consists of
an 18 node cluster, with one master node and 17 slave
nodes. Each node is equipped with two Intel(R) Quad
Core E5620 Xeon(R) processors at 2.4GHz and 24GB
RAM. For the master node, a 2TB disk is mounted while,
for each slave node, two 2TB disks are equipped. The
cluster runs under Redhat Enterprise Linux Server 6.0,
Java 1.6.0 and Hadoop-0.20.205.0. For the private cloud
(http://cs-cloud.nju.edu.cn), there is totally 20TB storage
capacity in the system. A user can apply for an 8-core
processor, 20G memory and 100GB disk [34].
V. THE EXPERIMENT ENVIRONMENTS
A. Experimental Case Study

According to ClubCF, experimental process is promoted by

two stages as we specified in Section III: Clustering stage and
collaborative filtering stage.
307

IEEE TRANSACTIONS ON

EMERGING TOPICS
IN COMPUTING

Hu et al.: ClubCF: Clustering-Based Collaborative Filtering Approach

TABLE 4. Characteristic similarity matrix (keeping three decimal

TABLE 3. Case of mahsup services.

places).

TABLE 5. Initial similarity matrix (k = 7).

In the first stage, characteristic similarities between

mashup services are first computed. Then, all mashup
services are merged into K clusters using Algorithm 1. In the
second stage, rating similarities between mashup services that
belong to the same cluster are computed. As PCC may overestimate the rating similarities, enhanced rating similarities
are calculated. Then some mashup services whose enhanced
rating similarities with the target mashup service exceed a
threshold are selected as neighbors of the target mashup
service. At last, the predicted rating of the target mashup
service is computed.
1) Deployment of Clustering Stage
Step 1.1: Stem Words
Generally, a mashup service si is described with some tags
and functionalize with some APIs [35]. As an experimental
case, seven concrete mashup services (i.e., s1 , s2 , s3 , s4 , s5 , s6
and s7 ) the corresponding tags and APIs are listed in Table III.
APIs of si are put into Fi , tags of si are put into Di . Tags in Di
0
are stemmed using Porter stemmer and put into Di .
Step 1.2: Compute Description Similarity And
Functionality Similarity
Description similarities between mashup services are computed using formula (1). For instance, there are one same
stemmed tag (i.e., book) among the six different
stemmed
T
tags in D2 and D5 , therefore, D_sim(s2 , s5 ) =

0
0
D5 |
0 S 0
|D2 D5 |

|D2

= 16 .

Functionality similarities between mashup services are

computed using formula (2). Since there is only one
API (i.e., AmazonTProduct Advertising) in F2 and F5 ,
|F
F |
F_sim(s2 , s5 ) = |F2 S F5 | = 1.
2

Step 1.3: Compute Characteristic

Similarity
Characteristic similarity is the weight sum of the description similarity and functionality similarity, which is computed
using formula (3). Without loss of generality, the weight of
308

description similarity is set to 0.5. Then the characteristic

similarity between s2 and s5 is computed as C_sim(s2 , s5 ) =
D_sim(s2 , s5 ) + (1 ) F_sim(s2 , s5 ) = 0.5 61 +
0.5 1
= 0.583. It should be noted that all the computation
results retain 3 digits after the decimal point, thereafter.
Characteristic similarities between the seven mashup
services are all computed by the same way, and the results
are shown in Table IV.
Step 1.4: Cluster Services
In this step, Algorithm 1 is processed in the specified order.
Initially, the seven services s1 s7 are put into seven clusters
C1 C7 one by one and the characteristic similarities between
each pair of services in Table IV are assigned to similarity of
the corresponding clusters. The highlighted data in Table V is
the maximum similarity in the similarity matrix.
The reduction step of Algorithm 1 is described as follows.
Step 1. Search for the pair in the similarity matrix with the
maximum similarity and merge them.
Step 2. Create a new similarity matrix where similarities
between clusters are calculated by their average
value.
Step 3. Save the similarities and cluster partitions for later
visualization.
Step 4. Proceed with 1 until the matrix is of size K , which
means that only K clusters remains.
VOLUME 2, NO. 3, SEPTEMBER 2014

IEEE TRANSACTIONS ON

EMERGING TOPICS
IN COMPUTING

Hu et al.: ClubCF: Clustering-Based Collaborative Filtering Approach

TABLE 6. Algorithm 1: reduction step 1 (k = 6).

