An Enhanced Approach for Detecting User's Behavior Applying Country-Wise Local Search

IJSRD - International Journal for Scientific Research & Development| Vol. 3, Issue 10, 2015 | ISSN (online): 2321-0613
All rights reserved by www.ijsrd.com 285
An Enhanced Approach for Detecting User’s Behavior Applying
Country-Wise Local Search
Abhishek Kumar Shukla1
G.S. Dhakad2
1
M. Tech Student 2
Assistant Professor
1,2
Department of Computer Science &Engineering
1,2
NITM, Gwalior, India
Abstract— The development of the web in past few years
has created a lot of challenge in this field. The new work in
this field is the search of the data in a search tree pattern
based on tree. Various sequential mining algorithms have
been devoloped till date. Web usage mining is used to
operate the web server logs, that contains the navigation
history of the user. Recommendater system is explained
properly with the explanation of whole procedure of the
recommendater system. The search results of the data leads
to the proper ad efficient search. But the problem was the
time utilization and the search results generated from them.
So, a new local search algorithm is proposed for country-
wise search that makes the searching more efficient on local
results basis. This approach has lead to an advancement in
the search based methods and the results generated.
Key words: Sequential pattern mining, web mining,
recommendation system
I. INTRODUCTION
Data mining is an extension of traditional data analysis and
statistical approaches. In addition it integrates analytical
techniques taken from a range of disciplines including, but
that is not limited to pattern matching numerical analysis,
and areas of artificial intelligence such as neural networks,
genetic algorithms and machine learning, [1].
The Internet has grown into a gold mine that
comprises or dynamically generates information beneficial
to E-businesses With the help of rapid growth in sources
information available on the World Wide Web(www) and
the rapidly growing pace of adoption to Internet
commerce[2].
Web mining can be classifies as the discovery and
analysis of useful information derived from the World Wide
Web(WWW), which is based on different ways to gather
information and different emphasis. Web Usage Mining can
be determined as the discovery and analysis of user access
patterns, by associated data from a specific Web site and
mining of log files.
Web usage mining can be accomplished first by
reporting visitors traffic information based on Web server
log files and other source of traffic data (as discussed
below). Web server log files were utilized initially by the
system administrators and webmasters for the purposes of
“what kind of errors are being generated, how much traffic
they are getting and how many requests fail”, etc. Though,
Web server log files can also trace and record the visitor‟s
on-line behavioral aspects. For example, after some basic
traffic analysis, the log files can help us to answer questions
such as “What pages are the most and least popular? from
which search engine are visitors coming?”[3]. Web log file
is one way to gather Web traffic data. The another way is to
“sniff” TCP/IP packets as they cross the network, and to
“plug in” to each Web server.
After the Web traffic data is gained, it may be
combined with other relational databases, over which the
implementation of data mining techniques can be done.
Recommender system is a subclass of filtering
information system that pursue to predict the 'rating' or
'preference' that user would give to an item. Recommender
systems have become extremely common in recent years,
and can be applied in a variety of applications. The most
popular ones are probably music, books, news, research
articles, movies, music, news, books, research articles,
search queries, social tags, and products in general.
II. WEB MINING
Web can be explained as a complex system with
interconnected elements and mining is the process to extract
information or data, so web mining is the process of data
extraction from this interconnected system. Web mining can
be separated into three different types – Web usage mining,
Web content mining and Web structure mining [4].
A. Web Content Mining
It refers to mine the information from weblog or record set.
In this technique content like picture text and graphs are
scanned to find out the significance of the content from the
requested query. Web page Content mining is a traditional
search process of web page via content [5].
B. Web Structure Mining
This technique refers to mine the information from the
linkage structure of the Webpage then this linkage
information is used to capture the list of interesting patterns.
Web Structure mining mines the structures like HTML or
XML tags and extracts information from the actual
organization of the pages. To give weight to the page, it uses
interconnections between web pages [5].
C. Web Usage Mining
This is the application of data mining techniques. Web
Usage miming refers to discover interesting usage patterns
from Web data in order to understand and better serve the
requirements of Web-based applications. Web usage mining
refers to mine the information gathered from the web access
behavior of the customers at the same time as they are
investigating and skimming through the Web[6]. Web usage
mining is the application of data mining techniques to
appreciate the web usage patterns. It mines data from logs,
cookies, bookmarks, , scrolls, mouse click, user sessions,
user profiles, user queries etc. The three phases of web
usage mining are preprocessing, pattern discovery, and
pattern analysis [6].

