Providing highly accurate service recommendation for semantic clustering over big data

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 1889
Providing highly accurate service recommendation for semantic
clustering over big data
Neha D. Patil1, Dr. D. S. Bhosale2
1PG Student, Ashokrao Mane group of institution, Vathar
2Associate Professor, Ashokrao Mane group of institution, Vathar
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Numerous approaches have been proposed to
provide recommendations. Manifestly, recommendation
system has a variety of properties that may need experiences
of a user, such as user prediction, rating, trust, etc. On the
internet, where the number of choices is enormous, there is a
need to filter, prioritize and efficiently deliver relevant
information to mitigate the problem of many internet users.
Recommended systems are one of information filtering
systems, estimating the items that may be of additional
interest to user within a big set of items based on a user's
interests. Recommended systems are currently useful in both
the research and in the commercial areas. The paper presents
an approach for Recommended System to generate
meaningful recommendations of a collectionofusersforitems
that might interest them. This approach uses adaptive
recommender system which combines two recommendation
techniques to increase the overall performance. The mainaim
of using multiple recommendationtechniquestoovercome the
drawbacks of the traditional techniques in a combined model.
The anatomy is based on the hierarchy and input/output
relations of recommenders. The present system improves the
speed and accuracy of recommendation in big data
application.
Key Words: Adaptive Recommendation System,
clustering, data mining and Big data.
1.INTRODUCTION
Big Data relates large-volume, growing and complex data
sets with multiple and independent sources. In Big Data
applications, data collection has increased terribly and it is
beyond the ability of commonly used software to capture,
manage, and process that data [3]. The most crucial
challenge to Big Data applications is to inspect the large
volumes of data and get useful information orknowledgefor
future actions. Service users nowadaysencounterunrivalled
difficulties in finding ideal services from the enormous
services. These days, it is common for people to choose web
as the platform to buy or sell something. Therefore, there
exist many online shops in different forms, varying from
private websites to eCommerce forums. This leads to both
advantages and disadvantages for customers in different
ways [1] The main advantage is that a customer has more
options to buy from. At the same time, it can also have the
drawback, because with many options customer will face
difficulty to choose one single product keeping in view
various criteria e.g. which shop has good customer service,
and who offers the best price. Therefore, the big issueisthat
there is no one-stop place to search wide information about
e-Commerce. The information which is required related to
online selling and buying includes list of products, list of
online shops and a set of recommendations about choosing
product and shop.
Recommender system is information filtering system that
deals with the problem of information excess [7]by filtering
vital information out of large amount of dynamically
generated information as per user’s preferences, observed
behavior about item or interest [9]. Recommender system
has capacity to forecast whether a user would select an item
on the user’s profile. Collaborative filtering (CF) techniques
such as item-, user- and utility-based are the governing
techniques applied in RSs. However, traditional CF
techniques are sound and have been successfully applied in
many RSs. They face two main challenges in big data
application:1) to explore useful recommendations from so
many services and 2) to take a decision within limited time.
A critical step in traditional CF algorithms is to compute
likeness between every pair of users and/or services which
may take long a time, also beyond the processing capability
of current RSs. The ratings of dissimilar users or services
may influence the accuracyofpredictedratings.Onesolution
is to reduce the number of services thatneedtobeprocessed
in real time. Clusteringaresuchtechniquesthatcandecrease
the data size by a large factor by grouping similar services
together. Therefore, the paper proposes a clustering and
collaborative filtering with adaptive recommendation
technique. Clustering is approach that separate big data into
manageable partitions [4]. Besides, since the ratings of
similar services within a clusteraremorepertinentthanthat
of dissimilar services, the recommendation accuracy based
on users’ ratings may be enhanced. Despite the success of
filtering techniques, they exhibit cold-start, sparsity and
scalability problem. This paper proposes an adaptive
recommendation system that combines item- and
knowledge-based filtering techniques to increase the
accuracy and performance of RSs.