TABLE 8. Algorithm 1: reduction step 3 (k = 4).

TABLE 9. Algorithm 1: reduction step 4 (k = 3).

TABLE 7. Algorithm 1: reduction step 2(k = 5).

TABLE 10. Rating matrix.

Let K = 3 as the termination condition of Algorithm 1, the

reduction steps are illustrated in Table VITable IX.
As for reduction Step 1 as shown in Table VI, since the
maximum similarity in the similarity matrix is dC2 ,C5 , C2
and C5 are merged into (C2 , C5 ). And the similarity between
(C2 , C5 ) and other clusters is calculated by their average

value. For example,
d(C2 ,C5 ),C3 = (dC2 ,C3 + dC5 ,C3 ) 2 =

(0 + 0.063) 2
= 0.032.
As for reduction Step 2 as shown in Table VII, since
the maximum similarity in the similarity matrix is dC3 ,C4 ,
C3 and C4 are merged into (C3 , C4 ). And the similarity
between (C3 , C4 ) and other clusters is calculated by their
average value. For example, d(C2 ,C5 ),(C3 ,C4 ) = (d(C2 ,C5 ),C3 +
d(C2 ,C5 ),C4 )/2 = (0.032 + 0.042)/2 = 0.037.
As for reduction Step 3 as shown in Table VIII, since the
maximum similarity in the similarity matrix is dC1 ,(C3 ,C4 ) , C1
and (C3 , C4 ) are merged into (C1 , C3 , C4 ). And the similarity between (C1 , C3 , C4 ) and other clusters is calculated by
their average value. For example, d(C1 ,C3 ,C4 ),C6 = (dC1 ,C6 +
d(C3 ,C4 ),C6 )/2 = (0 + 0.042)/2 = 0.021.
As for reduction Step 4 as shown in Table IX, since the
maximum similarity in the similarity matrix is d(C1 ,C3 ,C4 ),C7 ,
(C1 , C3 , C4 ) and C7 are merged into (C1 , C3 , C4 , C7 ). And
the similarity between (C1 , C3 , C4 , C7 ) and other clusters is calculated by their average value. For example,
d(C1 ,C3 ,C4 ,C7 ),(C2 ,C5 ) = (d(C1 ,C3 ,C4 ),(C2 ,C5 ) + dC7 ,(C2 ,C5 ) )/2 =
(0.019 + 0.050)/2
= 0.035.
Now, there are only 3 clusters remaining and the algorithm
is terminated.
By using Algorithm 1, the seven mashup services are
merged into three clusters, where s2 and s5 are merged
into a cluster named C1 , s1 s3 , s4 and s7 are merged into a
VOLUME 2, NO. 3, SEPTEMBER 2014

TABLE 11. Rating similarities and enhanced rating similarities

with s4 .

TABLE 12. Rating similarities and enhanced rating similarities

with s1 .

cluster named C2 , and s6 is separately merged into a cluster

named C3 .
2) Deployment of Collaborative Filtering Stage
Step 2.1: Compute Rating Similarity
Suppose there are four users (i.e., u1 u2 u3 u4 ) who rated
309

