A Privacy-Preserving QoS Prediction Framework for Web Service Recommendation

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395 -0056
Volume: 04 Issue: 02 | Feb -2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 5.181 | ISO 9001:2008 Certified Journal | Page 386
A Privacy-Preserving QoS Prediction Framework for Web Service
Recommendation
Pallavi P. Gupta1, Sarika M. Chavan2
1 Dept of CSE (Deogiri Institute of Engineering and Management Studies) Aurangabad.
pallavi.gupta26_07@yahoo.co.in
2 Dept of CSE (Deogiri Institute of Engineering and Management Studies) Aurangabad.
sarikasolanke@dietms.org
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Web service recommendation has become a hot
topic even in basic research in IT. The most popular technique
is the collaborative filtering (CF) on the basis of a quality of
service value. With the increasing presence and adoption of
web services over the World Wide Web, the quality of service
(QoS) is becoming more important to the description Non-
functional characteristics of Web services. Severalapproaches
for the selection of Web services and recommendation via
collaborative filtering were studied; here we are going to
investigate these techniques with the pros and cons of
Techniques. Also based on these comments, we will propose a
new technique for predicting the Web service selection based
on known quality of service values and unknown we explain in
our future work.
Key Words: Web Service, Service Computing, Collaborative
filtering, QoS values, Web service recommendation; QoS
prediction; collaborative filtering; privacy preservation …
1.INTRODUCTION
Web services are software components to support
interoperable machine-to-machine interaction over a
network. The increasing presence and acceptance of Web
services on the World Wide Web demand effective
recommendation and selection techniques that recommend
the optimum web service users from a variety of available
web services. With the number of Web services to increase
Quality of Service (QoS) [1] is generally used to describe
non-functional properties of Web services. Among the
different QoS properties of Web services, some features are
user independent and have identical values for different
users (for example, price, popularity, availability, etc.). The
values of the user independency of QoS properties are
typically offered byservice providersorthird-partyregisters
(for example, UDDI). On the other hand some QoS features
users are dependent and have different values for different
users (for example, response time, Invocation failure rate,
etc.). Client-side Web service evaluation requires real web
service calls and encounters the following drawbacks:
1) First, real Web service invocations impose costs for
service users and consume the resources of the service
provider. Some web service calls can also be charged.
2) Secondly, it can exist on many Web service candidate
analyzed and some appropriate web services in the
evaluation list may not be detected and recorded by the
service user.
3) Finally, most service users are not experts in web service
evaluation and the common time-to-market constraints
limiting an in-depth review of the target web service.
However, without sufficient client-side evaluation, exact
values of the user-specific QoS properties cannot be
obtained. Optimal Web service selection and
recommendation are so difficult to achieve.
2. RECOMMENDER SYSTEM
User needs a special system which can understand their
interests and suggest them the best usable services. In this
case, Recommender systems can help users with the most
suitable items to their interests, have been consideredasone
of the best solutions. Based on the functionality, the
recommender systems can be classified as collaborative
filtering, content based filtering, Hybrid models[2].
Recommender systems can help consumers and the most
valuable items by calculating the similarities among other
consumers with collaborative filtering algorithms.
2.1 Collaborative Filtering Methods
The process of identificationof similar users,relatedWeb
services and recommend what similar users like is called
collaborative filtering. The Web services for the user are
based on the previous Web service history. A user can hardly
recall all the services that the QoS (i.e. round-trip time RTT)
represents, values of services that the user has not called are
unknown. Therefore, and accurate Web service QoS forecast
is very necessary for service user providers. Based on the
predicted QoS values the desired service selection can be
made. Collaborative Filtering[3] was initially proposed by
Rich and has been widely used in service recommendation
systems. In Web service recommendation, the primary
question of the CF is to find a group of similar users, a group
of similar services and user-service-matrix on the QoS value
of services that is build by users. The user service matrix is
actually very sparse in practice. Based on such a sparse
matrix, the prediction accuracy of QoS values of services will
decrease considerably. So we initially expected the QoS