2.RELATED WORK
Li et al [6] proposedtointegratemultidimensional clustering
into a collaborative filtering recommendation model.
Background data in the form of user and item profiles were
collected and clustered using the proposed algorithm in the
first stage. Then the poor clusters with similar featureswere
deleted while the suitable clusters were selected based on
cluster pruning. At the last stage, an item prediction was
done by performing a weighted average of the deviations
from the neighbor’s mean. An approachlikethiswaslikelyto
trade-off on increasing the variety of recommendations
while maintaining the accuracy of the recommendations.
Pazzani et al. [12] developed an intelligentagentthattries to
evaluate which pages in the website will appeal a user by
applying a naïve Bayesian classifier. The agent allows a user
to give training instances by rating different pages as either
hot or cold. Jennings and Higuchi describe a neural network
that models the interests of a user in a Usenet news
environment.
Pham et al. [8] proposed tousenetwork clusteringtechnique
for social network of users to recognize their neighborhood.
And then use the conventional CF algorithms to explore the
recommendations. This work relies on social interaction
between users.
Simon et al. [10] used a high-dimensional parameter-free,
divisive hierarchical clustering algorithm that needs only
implicit feedback on past user purchases to find the
relationships within the users. Depending on the results of
clustering, products of high interest were suggested to the
users. But, suggested feedback does not always provide
accurate information about the user’s choice.
In Ziegler, et al. [5], a hybrid collaborative filtering approach
was proposed to capitalize on bulk taxonomic information,
designed for extracting product classification to address the
data sparsity problem of CF recommendation, based on the
generation of profiles via inference of super-topic score and
topic diversification.
A hybrid recommendation technique is also proposed in
Ghazantar and Pragel-Benett [4] and this uses the content-
based profile of the individual user to find similar users
which are used to make predictions.
Clustering techniques have received attentioninmanyfields
of study such as engineering, medicine, biology and data
mining. The goal of clustering is to collect data points.The K-
means algorithm is oneofthemostcommontechniquesused
for clustering. But, their results of K-means depend on the
initial state and converge to local optima. To overcome local
optima obstacles, a lot of studies have been done in
clustering. T. Nickname, Taherian Fared proposed [11] a
paper that presents an efficient hybrid evolutionary
optimization algorithm based on combining Modify
Imperialist Competitive Algorithm (MICA) andK-means(K),
which is called K-MICA, for optimizing clustering N objects
into K clusters. Then new Hybrid K-ICA algorithm is tested
on several data sets and its performance is compared with
those of MICA, ACO, PSO, Simulated Annealing (SA), Genetic
Algorithm (GA), Tab Search(TS), Honey Bee Mating
Optimization (HBMO) and K-means. The simulation results
show that the proposed evolutionaryoptimizationalgorithm
is robust and suitable for handling data clustering.
With the fact of service computing and cloud computing,
more and more services are appearing on the Internet,
generating huge volumes of data, such as trace logs, QoS
information, service and relationship etc. The enormous
service-generated data has becometoolargeandcomplex
to be effectively processed by conventional approaches.
How to store, manage, and create value from the service-
oriented big data has become an important research
problem. On the other hand, with the increasingly large
amount of data, a single infrastructure which provides
common functionality for managing and analyzing
different types of service-generated big data is urgently
required.
To address this challenge, Zibin Zheng, Jieming Zhu and
Michael R. Lyu [2] proposed a paper that provides an
overview of service-generated big data and Big Data-as-a-
Service. First, three types of service-generated big data are
exploited toenhancesystemperformance. Then,BigData-as-
a-Service, including Big Data Infrastructure-as-a-Service, Big
Data Platform-as-a-Service,andBigData Analytics Software-
as-a-Service is employedtoprovidecommonbigdata related
services (e.g. accessing the service generated big data and
data analytics results) to users to enhance efficiency and
reduce cost.
3.PROPOSED WORK
Here, designed system uses an adaptive recommendation
system which is making automatic predictions about the
interests of a user by gathering preferences from many
users. The aim of this system is to recommend new items to
the user or forecast the utility of a certain item, based on
user’s previous likings and on the opinions of other like-
minded users. AdaptiveRecommendationsystemdominates
content based recommender system and collaborative
filtering recommender, as the relative accuracy of the
recommender is comparatively high.