IEEE TRANSACTIONS ON

EMERGING TOPICS
IN COMPUTING
the seven mashup services. A rating matrix is established as
Table X. The ratings are on 5-point scales and 0 means the
user did not rate the mashup. As u3 does not rate s4 (a not-yetexperienced item), u3 is regarded as an active user and s4 is
looked as a target mashup. By computing the predicted rating
of s4 , it can be determined whether s4 is a recommendable
service for u3 . Furthermore, s1 is also chosen as another target
mashup. Through comparing the predicted rating and real
rating of s1 , the accuracy of ClubCF will be verified in such
case.
Since s4 and s1 are both belong to the cluster C2 , rating similarity and enhanced rating similarity are computed between
mashup services within C2 by using formula (4) and (5). The
rating similarities and enhanced rating similarities between s4
and every other mashup service in C2 are listed in Table XI
while such two kinds of similarities between s1 and every
other mashup service in C2 are listed in Table XII.
Step 2.2: Select Neighbors
Rating similarity is computed using Pearson correlation
coefficient which ranges in value from 1 to +1. The value
of 1 indicates perfect negative correlation while the value
of +1 indicates perfect positive correlation. Without loss of
generality, the rating similarity threshold in formula (6) is
set to 0.4.
Since the enhanced rating similarity between s4 and s1 is
0
0.467 ( i.e., R_sim (s4 , s1 ) = 0.467) and the enhanced rating
0
similarity between s4 and s3 is 0.631 (i.e., R_sim (s4 , s3 ) =
0.631), which are both greater than , s1 and s3 are chosen as
the neighbors of s4 , i.e., N (s4 ) = {s1 ,s3 }.
Since the enhanced rating similarity between s1 and s3 is
0
0.839 (i.e., R_sim (s1 , s3 ) = 0.839) and the enhanced rating
0
similarity between s1 and s4 is 0.467 (i.e., R_sim (s1 , s4 ) =
0.467), which are both greater than , s3 and s4 are chosen as
the neighbors of s1 , i.e., N (s1 ) = {s3 ,s4 }.
Step 2.3: Compute Predicted Rating
According to formula (7), the predicted rating of s4 for u3 ,
i.e., Pu3 ,s4 = 1.97 and the predicted rating of s1 for u3 , i.e.,
Pu3 ,s1 = 1.06.
Thus, s4 is not a good mashup service for u3 and will not
be recommended to u3 . In addition, as the real rating of s1
given by user u3 is 1 (see Table X) while its predicted rating
is 1.06, it can be inferred that ClubCF may gain an accurate
prediction.
B. Experimental Evaluation

To evaluate the accuracy of ClubCF, Mean Absolute

Error (MAE), which is a measure of the deviation of recommendations from their true user-specified ratings, is used
in this paper. As Herlocker et al. [36] proposed, MAE is
computed as follow:

Pn 


i=1 ra,t P(ua , st )
(8)
MAE =
n
In this formula, n is the number of rating-prediction pairs, ra,t
is the rating that an active user ua gives to a mashup service st ,
P(ua , st ) denotes the predicted rating of st for ua .
310

Hu et al.: ClubCF: Clustering-Based Collaborative Filtering Approach

In fact, ClubCF is a revised version of traditional itembased CF approach for adapting to big data environment.
Therefore, to verify its accuracy, we compare the MAE of
ClubCF with a traditional item-based CF approach (IbCF)
described in [26]. For each test mashup service in each fold,
its predicted rating is calculated based on IbCF and ClubCF
approach separately.

FIGURE 2. Comparison of MAE with IbCF and ClubCF.

(a) = 0.1. (b) = 0.2. (c) = 0.3. (d) = 0.4.

The mashup services published on ProgrammableWeb

focus on six categories which labeled with keywords:
photo, google, flash, mapping, enterprise, and
sms. Therefore, without loss of generality in our experiment, the value of K , which is the third input parameter of
Algorithm 1, is set to 3, 4, 5, and 6, respectively. Furthermore, rating similarity threshold is set to 0.1, 0.2, 0.3 and
0.4. Under these parameter conditions, the predicted ratings
of test services are calculated by ClubCF and IbCF. Then
the average MAEs of ClubCF and IbCF can be computed
using formula (8). The comparison results are shown in
Fig. 2(a)(d), respectively. There are several discoveries as
follows.
While the rating similarity threshold < 0.4, MAE
values of ClubCF decrease as the value of K increases.
Since services are divided into more clusters, the
services in a cluster will be more similar with each
other. Furthermore, neighbors of a target service are
chosen from the cluster that the target service belongs
to. Therefore, these neighbors may be more similar to
the target service. It results in more accurate prediction.
In contrast, plays no role in IbCF because it is not
considered in IbCF.
When < 0.4, MAE values of ClubCF and IbCF both
decrease as the value of increases. It is due to that
the neighbors will be more similar to the target service
when the value of increases. It also results in the
predicted ratings of target services computed according
to the history ratings of the neighbors are approximate
to their actual value.
When < 0.4, MAE values of ClubCF are lower than
IbCF. In ClubCF, services are first clustered according to
their characteristic similarities. And then rating similarVOLUME 2, NO. 3, SEPTEMBER 2014

IEEE TRANSACTIONS ON

EMERGING TOPICS
IN COMPUTING

Hu et al.: ClubCF: Clustering-Based Collaborative Filtering Approach

ities are measured between services in the same cluster.

