Temporal Influences-Aware Collaborative Filtering for QoS-Based Service
Recommendation
Jinglin Li, Jie Wang, Qibo Sun, Ao Zhou
State Key Laboratory of Networking and Switching Technology
Beijing University of Posts and Telecommunications, Beijing, China
{
jlli,
jayw, qbsun,
aozhou
}@bupt.edu.cn
Abstract—As service computing becomes increasingly
prevalent, the number of web services grows rapidly. It
becomes very important to recommend suitable, personalized
web services to users. Collaborative Filtering based on Quality
of Service (QoS) has been widely used for service
recommendation, and variety of factors such as location,
environment are taken into account to improve the accuracy of
recommendation. However, temporal influences, which is one
of key factors affecting the QoS, are not fully considered by the
investigators. In this paper, we propose a novel temporal
influences-aware collaborative filtering method which designs
an enhanced temporal influences-aware similarity
measurement to predict QoS values. Finally, we conduct a
series of experiments to evaluate the effectiveness of our
method, and results show that our method outperforms other
state-of-the-art methods.
Keywords-service recommendation; temporal influences-
aware; QoS; collaborative filtering
I. INTRODUCTION
Service computing as a promising computing paradigm
for software engineering and distributed computing has
gotten lots of attention in recent years. An application always
combines multiple services to form an integrated service to
satisfy various needs of users. The quality of the invoked
web services greatly influences the quality of integrated
service. Therefore, to build a high quality composited service
application, it is very important to identify and select an
appropriate web service.
With the exponential growth of web services, there are
many web services with similar or identical functionalities,
but different QoS. Since QoS describes the non-functional
characteristics of web services including availability, through
put, response time etc. [1], it is employed as an important
factor at the time of analogous service recommendation. QoS
based service recommendation can come to the aid of service
users to select optimal QoS performance services that meet
needs. However, it is a difficult task due to:
• Service providers rarely deliver the declared QoS
values, since web services are loosely-coupled,
located in different location and probably subject to
different development, verification, as well as testing
process [2].
• QoS values are highly related to invocation time, for
the reason that service status (e.g., number of clients,
workload etc.) and the network environment (e.g.
network latency, bandwidth etc.) change
dynamically over time, and have a great influence on
QoS performance.
Recently, Neighborhood-based Collaborative Filtering
(CF) has become the most prevalent method for
personalized service QoS prediction. CF includes user-
based, item-based and hybrid approaches, the user-based
method employs historical QoS ratings from similar users to
predict the missing QoS values, while the item-based
method utilizes historical QoS experience from similar
services to make QoS prediction. The hybrid method
combines both user-based and item-based method to achieve
a better accuracy. Furthermore, not only historical QoS
information but also context information is considered for
better accuracy. The most common discussed context factor
is location, but there are few researchers considering the
influence of time information accurately enough, which is
one of key factors affecting the QoS of web services.
To further enhance the performance of web service
recommendation, this paper proposes a novel temporal
influences-aware Collaborative Filtering method for service
QoS prediction. The key contributions of our work are as
follows:
1) In order to get high quality recommendation,
temporal dynamic characteristics of QoS values is
adequately considered in our similarity
measurement algorithm.
2) Experiments are conducted to evaluate the
prediction accuracy of our temporal influences-
aware CF method.
The rest of this paper is organized as follows. Section II
reviews related work. Section III describes our temporal
influences-aware Collaborative Filtering algorithm in detail.
Section IV presents experiments for evaluating our method
and comparisons with other CF algorithms. Section V
concludes this paper and outlines our future work.
II. R
ELATED WORK
Web service recommendation aims to recommend a
qualified web service with optimal QoS values to a service
user based on invocation records of web services.
Neighborhood-based CF methods which includes three main
categories, user-based, item-based and hybrid are widely
employed to the service QoS prediction. Zheng et al
proposes a hybrid CF method combining user-based CF and
2017 IEEE 14th International Conference on Services Computing
2474-2473/17 $31.00 © 2017 IEEE
DOI 10.1109/SCC.2017.67
471