A. System Architecture
Fig 1: Architecture of Adaptive recommendation
system for semantic cluster in big data
The proposed architecture consists of two modules.
1. Admin Module
Admin can login and manage the categories of products.The
admin user canadd/update productinformation,images and
description. Also, the admin has right to approve the order
requested by visitor, generate invoices and pass to dispatch
team. Admin module has the option to add/update
advertisements. Also,admincanprovidethe productdetails,
advertise details and order details to visitor through web
services.
Cluster Generation (ComputeDescriptionSimilarity and
Functionality Similarity):
A. Description similarity: Description similarity is
computed by a Jaccard similarity coefficient (JSC) which is a
statistical measure of similarity between sample sets.
B. Functionality similarity: Functionality similarity is also
computed by a Jaccard similarity coefficient (JSC), similarly
as description similarity.
2. Visitor Module
In the proposed system visitor module isthesecondmodule.
First, visitor can register and login to the web portal. Visitor
user can view product list by categories. Also, the visitor
module has the option to view details of product and
purchase the product and notify by mail. Visitor can give
rating to the product and he can give a prediction about that
product.
Service Recommendation using (AdaptiveCollaborative
Filtering Techniques)
The recommendation system builds a database (user-item
matrix) of preferences for items by user. It then matches
users with preferences by calculating similarities between
their profiles to make recommendations. Such users build a
group called a neighborhood. A user gets recommendations
to those items that he has not rated before, but that was
already positively rated by users in his neighborhood.
Recommendation that is produced can be either prediction
or recommendation. Prediction is a numerical value, Rij,
expressing the predicted score of item j for the user i, while
recommendation is a list of top N items that theuserwill like
the most. The adaptive system combines content-based
filtering and Item-based collaborative filtering.
In content-based filtering technique, recommendation is
made based on the user profiles using features extracted
from the content of the items evaluated in the past. Items
that are mostly related to the positively rated items are
recommended to the user.
Item-based approach computes for each user-item
correlation with all other items and aggregates for eachuser
the ratings for item that are already highly correlated.
4.SCOPE OF THE WORK
The main goal of the proposed work is providing accurate
recommendations to users.
1.To present the new adaptive algorithm to improve the
Scalability, Accuracy, Memory consumption.
2.Present an approach that provides therecommendation to
users even they are new in system by removing the cold-
start problem from existing algorithm.
3.Present a system to improve the speed and accuracy of
recommendation system in big data application.
5. CONCLUSION
In this paper, Recommendationtechniqueswerestudiedand
a new system is proposed, that uses an adaptive
recommendation approach which combines content-based
filtering and Item-based collaborative filtering. Adaptive
algorithm improves the Scalability, Accuracy and Memory
consumption. Also, removes cold-start, data sparsity and
scalability problem. The new proposed system will perform
clustering and provide accurate recommendations using
adaptive approach.
6.REFERENCES
[1] NanangHusin,” Internet user Behavior Analysis in online
shopping on Indonesia”, processing of the 2011
international conference on Advanced Computer
Science and information System (ICASIS).

[2] Z. Zheng, H. Ma, M. R. Lyu, and I. King, “QoS-aware web
service recommendation by collaborative filtering,”
IEEE Trans. Services Compute, Vol.4,no.2,pp.140-152,
Feb. 2011.
[3] M. A. Beyer and D. Laney, “The importance of ‘big data:
A definition “Gartner Inc., Stamford, CT, USA, Tech. Rep,
2012.
[4] T. C. Havens, J. C. Bezdek, C. Leckie, L. O. Hall, and M.
Palaniswami, “Fuzzy c-means algorithms for very large
data,” vol. 20, no. 6, pp. 1130-1146, Dec. 2012.
[5] Ziegler CN, Lausen G, Schmidt-Thieme L. Taxonomy-
driven computation of product recommendations in:
Proceeding of the 13th international conference on
information and knowledge management (CIKM 04),
Washington DC, USA;2004
[6] 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. 169-174.
[7] Konstan JA. Riedl J. Recommender system: from
algorithms to user experience.User model User-Adapt
Interact 2012.
[8] M. C. Pham, Y. Cao, R. Klamma, and M. Jarke, “A
clustering approach for collaborative filtering
recommendation using social network analysis,”J.Univ.
Compute. Sci., vol. 17, no. 4, pp. 583-604, Apr. 2011.
[9] Pan C. Li W. Research paper recommendation with
topic analysis. In computer design and Application, IEEE
2010;4. pp, v4-264.
[10] 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. 748-755.
[11] T. Nickname, E. Taherian Fared, N. Pourjafarian, and A.
Rousta, “An efficient algorithm based on modified
imperialist competitive algorithmandK-meansfordata
clustering,” Eng. Appl. Artif. Intell., vol. 24, no. 2, pp.
306-317, Mar. 2011.
[12] Pazzani MJ A framework for collaborative content-
based and demographic filtering. Artific Intell Rew 1999.

Providing highly accurate service recommendation for semantic clustering over big data

More Related Content

What's hot (19)

Similar to Providing highly accurate service recommendation for semantic clustering over big data (20)

More from IRJET Journal (20)

Recently uploaded (20)

Providing highly accurate service recommendation for semantic clustering over big data