Since ratings of characteristic-similar services are more
relevant with each other, it is more accurate to compute
rating similarities between services in the same cluster.
And the neighbors chosen based on the rating similarities
are more similar to the target services. Consequently,
the predicted ratings of the target services will be more
precise than that of IbCF.
While = 0.4, MAE values of ClubCF and IbCF both
increase. When k = 5 and k = 6, MAE values of
ClubCF are even more than that of IbCF. By checking the intermediate results of these two approaches,
we found that a lot of test services have few or even
no neighbors when the rating similarity threshold is set
to 0.4, especially when neighbors have to be selected
from a smaller cluster. It results in comparatively large
deviations between the predict ratings and the real ratings. With more sparse user-rating data, it will be more
difficult to find high-similar services. Therefore, if sparsity is inevitable, rating similarity threshold should be
adjusted dynamically according to the accuracy requirement of application or the sparsity of the dataset.
In addition, to evaluate the efficiency of ClubCF, the online
computation time of ClubCF is compared with that of IbCF,
as shown in Fig. 3(a)(d). There are several discoveries as
follows.

FIGURE 3. Comparison of Computation Time with ClubCF and

IbCF. (a) = 0.1. (b) = 0.2. (c) = 0.3. (d) = 0.4.

In all, ClubCF spends less computation time than Itembased CF. Since the number of services in a cluster is
fewer than the total number of services, the time of rating
similarity computation between every pair of services
will be greatly reduced.
As the rating similarity threshold increase, the computation time of ClubCF decrease. It is due to the number
of neighbors of the target service decreases when
increase. However, only when = 0.4, the decrease
of computation time of IbCF is visible. It is due to the
number of neighbors found from a cluster may less than
that of found from all, and then it may spend less time
on computing predicted ratings in ClubCF.

VOLUME 2, NO. 3, SEPTEMBER 2014

When = 0.4, as K increase, the computation time

of ClubCF decrease obviously. Since a bigger K means
fewer services in each cluster and a bigger makes
less neighbors, the computation time of predicted ratings
based on less neighbors may decrease.

According to the computation complex analysis in

Section III-C and these illustrations of experimental results,
it can draw a conclusion that ClubCF may gain good scalability via increase the parameter K appropriately. Along with
adjustment of , recommendation precision is also improved.
VI. RELATED WORK

Clustering methods for CF have been extensively studied

by some researchers. Mai et al. [37] designed a neural
networks-based clustering collaborative filtering algorithm in
e-commerce recommendation system. The cluster analysis
gathers users with similar characteristics according to the
web visiting message data. However, it is hard to say that
a users preference on web visiting is relevant to preference
on purchasing. Mittal et al. [38] proposed to achieve the
predictions for a user by first minimizing the size of item
set the user needed to explore. K -means clustering algorithm was applied to partition movies based on the genre
requested by the user. However, it requires users to provide
some extra information. Li et al. [39] proposed to incorporate
multidimensional clustering into a collaborative filtering recommendation model. Background data in the form of user
and item profiles was collected and clustered using the proposed algorithm in the first stage. Then the poor clusters with
similar features were deleted while the appropriate clusters
were further selected based on cluster pruning. At the third
stage, an item prediction was made by performing a weighted
average of deviations from the neighbors mean. Such an
approach was likely to trade-off on increasing the diversity
of recommendations while maintaining the accuracy of recommendations. Zhou et al. [40] represented Data-Providing
(DP) service in terms of vectors by considering the composite relation between input, output, and semantic relations
between them. The vectors were clustered using a refined
fuzzy C-means algorithm. Through merging similar services
into a same cluster, the capability of service search engine
was improved significantly, especially in large Internetbased service repositories. However, in this approach, it is
assumed that domain ontology exists for facilitating semantic
interoperability. Besides, this approach is not suitable for
some services which are lack of parameters. Pham et al. [41]
proposed to use network clustering technique on social
network of users to identify their neighborhood, and then
use the traditional CF algorithms to generate the recommendations. This work depends on social relationships between
users. Simon et al. [42] used a high-dimensional parameterfree, divisive hierarchical clustering algorithm that requires
only implicit feedback on past user purchases to discover
the relationships within the users. Based on the clustering
results, products of high interest were recommended to the
311