An Enhanced Approach for Detecting User’s Behavior Applying Country-Wise Local Search
(IJSRD/Vol. 3/Issue 10/2015/069)
Fig. 1: Web Mining Classification
Sequential pattern is a sequence of itemsets that
frequently occurred in a specific order, all items in the same
itemsets are supposed to have the same transaction time
value or within a time gap.
III. PROCEDURE OF RECOMMENDATION SYSTEM
Recommender systems now became famous commercially
and in the research community, where many approaches
have been recommended for providing recommendations. In
numerous cases a system designer desires to employ a
recommendation system must choose between a set of
candidate approaches. A first step towards choosing an
appropriate algorithm is to resolve which properties of the
application need focus upon when making this choice.
Indeed, recommendation systems have a variety of
properties that may affect user experience such as
scalability, accuracy, robustness, and so forth.
Recommender systems are having the effect of guiding
users in a personalized way as the useful and interesting
objects in a large space of possible options [7]. As the input
Recommendation algorithms are using customer‟s interests
to generate a list of recommended items. Amazon.com, uses
recommendation algorithms to personalize the online store
for each n every customer [8]. Recommending items related
issues has been studied extensively and mainly the two
paradigms have emerged. First is Content-based
recommendation systems that tries to identify recommend
items similar to those a given user has liked before in the
past, whereas the second systems designed recommendation
systems, according to the collaborative recommendation
paradigm identifies those users whose preferences are
similar to the given user and recommend items they have
liked [9].
A. Types of Recommendation System:
1) Content-Based Recommendation System:
The system learns to those recommend items that are similar
to the ones that the user liked in the past. The similarity of
items is calculated based on the features associated with the
compared items. For example, if a user has positively rated a
movie that belongs to the comedy genre, then the system can
learn to recommend other movies from this genre. The
overview of content based recommender systems, imposing
some order among the extensive and diverse features
involved in their design and implementation. Content-based
Recommendation system discusses trends and the future
research which might lead towards the next generation of
recommender systems. Collaborative filtering: The simplest
and original implementation of this approach [10]
recommends to the active user the items that other users
with similar tastes liked in the past. The similarity in taste of
two users is calculated based on the similarity in the rating
history of the users. This is the reason why [11] refers to
collaborative filtering as “people-to-people correlation.”
Recommendation procedure is partitioned into following
steps:-
1) Collect Web logs
2) Apply pre-processing
3) Prepare processed data base
4) Apply Pattern mining algorithm
5) Construct Pattern tree
6) Recommendation rule generation
a) Web logs:
First we will collect Web logs (Web server access Log) of
any Web-site. Web logs are the text file (ASCII) files
contain information about user IP Address, User Name,
URL that Referred, Access Request, Time Stamp; error
codes etc. and generally resides in the web servers.
b) Apply Preprocessing:
First step is preprocessing. Preprocessing need to be carried
out before performing web mining calculations on the Web
server logs.
B. Preprocessing comprises of following methods [12]:
1) Data Cleansing:
During data cleansing process Irrelevant records are
eliminated. Since the target of web usage mining is to
retrieve traversal pattern, following two kinds of records
found unnecessary and should be removed:
1) The records having suffixes for filename as GIF,
JPEG, CSS.
2) While examining the web log for status field of every
record, it eliminates the record with status code range
more than 299 and below then 200.
2) User and Session Identification:
It‟s aim is to extract different user sessions from the original
web access log.
1) If all the IP addresses are same, different users could
be found by user agents and client IP address indicated
by the different operation systems and browsers.
2) If all of the browsers, IP address, and operating
systems are same, information related to referrer
should be taken into account. The ReferURI is
checked, new user‟s session is identified if the URL in
the ReferURI is „-‟ indicates that field hasn't been
accessed before.
3) Content Retrieval:
It retrieves content derived from user‟s query request i.e.
cs_uri_query.forEg:Query:http/1www.cs.depaul.edu/courses
/syllabus.asp?course=323-21-603&q=3&y=2002&id=671.
Retrieve the content such as syllabus/courses.asp which
helps in fast searching of unvisited pages i.e; pages of
another user‟s which are similar to user‟s interest.
4) Path Completion:
It should be used for obtaining the complete user access
path. If in the begning of session, Referrer as well URI has

