International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 3 Issue: 7

ISSN: 2321-8169
4431 - 4436

_______________________________________________________________________________________________

Personalized Web Page Recommendation Using Ontology

Arundhati Patil

Prof. Supriya Sarkar

ME II (Computer),
SKN, Sinhgad Institute of Technology and Sciences,
Lonavala,
Email: [email protected]

Computer Dept.
SKN, Sinhgad Institute of Technology and Sciences,
Lonavala,
Email: [email protected]

Abstract:- In this network era, Web Page Recommendation and web page Recommendation systems can take advantage of semantic network
reasoning-capabilities to overcome common limitations of current systems and improve the recommendations quality. This paper presents a
personalized-web-recommendation system, a system that makes use of representations of items and user-profiles based on ontology in order to
provide semantic applications with personalized services. The recommender uses domain ontology to enhance the personalization: on the other
hand, users interests are modeled in a more effective and accurate way by applying a domain-based inference method; on the other hand, the
stemmer algorithm used by our content-based filtering approach, which provides a measure of the affinity between an item and a user, is
enhanced by applying a semantic similarity method. Web Usage Mining plays an important role in web page recommender systems and web
personalization system. In this paper, we propose an effective personalized web recommendation system based on ontology and Web Usage
Mining. The proposed approach integrates semantic knowledge into Web Usage Mining and personalization processes.
Keywords: Web Usage Mining, Ontology, History Knowledge, Semantic Network, Domain Knowledge and Web-Page Recommendation.

__________________________________________________*****_________________________________________________
1.

INTRODUCTION

Web-Page Recommendation Systems are becoming

more useful and popular. Network is getting wider and
wider. Websites are growing in numbers. So it is
challenging task of the webmasters to organize the contents
of the particular website in the network to gather the needs
of the users. This paper provides a better solution to the
Web Mining Process for recommendation purpose. The
integration of semantic knowledge system is, in fact, the
primary challenge for the next generation of personalization
systems. It is now common for Web users to encounter sites
that provide dynamic recommendations for products and
services, targeted banner advertising, and individualized link
selections. Indeed, nowhere is this technique is more
apparent as in the business-to-consumer in e-commerce
arena.
The reason is that, in today's highly competitive ecommerce environment, the success of a site often depends
on the site's ability to retain visitors and turn casual
browsers into potential customers. An Automatic
personalization and recommender system technologies have
become critical tools, because they help engage visitors at a
deeper and more intimate level by tailoring the site's
interaction with a visitor to her needs and interests. Web
personalization can be defined as any action that tailors the
Web experience to a particular user, or set of users. Before
getting deep into the main concept, lets understand the
additional concepts which are going to be used throughout
the paper.

The concepts which will be used in the whole paper are

given below:

Web Mining Process

Semantic Network
Ontology
Domain Knowledge
Web Page Recommendation
Knowledge Base Construction
The goal of this technique is to find the instances of
the concepts and relations from the Web sites domain,
because of that they can be used to perform further data
mining tasks in the system. Learning algorithms play an
important role in this technique.
2.

LITERATURE SURVEY

Classical Techniques [2] of web page recommendation

includes the result which is generated in the form of web
pages, so we can say that there is no semantic meaning of
common navigation profile system. There are many
problems which are included in the classical that is old
Systems. Now one of these problems is NEW PAGE
PROBLEM in the network. The classical systems failed to
recommends the newly added pages[3,5,6] or the products
to the visitors since these pages or products are not in the
current common navigation profiles.
So, to overcome the New page Problem the common
navigation profile can be extracted in terms of semantic
meaning or information. For that particular purpose
ontology should be used. But in classical systems ontology
was not used. Another problem is clustering [4, 6] in the
4431