values of the matrix of the search for historical QoS data for
similar user or similar services lacking and recommend Web
services at the optimal QoS values to the active user.
Collaborative Filtering algorithm uses two processes:
a) Prediction process[3][4] where a numerical value
expressing the predicted probability of web services that
cannot be upheld certain users. This predicted value is in the
same scale as opinion by the same user supplied values.
b) Recommendation process[3] where a list of N items
that the active users like the most is recommended. This
recommended list has those users who do not already have
access to Web services. This interface of collaborative
filtering algorithm Top N recommendation [13] is called
Collaborative filtering process and is as shown in the
following figure 1.
Fig 1. Web service recommendation process
It is impractical for every user to actively measure QoS
values due to the expensive overhead of invoking a large
number of services. To address this issue, collaborative QoS
prediction has recently been proposed, and becomes a key
step to QoS-based Web service recommendation [3], [4], [5].
Specifically, two types of CF approaches have been studied
for QoS prediction of Web services[5] in recent literature.
There are two types of collaborative filtering algorithms:
1. Model Based Collaborative Filtering
2. Memory Based Collaborative Filtering
2.1.1 Model-Based Collaborative Filtering
It involves building a model based on the dataset of
ratings. In other words, we extract some information from
the dataset, and use that as a "model" to make
recommendations[5] without having to use the complete
dataset every time. This approach potentially offers the
benefits of both speed and scalability. Using model-based
algorithms wecan study the collection of QoS, a modelwhich
is then used for QoS predictions. Model-based CF algorithms
include Bayesian models (probabilistic) and clustering
models [6]. Model-based CF technique [6] deliver a
predefined model adjust the observed QoS data, andthenthe
trained model can be used to predict the unknown QoS
values. Matrix factorization[7] is one of the most popular
model-based CF approaches that were first introduced to
address the QoS prediction problem. Matrix factorization
model [7] treats the problem well sparsely and generally
achieved better performance than neighborhood-based
approaches.Typicalexamplesincludeuser-basedapproaches
(e.g., UPCC [8]) that leverage the QoS information of similar
users for prediction.
2.1.2 Memory Based Collaborative Filtering
Memory-based algorithms approach the collaborative
filtering problem by using the entire database. As described
by Breese et. al [9], it tries to find users that are similar to the
active user (i.e. the users we want to make predictions for),
and uses their preferences to predict ratings for the active
user.Memory-basedalgorithmstomakepredictionsstoredin
memory through the use of data (users, services and QoS
data). They can be classified in nearest neighbor algorithms
and top-N recommendationalgorithms.Neighboralgorithms
are the most commonly used memory based CF algorithms.
Users similar to the current user in terms of preferences are
called neighbors. This type of CF approaches use the
observed QoS data to calculate the similarity values between
users or services and use themfurtherforQoSforecasts.Top-
N recommendation is to recommend a number of N top Web
services, this will be to a specific user of interest.AnalyzeTop
N recommendation[10] techniques to correlate the user
service matrix different users or services and use them to
calculate the recommendations.
Irjet template sample paragraph Irjet template sample
paragraph. Irjet template sample paragraph Irjet template
sample paragraph Irjet template sample paragraph Irjet
templatesample paragraph. Irjettemplatesampleparagraph
paragraph Irjet template sample paragraph Irjet template
sample paragraph. Irjet template sample paragraph Irjet
templatesample paragraph Irjet templatesampleparagraph
paragraph.
3. RELATED WORK
3.1. QoS aware Web service recommendation
History:
As the number of Web services available on the
Internet increases quickly, service consumers pay more
attention to QoS instead of functionality than before. QoS
mainly consists of non-functionalattributessuchasresponse
time, throughput, availability, etc. It has been widely used in