some data value, eliminate value of Referrer with addition
of „-„.
5) Prepare processed data base
The database which was preprocessed is now formatted and
stored in the database. The data which is stored is now in the
cleaned form and in well representable form.
6) Apply Pattern mining algorithm
Pattern mining algorithm deal with the processed data base.
Processed data base is capture by the pattern mining
algorithm for the generation of frequently accessed pattern.
7) Construct Pattern tree
Frequently accessed patterns construct a pattern tree (nodes
of the tree are represented by frequently accessed patterns)
for the recommendation rule generation. Sequence of pattern
tree can be matched with the current access sequence of the
user and then generate recommendation rule.
8) Recommendation Rule Generation
For the recommendation, we will apply Algorithm to the log
data to capture the list of frequently used items in the form
of sets, then Calculate the Confidence and Support for
different url of a website [13]. Then on the basis of the
frequent data item list we will recommend (i.e. suggestion) a
page to the user which is the best match of the user query.
IV. SEQUENTIAL PATTERN MINING ALGORITHMS
This algorithm was first introduceded by R. Agrawal and R.
Srikanth [14]. To reduce the search space for finding
frequent sequence they both have also proposed a
generalized algorithm for sequential pattern mining GSP
[15]. Closed itemset mining and Sequential pattern mining
are associated with Closed sequential pattern mining. From
a sequential database, total set of sequential patterns are
discovered by sequential pattern mining .
A brief summary of the is given below with the help of flow
diagram.
Fig. 2: Sequential Pattern-Mining Algorithms
On the basis of Apriori method algorithms such as
SPAM [16], SPADE [17], and GSP were developed.
Apriori-based methods use candidate generate-and-test
framework, to make use of the downward closure property.
Algorithms like FreeSpan [18] and PrefixSpan [19] were
developed based on pattern-growth model. Algorithms like
Pattern-growth follows an incremental mechanism for
generating construct projections and possible frequent
sequences of the database for reducing the search space.
The sequential pattern mining efficiency is
increased with the help of algorithms like SPAM, SPADE
and PrefixSpan, in terms of time and space complexity.
SPAM and PrefixSpan work on depth-first search where as
SPADE work with breadth-first search. The sequential
pattern are generated using a simple join on id-lists, SPADE
uses vertical data format but Prefix Span uses horizontal
data format with pattern growth model. Spam has the
advantage that it runs faster then SPADE and Prefix due to
use of vertical bitmap representation but it also comes up
with drawback of more memory consumption.
Closed itemset mining technique was developed to
generate closed itemset with no supersets of same set.
Closed itemset mining generates smaller set of results as
compared to frequent itemset mining. Close [20] mines
closed itemsets and is based on Apriori approach. At first
Close employs bottom-up search and then calculates the
closure of all generators.
FPclose[21] mines closed itemsets and is based on
FP-growth method. During the mining process, frequency
count of whole dataset store by FPtree used by FPclose.
FPclose uses another data structure i.e. Closed Frequent
Itemset tree by traversing all closed itemsets, to examine
the closure of a frequent itemset as well as to decrease the
search space and the no of subset testing actions.
CLOSET[22] and CHARM [23] uses space efficient depth
first search. To generate compressed database representation
to mine closed itemset, FP tree is used by CLOSET where
as CHARM uses compact vertical tid list structure also
known as diffset.
CLOSET+[24] mines closed itemsets by using item
skipping and subset-checking techniques. Item skipping
technique speed up the mining by pruning the search space.
While subset-checking technique performs the closure-
checking and reduces the memory consumption. The
CLOSET+ performs better from CLOSET and CHARM in
terms of scalability , execution time and memory usage.
Now the recent focus in sequential pattern mining
has been diverted from mining the complete set of
sequential patterns to mine the closed sequential patterns for
achieving a more compact output set with better efficiency.
A pattern is a closed pattern if it does not have any super
pattern with the same support.
For closed sequential pattern mining the two well
known algorithms are CloSpan [25] and BIDE [26]. Both
these techniques uses PrefixSpan framework. As the above
closed itemset mining techniques, CloSpan also uses a
candidate maintenance-and-test paradigm. The mining is
done in two phases. In the first phase CloSpan generates a
closed sequential pattern candidate set and keeps in a prefix
sequence lattice where as in second phase, it eliminates non
closed sequential patterns by performing post pruning. It
uses an efficient search space pruning method, known as
equivalence of projected databases. It performs
subsequence/ supersequence checking efficiently by using
the size of the projected databases as the hash key.
CloSpan produces minimum number of patterns
compared to sequential pattrn mining while preserving the
same expressive power and runs faster than PrefixSpan.