IEEE TRANSACTIONS ON

EMERGING TOPICS
IN COMPUTING
users. However, implicit feedback does not always provide
sure information about the users preference.
In ClubCF approach, the description and functionality
information is considered as metadata to measure the characteristic similarities between services. According to such
similarities, all services are merged into smaller-size clusters. Then CF algorithm is applied on the services within
the same cluster. Compared with the above approaches, this
approach does not require extra inputs of users and suits
different types of services. Moreover, the clustering algorithm used in ClubCF need not consider the dependence of
nodes.
VII. CONCLUSION AND FUTURE WORK

In this paper, we present a ClubCF approach for big data

applications relevant to service recommendation. Before
applying CF technique, services are merged into some clusters via an AHC algorithm. Then the rating similarities
between services within the same cluster are computed.
As the number of services in a cluster is much less than that
of in the whole system, ClubCF costs less online computation
time. Moreover, as the ratings of services in the same cluster
are more relevant with each other than with the ones in other
clusters, prediction based on the ratings of the services in the
same cluster will be more accurate than based on the ratings
of all similar or dissimilar services in all clusters. These two
advantageous of ClubCF have been verified by experiments
on real-world data set.
Future research can be done in two areas. First, in the
respect of service similarity, semantic analysis may be
performed on the description text of service. In this way, more
semantic-similar services may be clustered together, which
will increase the coverage of recommendations. Second, with
respect to users, mining their implicit interests from usage
records or reviews may be a complement to the explicit
interests (ratings). By this means, recommendations can be
generated even if there are only few ratings. This will solve
the sparsity problem to some extent.
REFERENCES
[1] M. A. Beyer and D. Laney, The importance of big data: A definition,
Gartner Inc., Stamford, CT, USA, Tech. Rep., 2012.
[2] X. Wu, X. Zhu, G. Q. Wu, and W. Ding, Data mining with big data, IEEE
Trans. Knowl. Data Eng., vol. 26, no. 1, pp. 97107, Jan. 2014.
[3] A. Rajaraman and J. D. Ullman, Mining of Massive Datasets.
Cambridge, U.K.: Cambridge Univ. Press, 2012.
[4] Z. Zheng, J. Zhu, and M. R. Lyu, Service-generated big data and big dataas-a-service: An overview, in Proc. IEEE Int. Congr. Big Data, Oct. 2013,
pp. 403410.
[5] A. Bellogn, I. Cantador, F. Dez, P. Castells, and E. Chavarriaga,
An empirical comparison of social, collaborative filtering, and hybrid
recommenders, ACM Trans. Intell. Syst. Technol., vol. 4, no. 1,
pp. 137, Jan. 2013.
[6] W. Zeng, M. S. Shang, Q. M. Zhang, L. L, and T. Zhou, Can
dissimilar users contribute to accuracy and diversity of personalized recommendation? Int. J. Modern Phys. C, vol. 21, no. 10,
pp. 12171227, Jun. 2010.
[7] T. C. Havens, J. C. Bezdek, C. Leckie, L. O. Hall, and M. Palaniswami,
Fuzzy c-means algorithms for very large data, IEEE Trans. Fuzzy Syst.,
vol. 20, no. 6, pp. 11301146, Dec. 2012.
312