IJRITCC | July 2015, Available @ http://www.ijritcc.org

_______________________________________________________________________________________

International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 3 Issue: 7

ISSN: 2321-8169
4431 - 4436

_______________________________________________________________________________________________
existing systems. In this clustering number of recommended
pages get increased. Non Useful or Unrelated pages or links
are recommended which user never prefers[2, 7].
Previous techniques used the RDF[9] and LCS algorithms.
LCS: Longest Common Subsequence
RDF: Resource Description Framework
The comparison of two sequences to determine
their similarity is one of the fundamental problems in pattern
matching technique. The Longest Common Subsequence
method generates a sequence or a list of recommended
products to the user. It is useful for the latest trend shopping
that is online shopping purpose but what about the other
users who does not use such online shopping websites but
they still want the personalization in their web page
recommendation system.The Numerical or numbering
features should be represented by single components of
vectors representing items. These components hold the exact
value of that feature. Unfortunately, classifiers of all types
tend to take a long time to construct. For example, if we
wish to use decision trees, we need one tree per user.
Constructing a tree not only requires that we look at all the
item profiles, but the problem is we have to consider many
different predicates, that could involve complex
combinations of features. Hence, this approach tends to be
used only for relatively small problem sizes.
The navigation of the users is largely driven by the
semantics. Every time while user is searching he actually
aims at finding some information concerning a particular
different subject. There are many methods to extract
keywords that characterize the web content. Which should
be the exact matching between the terms determines the
similarity between documents. Previous systems use this
approach to find the similarity between documents. But by
using this approach, only the binary similarity is achieved.
That means no actual semantic similarity is considered.
Semantic similarity is very important as far as similarity
between documents is considered.
Many number of research approaches integrate
other information sources. We require more abstract
representation which enables a more flexible and uniform
document matching process. It uses the semantic web
structures, such as ontology. Early systems possess a very
common problem of caching of web pages. When a web
user searches for an already cached page, this action is not
recorded in the web sites log.
3.

PROBLEM STATEMENT

The problem in the existing systems is that, there was

no more personalization based on ontology and domain or
history knowledge. So that the personal recommendation or

personalized recommendation system. This system uses

domain ontology and history knowledge to enhance the
personalization should be there to use for web page
recommendation purpose.
4.

PROPOSED MODEL

Modules

Creating Search history

Query clustering
Query reformulation
History grouping

These modules are explained in detail as follows:

Creating Search history

Any personal documents such as browsing history and

emails on a users computer could be the data source for
user profiles. This focus on frequent terms limits the
dimensionality of the document set, which further provides a
clear description of users interest. This module allows the
search engine to better understand a users session and
potentially tailor that users search experience according to
the needs. Once query groups, which have been identified,
that the search engines can have a very good representation
of the search context behind the current query using queries
and clicks in the corresponding query group.

Query clustering

Users queries can be classified into different query

clusters. The Concept-based user profiles are employed in
the clustering process to achieve personalization effect. The
most similar pair of concept nodes and after that, merge the
most similar pair of query nodes, and so on. Each individual
query submitted by each user is treated as an individual
node and each query with a user identifier. We perform the
grouping in a similar dynamic fashion, and whereby we first
place the current query and clicks into a query group

Query reformulation

To ensure that each query group contains closely related

and relevant queries and clicks, so it is important to have a
suitable relevance between the current query groups. We can
assume that users generally issue very similar queries and
clicks within a short period of time. Then search history of a
large number of users contains signals regarding query
relevance, like which queries tend to be issued closely
together. This technique captures the relationship between
queries frequently leading to clicks on similar URLs. Query
reformulation graph and the query click graph from search
logs, as well as how to use them to determine relevance
between queries or query groups within a users history.
4432

IJRITCC | July 2015, Available @ http://www.ijritcc.org

_______________________________________________________________________________________

International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 3 Issue: 7

ISSN: 2321-8169
4431 - 4436

_______________________________________________________________________________________________

History grouping

Query groups is to first treat every query in a users

history as a query group, and then merge these query groups
in an iterative fashion (in a k-means). However, this is
impractical in our scenario for two reasons. The first is, it
may have the undesirable effect of changing a users
existing query groups, which are used for undoing the users
own manual efforts in organizing the history. Second, it
involves a high-computational cost, for that we would have
to repeat a large number of query group similarity
computations for every new query.
5.