CloSpan is not scalable because it performs under candidate
maintenance-and-test paradigm. In CloSpan large amount
of closed sequential pattern candidates will consume more
memory space utilization and it leads to large search space
for checking new pattern of closure, particularly low support
threshold values or long patterns.
BIDE generated closed sequential patterns without
maintaining candidates. The Non closed sequences are
eliminated by using BI-Directional Extension closure
checking scheme. Search Space is pruned by the BackScan
pruning method. It consumes less memory and runs faster
compared to previously announced closed sequential pattern
mining techniques, specially when the support is minimum.
Scalability is better compared to previous techniques.
FMCSP [27] uses breadth first search techniques and
generate output on frequent closed patterns online. It
removes the drawback of the candidate maintenance-and-
test paradigm and generate more search space in compared
to closed pattern mining algorithms explained above. It uses
equivalence class to reduce the runtime.
V. RELATED WORK
As of late, to forecast navigation patterns of users and their
inclination, many Web Usage Mining systems have been
proposed. In this section we review some of the research
work related to web recommendation.
S.Vijayalakshmi et al [28] proposed a new
technique for frequent sequential pattern (FSP) mining,
named as AWAPT (Adaptive Web Access Pattern Tree).
To efficientl store all the sequences that contain a given
item, it combines Suffix tree and Prefix tree together.
During the mining, AWAPT removes recursive
reconstruction of intermediate WAP tree by transmitting the
binary codes to every node of it. To improve the efficiency
of AWAPT, an attempt has been done to its approach.
AWAPT completely removes the requirements to engage in
numerous reconstructions of intermediate WAP-trees during
mining and considerably decreases execution time.
Jian Pei et al [29] done survey about the
problemetic mining access patterns from Web logs
efficiently. They have introduced a novel data structure for
effcient mining of access patterns from pieces of logs, and
termed as Web access pattern tree, or WAP-tree in short. it
stores extremely compressed critical information for access
pattern mining. And they facilitates the development of
novel algorithms for mining access patterns in large set of
log pieces. This algorithm can efficiently extracts access
patterns from web logs.
Pragya Rajput, Joy Bhattacharjee, Roopali Soni[30]
proposed work helps personalize the website according to
users, country being its analytical parameter(“ web
personalization is a action that tailors the web experience to
any individual or several users “).Distinctive IP address are
provided by every internet user by the Internet Service
Provider. Every region has some specific Internet service
provider. For example, in India BSNL is the major ISP.
VI. PROPOSED WORK
It is a good practice if we provide a list of recommendation
or suggestion to the users related to their search query, for
making effective search when they interact with the web.
Frequency is calculated on the basis of popularity of the
searched keyword and popularity is calculated with the help
of analysis of web log. In this paper we propose an approach
for the recommendation that uses pattern mining apriori
algorithm for finding frequently access pattern. Apriori
algorithm generates a list of frequently accessed patterns.
The frequently accessed patterns are find out on the basis of
the local country-wise search algorithm implemented. Then
with the help of frequently accessed patterns construct a
pattern tree (nodes of the tree are represented by frequently
accessed patterns) for the recommendation rule generation.
Diverse estimation procedures like Precision, applicability
and satisfibility are considered to compute the performance
of the system.
A. Sequential Pattern Mining
Generally, every line of web logs (one access record)
comprises the following key information: requested URL,
and HTTP status code, client IP address, date-timestamp,
user ID. Prior to applying sequential pattern mining,
Preprocessing must be done on the web logs. The main data
preprocessing techniques used here containing session
identification and user identification as well as data
cleaning.
Sequential pattern mining is an significant subject of data
mining which is broadly applied in several zones [31].
Generally, sequential pattern mining is defined as the
determination of whole set of frequent subsequences in a
set of sequences [32, 33]. Sequential pattern can be
explained as the sequence of itemsets that frequently appear
in a specific order, such that, all items in that same itemset
are expected to have the same transaction time value or
within a time gap.
In our proposed work, closed sequential pattern algorithm
has been explained like this:-
1) Algorithm 1: Call_Seq()
B. Pattern tree construction
A Pattern-tree model is proposed for storing sequential web
access patterns compactly, so that it can be used for
matching with a user‟s current access sequence and
generating recommendation rules more efficiently in the
Recommendation Rules Generation component. In general,
the set of sequential web access patterns can be treated as a
set of strings over a finite alphabet E (the set of unique
access events). Sometimes, the edges of a Tree are labeled
instead of the nodes, but nevertheless refer to the same
structure in either case.
The procedure for constructing the pattern tree in
the proposed system is as follows:

1) Algorithm 2: Call_PTC()
C. Recommendation Rule Generation
It‟s elements searches for the optimum-matching access path
in the Pattern tree according to the user‟s existing access
sequence. In general, as longer the user‟s existing access
sequence is, the lower the probability is of extracting the
matching path from the Pattern-tree. Some initial items can
be eliminated to make the current access sequence shorter
than the depth of the Pattern-tree before the sequence
matching process begins.
D. Generation of Recommendations for User Access
Sequences
The constructed pattern tree is used for matching and
generating web links for recommendations for user access
sequences. For instance, the proposed system provides
recommendations R for user web access sequences <xy>
and <zx> as follows:
1) Case 1:
The constructed pattern tree is searched for sequential
patterns with the presence of User‟s web access sequence
<xy>. In the pattern tree (shown in figure 3), next to the root
node has the sequences <xyy: 2> and <xyz: 4> that are
subset of given user web access sequence <xy>. So the
postfixes of these sequential patterns i.e., <y> and <z> are
given as recommendations to the user ( R = y: 2 and z: 4).
2) Case 2:
For user web access sequence <zx>, the pattern tree (shown
in figure 3) is searched for pattern with the presence of
<zx>. As the pattern <zx> does not have any postfixes, it‟s
child node i.e., <y> is provided as recommendations for the
user ( R = y: 2).
3) Algorithm 3: Call_Rec_rule_gen()
The process of recommendation rule generation works n a
particular algorithm that calls various algorithms within
itself.
The pseudo-code for the recommendation system
proposed can be explained as below:-
VII. PERFORMANCE EVALUATION
A. Evaluation Measures
Let S= x1,x2…xl,xl+1…can be a web access sequence. For
the prefix sequence Sprefix = x1x2…xl (l ≥ MinLength), we
generate a recommendation rule Rec_Rule = {s1, s2, …,
sm} using the Pattern-tree, where all events are ordered by
their support. If xl+1∈Rec_Rule, Recommendation rules are
labeled as correct, and incorrect. If there exists xi∈Rec_Rule
(l+1≤ i ≤ l+1+m, m>0), Recommendation rules are deemed
as m-step satisfactory. Otherwise, labeled as m-step
unsatisfactory. Let R_Rule = {Rec_Rule1, Rec_Rule2, …,
Rec_Rulen} be a set of recommendation rules, where
Rec_Rulei (1≤ i ≤n) is a recommendation rule. |R_Rule| = n
is the total number of recommendation rules in R_Rule. We
define the following measures.
1) Definition 1:
Let Rc be the subset of R_Rule consisting of all correct
recommendation rules. The overall web
recommendation precision is defined as
precision=
This precision measures how probable a user will access one
of the recommended pages.
2) Definition 2:
Time Utilization = the net time utilized by the algorithm to
process the code.
Where, R_Rule are total rules generated.
Rc is subset of R_Rule which are correct rules.
Rs is subset of R_Rule that consists of all satisfactory rules.
B. Results Analysis
1) Base paper results
Total number of records : 1560
CountryNo CountryName
1 United States
2 India
3 Brazil
2) No. WebsiteName
1) http://bluerobot.info/nw2/grigosikyfaheku
2) http://buttons-for-website.com
3) http://buttons-for-website.comhttp://buttons-for-
4) website.com/
5) http://semalt.semalt.com/crawler.php
6) http://texec.org
7) http://texec.org/
8) http://texec.org/about.html
9) https://www.megakin.com/