Hu et al.: ClubCF: Clustering-Based Collaborative Filtering Approach

[8] Z. Liu, P. Li, Y. Zheng, and M. Sun, Clustering to find exemplar terms for
keyphrase extraction, in Proc. Conf. Empirical Methods Natural Lang.
Process., May 2009, pp. 257266.
[9] X. Liu, G. Huang, and H. Mei, Discovering homogeneous web service
community in the user-centric web environment, IEEE Trans. Services
Comput., vol. 2, no. 2, pp. 167181, Apr./Jun. 2009.
[10] H. H. Li, X. Y. Du, and X. Tian, A review-based reputation evaluation
approach for Web services, J. Comput. Sci. Technol., vol. 24, no. 5,
pp. 893900, Sep. 2009.
[11] K. Zielinnski, T. Szydlo, R. Szymacha, J. Kosinski, J. Kosinska, and
M. Jarzab, Adaptive SOA solution stack, IEEE Trans. Services Comput.,
vol. 5, no. 2, pp. 149163, Jun. 2012.
[12] F. Chang et al., Bigtable: A distributed storage system for structured
data, ACM Trans. Comput. Syst., vol. 26, no. 2, pp. 139, Jun. 2008.
[13] R. S. Sandeep, C. Vinay, and S. M. Hemant, Strength and accuracy
analysis of affix removal stemming algorithms, Int. J. Comput. Sci. Inf.
Technol., vol. 4, no. 2, pp. 265269, Apr. 2013.
[14] V. Gupta and G. S. Lehal, A survey of common stemming techniques and
existing stemmers for Indian languages, J. Emerging Technol. Web Intell.,
vol. 5, no. 2, pp. 157161, May 2013.
[15] A. Rodriguez, W. A. Chaovalitwongse, L. Zhe, H. Singhal, and H. Pham,
Master defect record retrieval using network-based feature association,
IEEE Trans. Syst., Man, Cybern., Part C, Appl. Rev., vol. 40, no. 3,
pp. 319329, Oct. 2010.
[16] T. Niknam, E. Taherian Fard, N. Pourjafarian, and A. Rousta, An efficient algorithm based on modified imperialist competitive algorithm and
K-means for data clustering, Eng. Appl. Artif. Intell., vol. 24, no. 2,
pp. 306317, Mar. 2011.
[17] M. J. Li, M. K. Ng, Y. M. Cheung, and J. Z. Huang, Agglomerative
fuzzy k-means clustering algorithm with selection of number of clusters, IEEE Trans. Knowl. Data Eng., vol. 20, no. 11, pp. 15191534,
Nov. 2008.
[18] G. Thilagavathi, D. Srivaishnavi, and N. Aparna, A survey on efficient
hierarchical algorithm used in clustering, Int. J. Eng., vol. 2, no. 9,
Sep. 2013.
[19] C. Platzer, F. Rosenberg, and S. Dustdar, Web service clustering using
multidimensional angles as proximity measures, ACM Trans. Internet
Technol., vol. 9, no. 3, pp. 11:111:26, Jul. 2009.
[20] G. Adomavicius and J. Zhang, Stability of recommendation algorithms,
ACM Trans. Inf. Syst., vol. 30, no. 4, pp. 23:123:31, Aug. 2012.
[21] J. Herlocker, J. A. Konstan, and J. Riedl, An empirical analysis of design
choices in neighborhood-based collaborative filtering algorithms, Inf.
Retr., vol. 5, no. 4, pp. 287310, Oct. 2002.
[22] A. Yamashita, H. Kawamura, and K. Suzuki, Adaptive fusion method
for user-based and item-based collaborative filtering, Adv. Complex Syst.,
vol. 14, no. 2, pp. 133149, May 2011.
[23] D. Julie and K. A. Kumar, Optimal web service selection scheme with
dynamic QoS property assignment, Int. J. Adv. Res. Technol., vol. 2, no. 2,
pp. 6975, May 2012.
[24] J. Wu, L. Chen, Y. Feng, Z. Zheng, M. C. Zhou, and Z. Wu,
Predicting quality of service for selection by neighborhood-based collaborative filtering, IEEE Trans. Syst., Man, Cybern., Syst., vol. 43, no. 2,
pp. 428439, Mar. 2013.
[25] Y. Zhao, G. Karypis, and U. Fayyad, Hierarchical clustering algorithms
for document datasets, Data Mining Knowl. Discovery, vol. 10, no. 2,
pp. 141168, Nov. 2005.
[26] Z. Zheng, H. Ma, M. R. Lyu, and I. King, QoS-aware web service recommendation by collaborative filtering, IEEE Trans. Services Comput.,
vol. 4, no. 2, pp. 140152, Feb. 2011.
[27] M. R. Catherine and E. B. Edwin, A survey on recent trends in cloud
computing and its application for multimedia, Int. J. Adv. Res. Comput.
Eng. Technol., vol. 2, no. 1, pp. 304309, Feb. 2013.
[28] X. Liu, Y. Hui, W. Sun, and H. Liang, Towards service composition based on mashup, in Proc. IEEE Congr. Services, Jul. 2007,
pp. 332339.
[29] X. Z. Liu, G. Huang, Q. Zhao, H. Mei, and M. B. Blake,
iMashup: A mashup-based framework for service composition,
Sci. China Inf. Sci., vol. 57, no. 1, pp. 120, Jan. 2013.
[30] H. Elmeleegy, A. Ivan, R. Akkiraju, and R. Goodwin, Mashup advisor:
A recommendation tool for mashup development, in Proc. IEEE Int. Conf.
Web Services, Oct. 2008, pp. 337344.
[31] O. Greenshpan, T. Milo, and N. Polyzotis, Autocompletion for mashups,
Proc. VLDB Endowment, vol. 2, no. 1, pp. 538549, Aug. 2009.
VOLUME 2, NO. 3, SEPTEMBER 2014