web service selection [11], [12] (Wang, Wang et al. 2013),
service composition (Feng, Ngan et al. 2013), service
recommendation (Cao, Wu et al. 2013; Jiang, Liu et al. 2011)
and other popular topics inthefieldofServicesComputing.In
this section, we present the related work of efficient QoS-
aware Web service [12] recommendation.
Fig2. Web Service QoS Prediction
They suggested that a Web service QoS value prediction
approach by the traditional user-based combination and
item-based collaborative filtering method. Their approach
does not require Web servicecalls and help by analyzingQoS
information of similar users Service users discover
appropriateWeb services. In its Web service[12]evaluations
in paper reports, to reduce the effect of Web service calls to
the real web services, theyselectedonlyoneoperationfroma
web service make for evaluations and use the power of this
operation to the performance by presenting the Webservice.
3.2. Web service recommendation based on
location aware Qos:
Existing approaches fail QoS variance according to user
locations to consider; and former recommender systems are
all black boxes provide only limited information about the
performance of the service candidates. Thus X. Chen, Z.
Zheng, X. Liu, Z. Huang, H. and Sun [13], [13] proposed
designed a novel collaborative filtering algorithm for large-
scale Web service recommendation on location aware QoS.
First, it combines the model-based and memory-based CF
algorithmsforWebservicerecommendation,clearlyshowing
therecommendationaccuracyandtimecomplexityimproved
compared to earlier service recommendation algorithms.
Second, they create a visually appealing interface to browse
the recommended web services, which allows a better
understanding of the service performance. Your algorithm
uses the property of QoS of users in different regions
clustering. Based on the feature region a refined nearest
neighbor algorithm is proposed to generate QoS forecasts.
The final service recommendations are on a map by putting
the underlying structureof QoS space to showandhelpusers
who accept recommendations.
Similarly, M. Tang, Y. Jiang, J. Liu and X. Liu [6] proposeda
method for location aware Collaborative Filtering Web
services for users to recommend sites of both users and
services.
Unlike existing user-based collaborative filtering to find
similar users for a target user, rather than searching for
group of users, they focus on the user physically close to the
target. Similarly, they also change existing service similarity
measurement of collaborative filtering which is used by
service location information based on a hybrid collaborative
filtering technology. After finding similar users and services
they use thesimilaritymeasuretopredictmissingQoSvalues.
Web service candidate with the top QoS values are
recommended to users. In place consciously method they
acquire first historical QoS data and the location information
of the active user. A location information handler deals with
the location informationofactiveusersandthetargetservice,
the QoS values are missing for the active user. The user
service matrix records every QoS experience the user's web
services he has called. To similar users, user similarity
measurement are based on the historicalQoSdataoftheuser
that is calculated, which are in the active user nearby,
determined by the location information handler. Likewise
Services similarity measurement is calculated based on the
QoS records of the services thatareclosetothetargetservice,
also determined by the location information handler.Similar
users and related services for the active user and target
service or CF algorithm and item-based are used for both
user-based forecasting and finding the missing QoS values of
the target service.
3.3. Web Service Recommendation Methods Based
on Personalized Collaborative Filtering
There were different methods of selecting Web services
and recommendation based on collaborative filtering, but
rarely do they take into account personal influence of users
and services. Therefore Y. Jiang, J. Liu, Tang, X. Liu [14]
provided a method of collaborative personalized
recommendation effective filtering [18] for Web service. A
significant portionofthesetechniquesisthecalculationofthe
measureof the similarity ofweb services. Unlike the Pearson
correlation coefficient (PCC) similarity measure, they
consider the personal impact of services where between
users and the personal impact of the measure of calculated
similarity of services. Based on the model of similarity
measure Web services, they develop a custom hybrid
effective collaborative filtering technology (HICP) for
integrating algorithm based on custom user and custom
algorithm based item.
Similarly, L. Shao, J. Zhang, Y. Wei, J. Zhao, B. Xie and Mei
H. [15] being aware of different experiences of consumers
quality of service, they hit a collaborative approach to
filtering based on mining similarity decision and forecasting
of consumer experiences.