10) https://www.megakin.com/about.html
11) https://www.megakin.com/contact.html
Etc.
1) Step 1:- User is asked to input the search sequence
2) Step 2:- User is asked to enter the suffix that he wants to
search.
3) Step 3:- Then, user is asked to enter m-step satisfactory
recommendation rules to be searched.
Elapsed time in tree creation 67.419610 seconds.
3) Output_rule
25 3
25 3
25 3
25 3
25 3
25 3
28 3
28 3
– Precision = 0.750000
– Satisfaction = 0.375
– Elapsed time in tree searching 63.0953 seconds
4) Proposed Paper Results
CountryNo CountryName
1) United States
2) India
3) Brazil
5) Country Name United States
a) WebsiteNo WebsiteName
4) http://texec.org/css/josefin-sans/stylesheet.css
5) http://texec.org/services.html
6) http://whois.domaintools.com/texec.org
7) http://www.google.co.in/url
8) http://www.google.com/m
9) http://www.google.com/search
10) http://www.google.com/url
6) Country Name India
a) WebsiteNo WebsiteName
2) http://buttons-for-website.comhttp://buttons-for-
website.com/
5) http://texec.org/about.html
6) http://texec.org/contact.html
7) http://texec.org/css/josefin-sans/stylesheet.css
8) http://texec.org/hoc/
9) http://texec.org/http://texec.org/
10) http://texec.org/school/

Etc… for all the countries.
Elapsed time is 20.627226 seconds.
1) Step 1: User is asked to enter the country name first so
as to search loacally.
2) Step 2: User is asked to input the sequence to be
searched locally.
3) Step 3: Enter the suffix for the local search.
4) Step 4:- Then, user is asked to enter m-step satisfactory
recommendation rules to be searched.
7) Output_rule
14 3
14 3
14 3
14 3
14 3
14 3
16 3
16 3
Precision = 0.625000
Satisfaction = 0.125
Elapsed time in tree searching is 10.894232 seconds.
VIII. CONCLUSION
In this paper we have discussed about the sequential pattern
mining and web mining. Various sequential mining
algorithms have been developed that have been developed
till date with all the pros and cons of every algorithm.
Recommender system with various steps used in this system
has also been discussed. A brief introduction to each
algorithm has been discussed.