IEEE TRANSACTIONS ON

Hu et al.: ClubCF: Clustering-Based Collaborative Filtering Approach

[32] M. Weiss, S. Sari, and N. Noori, Niche formation in the mashup ecosystem, Technol. Innov. Manag. Rev., May 2013.
[33] S. An, W. Liu, S. Venkatesh, and H. Yan, Unified formulation of linear
discriminant analysis methods and optimal parameter selection, Pattern
Recognit., vol. 44, no. 2, pp. 307319, Feb. 2011.
[34] W. Dou, X. Zhang, J. Liu, and J. Chen, HireSome-II: Towards privacyaware cross-cloud service composition for big data applications, IEEE
Trans. Parellel Distrib. Syst., 2013, to be published.
[35] S. Agarwal, and A. Nath, A study on implementing Green IT in
Enterprise 2.0, Int. J. Adv. Comput. Res., vol. 3, no. 1, pp. 4349,
Mar. 2013.
[36] J. L. Herlocker, J. A. Konstan, L. G. Terveen, and J. T. Riedl, Evaluating collaborative filtering recommender systems, ACM Trans. Inf. Syst.,
vol. 22, no. 1, pp. 553, Jan. 2004.
[37] J. Mai, Y. Fan, and Y. Shen, A neural networks-based clustering collaborative filtering algorithm in e-commerce recommendation system, in Proc.
Int. Conf. Web Inf. Syst. Mining, pp. 616619, Jun. 2009.
[38] N. Mittal, R. Nayak, M. C. Govil, and K. C. Jain, Recommender
system framework using clustering and collaborative filtering, in
Proc. 3rd Int. Conf. Emerging Trends Eng. Technol., Nov. 2010,
pp. 555558.
[39] X. Li and T. Murata, Using multidimensional clustering based
collaborative filtering approach improving recommendation diversity, in
Proc. IEEE/WIC/ACM Int. Joint Conf. Web Intell. Intell. Agent Technol.,
Dec. 2012, pp. 169174.
[40] Z. Zhou, M. Sellami, W. Gaaloul, M. Barhamgi, and B. Defude, Data
providing services clustering and management for facilitating service
discovery and replacement, IEEE Trans. Autom. Sci. Eng., vol. 10, no. 4,
pp. 116, Oct. 2013.
[41] M. C. Pham, Y. Cao, R. Klamma, and M. Jarke, A clustering approach
for collaborative filtering recommendation using social network analysis,
J. Univ. Comput. Sci., vol. 17, no. 4, pp. 583604, Apr. 2011.
[42] R. D. Simon, X. Tengke, and W. Shengrui, Combining collaborative
filtering and clustering for implicit recommender system, in Proc. IEEE
27th Int. Conf. Adv. Inf. Netw. Appl., Mar. 2013, pp. 748755.

EMERGING TOPICS
IN COMPUTING
WANCHUN DOU was born in 1971. He is a
Professor with the Department of Computer Science and Technology, Nanjing University, China.
He received the Ph.D. degree from the Nanjing
University of Science and Technology, China, in
2001. Then, he continued his research work as a
Post-Doctoral Researcher with the Department of
Computer Science and Technology, Nanjing University, from 2001 to 2002. In 2005, he visited the
Hong Kong University of Science and Technology
as a Visiting Scholar. His main research interests include knowledge management, cooperative computing, and workflow technologies.

JIANXUN LIU was born in 1970. He received

the M.S. and Ph.D. degrees in computer science
from the Central South University of Technology
and the Shanghai Jiao Tong University in 1997 and
2003, respectively. He is currently a Professor with
the Department of Computer Science and Engineering, Hunan University of Science and Technology. His current interests include workflow management systems, services computing, and cloud
computing.

RONG HU was born in Xiangtan, Hunan, China,

in 1977. She is currently pursuing the Ph.D. degree
in computer science and technology from Nanjing University, China. She received the M.A.
degree from the College of Information Engineering, Xiangtan University, in 2005. Her current
research interests include service computing and
big data.

VOLUME 2, NO. 3, SEPTEMBER 2014

313