4. FRAMEWORK OF QOS-AWARE WEB SERVICE
RECOMMENDATION
In this section, an online service searching scenario to show
the research problem of this paper. The basic idea of this
approach is that users closely located with each other are
more likely to have similar service experience than those
who live far away from each other. We employ the idea of
user-collaboration in our web servicerecommendersystem.
The more QoS information the user contributes, the more
accurate service recommendationstheuser canobtain,since
more user characteristics can be analyzed from the user
contributed information.Basedonthecollected QoSrecords,
our recommendation approach is designed as a two-phase
process. In the first phase, we divide the users into different
regions based on their physical locations and historical QoS
experience[15] on web services.Inthesecondphase, wefind
similar users for the current user and make QoS prediction
for the unused services. Services withthe best predictedQoS
will be recommended to the current user.
4.1. Location Information Representation,
Acquisition and Processing
This section discusses how to represent, acquire, and
process location information of both Web services and
service users, which lays a necessary foundation for
implementing location-aware Webservicerecommendation
method.
4.1.1. Location Representation:
We represent a user’s location as a [IP Address], [Country],
[IP No.], [AS], [Latitude], [Longitude]. Typically, a country
has many ASs and an AS is within one country only. The
Internet is composed of thousands of ASs that inter-
connected with each other.
However, users located in the same AS are not always
geographically close, and vice versa. Therefore, even if two
users are located in the same city, they may seem to be at
different ASs. This explains why we have chosen, AS instead
of other geographic positions, suchaslatitudeandlongitude,
to represent a user’s location.
4.1.2. Location Information:
Acquisition fetch the location information of both Web
services and service users can be easily done. Based on the
users’ IP addresses are already known,toobtainfull location
in-formation of a user, we only need to identify both the AS
and the country in which he is located based onIPaddress.A
number of services and databases are available for this
purpose (e.g. the Who is lookup service2). In this work, we
accomplished the IP to AS mapping and IP to country
mapping using the GeoLite Autonomous System Number
4.1.3. SimilarityComputationandSimilarNeighbor
Selection
Here we have defined notations for the convenience of
describing our method and algorithms. We implemented a
weighted PCC for computing similarity between both users
and Web services, which takes personal QoS characteristics
[16] into consideration. Finally, author has discussed
incorporating locations of both users and Web services into
the similar neighbor selection.
4.1.4. Similar Neighbor Selection:
This selection is a very important step of CF. In conventional
type of user-based CF, the Top-N similar neighbor selection
algorithm is used invariably [16]. It selects N users that are
most similar to the active user as neighbors. Similarly, the
Top-N similar neighbor selectionalgorithmcanbe employed
to select N Web services that are most similar to the target
Web service. Traditional Top-N algorithms ignore this
problem and still choose the top N most ones. Because ofthe
resulting neighbors are not actuallysimilartothetargetuser
(service), doing this will impair the prediction accuracy.
Therefore, abandoning those neighbors from the top N
similar neighbor set is better if the similarity is not greater
than zero. Secondly, as previously mentioned, Web service
users may happen to perceive similar QoS values on a few
Web services.
Considering the location-relatedness ofWebserviceQoS[5],
authors have incorporated the locations of users and Web
services into similar neighbor selection.
4.2. User-Based QoS Value Prediction:
Authors presented a user-based location-aware CF method,
named as ULACF[16]. Traditional user-based CF[17]
methods usually adopted for finding value predictions. This
equation, however, may be inaccurate for Web service QoS
value prediction. As Web service QoS factors such as
response time and throughput, which are objective
parameters and their values, vary largely. Therefore,
predicting QoS values based on theaverage QoS[17][18][19]
values perceived by the active user (i.e., r (u)) is flawed.
Intuitively, given two users that have the same estimated
similarity degree to the target user, the user nearer to the
target user should be placed more confidence in QoS
prediction than the other.
5. CONCLUSION
The assembly of the various QoS properties is significant for
the accomplishment of web service technology. Due to the
increasing popularity of Web services technology and the
latency of dynamic service selection and integration,several
service providers now provide parallel services. QoS is a
modified factor to discriminate functionally similar Web

