CSEIT172447 | Received : 12 July 2017 | Accepted : 23 July 2017 | July-August-2017 [(2)4: 141-147]
International Journal of Scientific Research in Computer Science, Engineering and Information Technology
© 2017 IJSRCSEIT | Volume 2 | Issue 4 | ISSN : 2456-3307
141
A Web Page Recommendation using Naive-Bayes Algorithm in
Hybrid Approach
1
S. Abirami,
2
J. Bhavithra,
3
Dr. A. Saradha
1,2
Department of Computer Science and Engineering, Dr. Mahalingam College of Engineering and Technology,
Pollachi, India
3
Department of Computer Science and Engineering, Institute of Road and Transport Technology, Erode, Tamunadu,
India
ABSTRACT
Web page recommendation has been emerging as a most important application area in mining. In order to predict
the users’ interests for effective recommendation two methods such as collaborative filtering and content based
filtering are considered. Content based filtering is applied by considering information including user’s profile and
the users’ past preferences. User preferences and similarity with other users are considered as primary factor in
collaborative filtering method. In probabilistic generative the unobserved user preferences are also considered along
with ratings and semantic content. To improve the accuracy and to still improve the user satisfaction this paper
applies Naïve- Bayes classifier along with content and collaborative based approach. Naive-Bayes classifier is
considered to be more efficient as it considers dynamic and adaptive features for accurate classification. The
features that are considered in Naive-Bayes classifier are independent to each other. The performance of the
proposed algorithm is measured using the precision and recall.
Keywords : Naive-Bayes Classifier, Content Based Filtering, Collaborative Filtering
I. INTRODUCTION
Web mining is the process where the information is
extracted from the web and it can evaluate the
effectiveness of particular web site. The information on
the web has been increasing, where recommendation
should be made effectively. In early days few
companies were generating data and others were
consuming. Nowadays, all of us were generating data
and all of us were consuming. The web mining requires
the recommendation system which extracts the required
knowledge from the correlated data, since the size of
the data is relatively high on the web. [20]
Web recommender system is one which it provides list
of web pages that are mostly liked for the users’. The
recommender system compares the similar and the
dissimilar data among the other content for the
effective recommendation. Recommender system gives
the list of recommendation using one of content and
collaborative based filtering. First, content based
approach the recommendation is made using the users’
information from their own profile and according to
their own interests. For instance, the user likes the web
service sa then the services related to that service will
be recommended. In Collaborative filtering, the
recommendation is made using the interests of other
users’ having the similar preferences. For instance, if
the user likes the service sa and sb. There will be many
other user who likes the service sa and sb and also like
the service sc. Probably the service sc will be
recommended to that user. [19]
The main goal of this paper is to reduce the complexity
of handling data and to increase the web service
recommendation, which reduces the users’ work on
giving their preferences (e.g., the user interests and
their personal information) and to satisfy their needs.
The cold start problem is solved where the user’s
interests is identified without any information given by
user. The accuracy level of prediction for