SYSTEM DESIGN AND IMPLEMENTATION

This paper proposes a model for personalized web page

recommendation based on Ontology, Semantic Network and
Domain Knowledge [10]. There will be semantic web usage
knowledge which an integration of both domain knowledge
and web usage knowledge. The previous implementation of
this model will demonstrate the results which produces
significantly higher performances than the previous web
usage mining and web page recommendation techniques.
This paper proposes a formal framework for integrating
full domain ontology with personalization process based on

Web usage mining. It is because the classical process was

not preferable by the user for the personalization purpose.
The Semantic Web Mining Process will provide an
infrastructure that enables not just web pages but databases,
programs, services, sensors, personal devices and even
household appliances to both produce and consume data on
the web.The system proposed here, which is used to
represent the items and user-profiles based on ontology in
order to provide semantic applications with personalized
services. The proposed recommender system is Domain
Independent system and it is implemented as a Web Service.
This kind of web service system uses both Implicit
Feedback method and Explicit Feedback Collection Method
which is used to obtain the users interest in particular
search area.
The first step of this proposed system is Feature
Extraction. Feature extraction technique approach extracts
features from web documents and constructs relevant
concepts. Then it builds the Ontology for the website usage
by using the extracted features from web documents. After
completing this, the semantic similarity of web documents is
taken into consideration to cluster the web documents into
different Semantic Themes the different themes may contain
different preferences.

Fig1. Proposed Architecture.

4433
IJRITCC | July 2015, Available @ http://www.ijritcc.org

_______________________________________________________________________________________

International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 3 Issue: 7

ISSN: 2321-8169
4431 - 4436

_______________________________________________________________________________________________
For this purpose basically two algorithms are used.
They are as follows:
A. TermNetWP
B. PrefixSpan
A. Algorithm:
Automatically construct a TermNetWP[1]
Input:
d (PageID)
O (TermNetWP)
Output:
T (list of domain terms)
Process:
Traverse through O to get the WPage instance
whosePageID is d
Set,T = term instances associated with Web-page di via
the belongto-Instance object property
Sort T in descending order of their occurrences
Return T
Three Steps are involved in this algorithm:
1. Collect the titles of visited Web-pages(TSC)
2. Extract term sequences from the Web-page
titles
3. Build the semantic network TermNetWP
B. Algorithm:
Prefix-Span
Input:
The Sequencedatabase S and the minimum threshold
MinSup.
Output:
Complete set of Sequential Patterns.
Process:
Step 1:

To find length of sequence patterns for the Sequence

database S considering the minimum support that has been
specified.
Step 2:
Divide the search space into the prefixes to search for
the ontology.
Step 3:
The subsets of sequential patterns can be searched by
constructing projected databases of the prefixes supported.
The searching is either can be depth first or breadth first
search.
Step 4:
Repeat the steps for sequential patterns until frequent
sequences of the database are not found.

6.

SYSTEM OVERVIEW

General and overall framework of web personalization for

Web Usage Mining consists of 3 phases:
1.
2.
3.

Data Preparation
Transformation
Pattern Discovery and Recommendation

These phases are the raw phases of the Semantic

Network[1,8] which is used to make relation between
concepts for the ontology. The first phase that is Data
Preparation phase is used to transform raw Web Log Files
into transaction data that can be processed by data mining
techniques.Ontology phase takes domain information and
web pages as input and generates site ontology.This is the
system which is used to represent the items and user-profiles
based on ontology in order to provide semantic applications
with personalized services.