REFERENCES
[1] U. M. Fayyad, G. P. Shapiro and P. Smyth, From Data
Mining to Knowledge Discovery in Databases. 0738-
4602-1996, AI Magazine (Fall 1996). pp: 37–53.
[2] J. Han and M. Kamber, Data Mining: Concepts and
Techniques. Second edition Morgan Kaufmann
Publishers.
[3] S.K. Pani, L.Panigrahy, V.H.Sankar, Bikram Keshari
Ratha, A.K.Mandal, S.K.Padhi, “Web Usage Mining: A
Survey on Pattern Extraction from Web Logs”,
International Journal of Instrumentation, Control &
Automation
[4] M. Eirinaki and M. Vazirgiannis, “Web mining for web
personalization”, ACM Transactions on Internet
Technology, Vol. 3, No. 1, 2003, pp. 1-27.
[5] Dunham, M. H. 2003. Data Mining Introductory and
Advanced Topics. Pearson Education.
[6] Ms.Dipa Dixit and Mr Jayant Gadge, “Automatic
Recommendation for Online Users Using Web Usage
Mining”, International Journal of Managing Information
Technology (IJMIT) Vol.2, No.3, August 2010.
[7] Burke, R.: Hybrid Recommender Systems: Survey and
Experiments. User Modeling and User-Adapted
Interaction 12(4), 331–370 (2002)
[8] Linden, G., Smith, B., York, J.: Amazon.com
Recommendations: Item-to-Item Collaborative Filtering.
IEEE Internet Computing 7(1), 76–80 (2003)
[9] Balabanovic, M., Shoham, Y.: Fab: Content-based,
Collaborative Recommendation. Communications of the
ACM 40(3), 66–72 (1997)
[10]Schafer, J.B., Frankowski, D., Herlocker, J., Sen, S.:
Collaborative filtering recommender systems. In: The
Adaptive Web, pp. 291–324. Springer Berlin /
Heidelberg (2007)
[11]Schafer, J.B., Konstan, J.A., Riedl, J.: E-commerce
recommendation applications. Data Mining and
Knowledge Discovery 5(1/2), 115–153 (2001)
[12]M. Perkowitz and O. Etzioni, "Towards adaptive Web
sites: Conceptual framework and case study," Artificial
Intelligence, vol. 118, pp. 245-275, 2000.
[13]M. Jalali, N. Mustapha, A. Mamat, Md N. Sulaiman,
“OPWUMP An architecture for online predicting in
WUM-based personalization system”, In 13th
International CSI Computer Science, Springer
Verlag,2008. 307.
[14]R.Agrawal and R.Srikant, “Mining Sequential Patterns,"
In Proceedings of International conferenceon data
engineering. pp. 3-14.
[15]Minghua Zhang, Ben Kao, Chi-Lap Yip, David Cheung,
“A GSP-based Efficient Algorithm for Mining Frequent
Sequences”
[16]J. Ayres, J. Gehrke, T. Yiu, and J. Flannick, “Sequential
pattern mining using a bitmap representation,”
Proceedings of ACM SIGKDD ‟02, pp. 429-435, July
2002.
[17]M. Zaki, “SPADE: An efficient algorithm for mining
frequent sequences,” Kluwer Academic Publisher‟s
Machine Learning, vol. 42, pp. 31-60, 2001.
[18]J. Han, J. Pei, B. Mortazavi-Asl, Q. Chen, U. Dayal, and
M.C. Hsu, “FreeSpan: Frequent patternprojected
sequential pattern mining,” Proc. ACM SIGKDD Int‟l
Conf. Knowledge Discovery and Data Mining (SIGKDD
‟00), pp. 355-359, Aug. 2000.
[19]J. Pei, J. Han, B. Mortazavi-Asl, and et al., “PrefixSpan :
Mining sequential patterns efficiently by prefix-
projected pattern growth,” Proc. Int‟l Conf. Data
Engineering (ICDE ‟01), pp. 215-224, Apr. 2001.