services. To make it more problematic understanding is the
progression of hiding unique data with arbitrary characters
or data. The Web service recommendation helpsusersfinda
mandatory service has become important topic in the
calculation of the service.
REFERENCES
[1] Jianxun Liu, Mingdong Tang, Member,IEEE,ZibinZheng,
Member, IEEE, Xiaoqing (Frank) Liu, Member, IEEE,
Saixia Lyu, “Location-Aware and Personalized
Collaborative Filtering for Web Service
Recommendation”, IEEE Transactions on services
computing, manuscript Val X , No XX, 2015
[2] W. Zhang, H. Sun, X. Liu, and X. Guo, “Temporal QoS-
aware web service recommendation via non-negative
tensor factorization,” in Proc. of the 23rd International
World Wide Web Conference (WWW), 2014, pp. 585–
596.
[3] X. Chen, Z. Zheng, Q. Yu, and M. R. Lyu, “Web service
recommendation via exploiting location and QoS
information,” IEEE Trans. Parallel Distrib. Syst., vol. 25,
no. 7, pp. 1913–1924, 2014.
[4] X. Chen, Z. Zheng, X. Liu, Z. Huang, and H. Sun,
‘‘Personalized QoS-Aware Web Service
Recommendation and Visualization,’’ IEEE Trans. Serv.
Comput., vol. 6, no. 1, pp. 35-47R1st Quart., 2013.
[5] R. Salakhutdinov and A. Mnih, “Probabilistic matrix
Factorization,” in Proc. of the 21stAnnual Conferenceon
Neural Information Processing Systems (NIPS), 2007.
[6] M. Tang, Y. Jiang, J. Liu, X. F. Liu: Location-Aware
Collaborative Filtering for QoS-Based Service
Recommendation. in Proc. 10th International
Conference on Web Services, Hawaii, USA, June 2012,
pp.202-209
[7] S. Wang, Q. Sun, and F. Yang, ‘‘Towards Web Service
Selection Based on QoS Estimation,’’ Int’l J. Web Grid
Serv., vol. 6, no. 4, pp. 424-443, Nov. 2012.
[8] G. Kang, J. Liu, M. Tang, X. F. Liu, K. K. Fletcher. "Web
Service Selection for Resolving Conflicting Service
Requests". Proceedings of International Conference on
Web Services. IEEE Computer Society, pp. 387-394,
2011.
[9] Breese, J. S., Heckerman, D., Kadie, C. (1998). Empirical
analysis of predictive algorithms for collaborative
filtering, Proceedings of the 14 th Conference on
Uncertainty in Artificial Intelligence (UAI 1998),
Madison, Wisconsin, USA, July 24-26, 1998, 43-52.
[10] Y. Zhang, Z. Zheng, M. R. Lyu, “WSExpress: a QoS-aware
search engine for Web services”, in Proc. 8th IEEE
International Conference on Web Services, Miami, FL,
USA, July, 2010, pp.83-90.
[11] M. Alrifai, and T. Risse, “Combining Global Optimization
with Local Selection for Efficient QoS-aware Service
Composition,” in Proc. of the International World Wide
Web Conference, Apr. 2009, pp. 881- 890.
[12] X. Su and T. M. Khoshgoftaar, “A survey of collaborative
filtering techniques,” Adv. Artificial Intelligence, vol.
2009, 2009.
[13] Z. Zheng, H. Ma, M.R. Lyu, and I. King. “WSRec: A
Collaborative Filtering Based Web Service
Recommendation System,” in Proc. 7th International
Conference on Web Services, Los Angeles, CA, USA, pp.
437- 444, 2009.
[14] X. Chen, Z. Zheng, X. Liu, Z. Huang, and H. Sun,
‘‘Personalized QoS-Aware Web Service
Recommendation and Visualization,’’ IEEE Trans. Serv.
Comput., vol. 6, no. 1, pp. 35-47R1st Quart., 2013.
[15] Shao. L., Zhang, J., Wei, Y., Zhao, J., Xie, B., Mei, H.
(2007). Personalized QoSpredictionforwebservicesvia
collaborative filtering, Proceedings of the 14th IEEE
International Conference on Web Services (ICWS2007),
Salt Lake City, Utah, USA, July 9-13, 2007, 439-446.
[16] T. Yu, Y. Zhang, and K.-J. Lin, “Efficient algorithms for
web servicesselection withend-to-endQoSconstraints,”
ACM Transactions on Web, 1(1):6, 2007.
[17] G. Xue, C. Lin, Q. Yang, W. Xi, H. Zeng, Y. Yu, and Z.
Chen,‘‘Scalable Collaborative Filtering Using Cluster-
Based Smoothing,’’in Proc. 28th Int’l ACM SIGIR Conf.
Res. Dev. Inf. Retrieval,2005, pp. 114-121.
[18] K. Karta, ‘‘An Investigation on Personalized
Collaborative Filtering for Web Service Selection,’’
Honours Programme Thesis, Univ. Western Australia,
Brisbane, Old., Australia, 2005.
[19] Z. Zheng, H. Ma, M. R. Lyu, and I. King “QoS-Aware Web
Service Recommendation by Collaborative Filtering”,
IEEE Trans. on Services Computing, 2011, vol.4, no.2,
pp.140-152..
BIOGRAPHIES
Pallavi Gupta is pursuing her M.E.
from Devgiri College of
Engineering, Aurangabad.
Sarika Chavan is working as
professor at Devgiri College of
Engineering ,Aurangabad.
r
Photo

A Privacy-Preserving QoS Prediction Framework for Web Service Recommendation

More Related Content

What's hot (17)

Similar to A Privacy-Preserving QoS Prediction Framework for Web Service Recommendation (20)

More from IRJET Journal (20)

Recently uploaded (20)

A Privacy-Preserving QoS Prediction Framework for Web Service Recommendation