Fig. 2. General Framework for Web Personalization[11]

Fig. 3. Online Personalization Component [11]
4434
IJRITCC | July 2015, Available @ http://www.ijritcc.org

_______________________________________________________________________________________

International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 3 Issue: 7

ISSN: 2321-8169
4431 - 4436

_______________________________________________________________________________________________
ACTUAL IMPLEMENTATION

Fig 4. Actual Implementation

7.

CONCLUSION

In conclusion, we can say that this paper proposes a

new method to offer a better Web Page Recommendation
System by using Semantic Enhancements and ontology,
which are called as Knowledge Representation Models for
Recommendation Purpose. One model is for Domain
Knowledge, second model is for ontology. This system is a
conceptual prediction model which is used to integrate the
web usage and domain knowledge. It is used to form a

weighted semantic network of frequently viewed pages or

terms. This system is proposed to predict next Web Page
requests of users through querying the knowledge.
REFERENCES
[1] ThiThanh Sang Nguyen, Hai Yan Lu, Jie Lu, Web-page
Recommendation based on Web Usage and Domain
Knowledge,Knowledge and Data Engineering IEEE
Transaction on September 2014
[2] B. Liu, B. Mobasher, and O. Nasraoui, Web Usage Mining,
in Web Data Mining: Exploring Hyperlinks, Contents, and
4435

IJRITCC | July 2015, Available @ http://www.ijritcc.org

_______________________________________________________________________________________

International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 3 Issue: 7

ISSN: 2321-8169
4431 - 4436

_______________________________________________________________________________________________
[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

Usage Data, Ed.: Springer-Verlag Berlin Heidelberg, 2011,

pp. 527-603.
S. Grimm, A. Abecker, J. Vlker, and R. Studer, "Ontologies
and the Semantic Web" , in Handbook of Semantic Web
echnologies,J. Domingue, D. Fensel, and J. A. Hendler, Eds.:
Springer-Verlag Berlin Heidelberg, 2011, pp. 507-580.
B. Liu "Information Retrieval and Web Search", in Web Data
Mining: Exploring Hyperlinks, Contents, and Usage Data, B.
Liu, Ed.: Springer-Verlag Berlin Heidelberg, 2011, pp.183236.
H. Dai and B. Mobasher, "Integrating Semantic Knowledge
with Web Usage Mining for Personalization", in Web Mining:
Applications and Techniques, A. Scime, Ed. Hershey, PA,
SA: IGI Global, 2005, pp. 276 - 306.
Brindha.S and Sabarinathan.P "An effective search on web
log from most popular downloaded content",International
Journal of Distributed and Parallel Systems (IJDPS) Vol.5,
No.1/2/3, May 2014.
Mahendra Thakur, Yogendra Kumar Jain, GeetikaSilakari,
"Query based Personalization in Semantic Web Mining",
(IJACSA) International Journal of Advanced Computer
Science and Applications,Vol. 2, No.2, February 2011
Sneha Y. S, G Mahadevan, and Madhura Prakash, "A
Personalized Product Based Recommendation System Using
Web Usage Mining and Semantic Web", International Journal
of Computer Theory and Engineering Vol. 4, No. 2, April
2012.
C.Ramesh1, Dr. K. V. Chalapati Rao, Dr. A.Goverdhan, "A
Semantically Enriched Web Usage Based Recommendation
Model", International Journal of Computer Science &
Information Technology (IJCSIT) Vol 3, No 5, Oct 2011.
V. SithaRamulu, Ch. N. Santhosh Kumar, K. Sudheer
Reddy,"A Study of Semantic Web Mining: Integrating
Domain Knowledge into Web Mining", International Journal
of Soft Computing and Engineering (IJSCE)ISSN: 22312307, Volume-2, Issue-3, July 2012.
Honghua Dai and Bamshad Mobasher, Integrating Semantic
Knowledge with Web Usage Mining for Personalization,
School of Computer Science, Telecommunication, and
Information Systems.

4436
IJRITCC | July 2015, Available @ http://www.ijritcc.org

_______________________________________________________________________________________