[20]N. Pasquier, Y. Bastide, R. Taouil, and L. Lakhal,
“Discovering frequent closed itemsets for association
rules,” 7th Int‟l Conf. on Database Theory, pp. 398-416,
Jan. 1999.
[21]Grahne G and Zhu J F, “Fast algorithm for frequent
itemset mining using FP-Trees,” IEEE Trans. on
Knowledge and Data Engineering, vol. 17, no. 10, pp.
1347-1362, 2005.
[22]J. Pei, J. Han, and R. Mao, “CLOSET: An efficient
algorithm for mining frequent closed itemsets,”
Proceedings of ACM DMKD ‟00, pp. 21-30, May 2000.
[23]M. Zaki and C. Hsiao, “CHARM: An efficient algorithm
for closed itemset mining,” Proceedings of SIAM‟s
SDM ‟02, pp. 457-473, Apr. 2002.
[24]J. Wang, J. Han, and J. Pei, “CLOSET+: Searching for
the best strategies for mining frequent closed itemsets,”
Proceedings of ACM SIGKDD ‟03, pp. 236-245, Aug.
2003.
[25]X. Yan, J. Han, and R. Afshar, “CloSpan: Mining closed
sequential patterns in large databases,” Proceedings of
SIAM‟s SDM ‟03, pp. 166-177, May 2003.
[26]J. Wang, J. Han, and Chun Li, “Frequent closed
sequence mining without candidate maintenance,” IEEE
TKDE, vol. 19, no. 8, pp. 1042-1056, Aug. 2007.
[27]Nancy P. Lin, Wei-Hua Hao, Hung-Jen Chen, Hao-En
Chueh and Chung-I Chang, “Fast mining of closed
sequential patterns,” WSEAS Transactions on
Computers, vol. 7, no. 3, Mar. 2008.
[28]S.Vijayalakshmi, V.Mohan and S.Suresh Raja, “Mining
Of Users‟ Access Behaviour For Frequent Sequential
Pattern From Web Logs”, International Journal of
Database Management Systems ( IJDMS ) Vol.2, No.3,
August 2010.
[29]Jian Pei, Jiawei Han, Behzad Mortazavi-asl, and Hua
Zhu, “Mining Access Patterns E_ciently from Web
Logs”, Natural Sciences and Engineering Research
Council of Canada (grant NSERC-A3723), the Networks
of Centres of Excellence of Canada (grant NCE/IRIS-3),
and the Hewlett-Packard Lab.
[30]Pragya Rajput, Joy Bhattacharjee, Roopali Soni, “A
Proposed Framework to Implicit Measures of User
Interests through Country and Predicting Users‟ Future
Requests in WWW”, International Journal of Soft
Computing and Engineering (IJSCE) ISSN: 2231-2307,
Volume-3, Issue-1, March 2013.
[31]Sizu Hou, Xianfei Zhang, "Alarms Association Rules
Based on Sequential Pattern Mining Algorithm," In
proceedings of the Fifth International Conference on
Fuzzy Systems and Knowledge Discovery, vol. 2,
pp.556-560, Shandong, 2008.
[32]R. Agrawal and R. Srikant, “Mining Sequential
Patterns”, In Proceedings of the 11th International
Conference on Data Engineering, pp. 3-14, Taipei,
Taiwan, 1995.

[33]Zhenglu Yang, Yitong Wang and Masaru Kitsuregawa,
"An Effective System for Mining Web Log", LNCS, vol.
3841, pp.40-52, 2006.

An Enhanced Approach for Detecting User's Behavior Applying Country-Wise Local Search

More Related Content

What's hot (20)

Similar to An Enhanced Approach for Detecting User's Behavior Applying Country-Wise Local Search (20)

More from IJSRD (20)

Recently uploaded (20)

An Enhanced Approach for Detecting User's Behavior Applying Country-Wise Local Search