29 ijcse-01238-7 sumathi

© 2015, IJCSE All Rights Reserved 153
International Journal of Computer Sciences and EngineeringInternational Journal of Computer Sciences and EngineeringInternational Journal of Computer Sciences and EngineeringInternational Journal of Computer Sciences and Engineering Open Access
Review Paper Volume-3, Issue-8 E-ISSN: 2347-2693
Privacy Enhanced Web Search Using MKSE in KNN
R.Sumathi1*
and V.Geetha2
1*,2
Department of Computer Science,
STET Women’s College, Mannargudi.
www.ijcseonline.org
Received: Jul /26/2015 Revised: Aug/06/2015 Accepted: Aug/23/2015 Published: Aug/30/ 2015
Abstract— Customized Web Look has established to make strides the quality of different look administrations on the Internet.
Due to tremendous data opportunities in the internet, the security assurance is exceptionally critical to preserve customer look
behaviors and their profiles. In the existing framework the summed up calculation specifically Covetous DP calculation were
connected to secure private data’s in modified look engine. The existing frame lives up to expectations failed to resist successive
and foundation information adversaries who has the broader foundation information such as richer relationship among topics.
The proposed framework introduces vector that point again quantization approach piecewise on the datasets which segmented
each column of datasets and that quantization approach is performed on each segment, utilizing the proposed approach which
later is again united to structure a transformed data set. The proposed work is implemented and is analyzed utilizing certain
parameters such as Precision, Recall, Frequency Measure, Distortion and Computational Delay.
Keywords— Security Protection, Customized Web Search, Profile, Vector, Quantization
I. INTRODUCTION
The web look motor is the most critical portal fat that point
again ordinary individuals looking at that point again
valuable data on the web. However, clients for the most part
experience failure and get improper results at the point
when look motors return material results that do not meet
their genuine intentions. A typical look motor gives
comparative set of results without considering of who
submitted the query. Therefore, the requirement arises to
have modified web look framework which gives outputs
appropriate to the customer as highly positioned pages.
Customized Web Look (PWS) is a general category of look
procedures which aims to give better look results, agreeing
to the person customer needs. So, fat that point again this
customer data has to be gathered and analyzed so that the
perfect look results needed fat that point again the customer
behind the issued question is to be given to the user. The
solution to this is Customized Web Look (PWS). It can for
the most part be categorized into two sorts specifically click
–log-based procedures and profile-based ones. The click-
log based procedures are fundamental and straightforward
.This framework performs the look based upon clicked
pages in the user’s question history. Although this
framework has been demonstrated to per structure reliably
and fundamentally well, it can just work on rehashed
questions from the same user, which is a solid limitation
and restricted at that point again certain applications. In
contrast, profile-based procedures make strides the look
experience with confused user-interest models created from
customer profiling techniques. Profile-based procedures can
be demonstrated more compelling at that point again
practically all sorts of queries, be that as it may are reported
to be improper under some situations. Although there are
reasons and considerations at that point again both sorts of
PWS techniques, the profile-based PWS has demonstrated
its more adequacy in improving the quality of web look
recently, with expanding usage of one’s individual and
behavioral data to record its users, which is for the most
part gathered implicitly with the help of question history,
browsing history, click-through data, bookmarks, customer
reports, and so on. Unfortunately, such sort of gathered
individual data can effectively uncover a entire scope of
user’s private life. The existing profile-based Customized
Web Look does not support runtime profiling. A customer
record is commonly summed up at that point again just once
offline, and used to customize all questions from a same
customer indiscriminatingly. Such “one record fits all”
method certainly has disadvantages given the mixed bag of
queries. One proof reported in is that profile-based
personalization might not indeed help to make strides the
look quality at that point again some commercial hoc
queries, though uncovering customer record to a server has
put the user’s security at risk. A better approach is to make
an online choice on whether to customize the question (by
uncovering the profile) and what to exposure in the
customer record at runtime. The existing procedures do not
take into account the customization of security
requirements. This presumably makes some customer
security to be overprotected while others insufficiently
protected. At that point again example, in, all the touchy
points are distinguished utilizing an outright metric called

International Journal of Computer Sciences and Engineering Vol.-3(8), PP(153-159) Aug 2015, E-ISSN: 2347-2693
surprise based on the data theory, expecting that the
intrigues with less customer record support are more
sensitive. However, this supposition can be doubted with a
fundamental counterexample: In the occasion that a
customer has a huge number of reports about “status,” the
surprise of this subject might commercial to a conclusion
that “status” is exceptionally general and not sensitive, in
spite of the truth which is opposite. Unfortunately, little
print that point again work can effectively address person
security needs amid the generalization. The calculation of
existing plan comprises of two heuristic rules by expecting
two terms tA and tB. The two heuristic rules used in
existing plan are
• Standard 1: Two terms that cover the record sets
with heavy overlaps might show the same interest.
The Jaccard capacity is used to ascertain the
closeness between two terms∪ Sim (tA, tB) = | D
(tA) ∩D (tB) | / | D(tA) D(tB) |. In the occasion
that Sim (tA, tB) > δ, where δ is another user-
indicated threshold, take tA and tB as comparative
terms representing the same interest.
• Standard 2: Specific terms regularly appear
together with general terms, be that as it may the
reverse is not true. At that point again example,
“badminton” tends to happen together with
“sports”, be that as it may “sports” might happen
with “basketball” at that point again “soccer”, not
necessarily “badminton”. Thus, tB is taken as a
child term of tA in the occasion that the condition
likelihood P(tA | tB )> δ, where δ is the same edge
in Standard 1. The existing plan calculation
comprises of two stages called Split and Buildup.
The taking after steps depict the Split process of
Client profile.
The problems with existing plan are the existing profile-
based PWS do not support runtime profiling, the existing
procedures do not take into account the customization of
security requirements, and the existing framework endures
from the modified security policy maintenance. Security
assurance range requires iterative customer communications
at that point again personalization. This created in
compelling results. They failed to secure data from
successive and foundation attackers.
The proposed plan contains Protection Preserving
Customized Web Look framework UPS (PP-PWS), which
can sum up prerecords at that point again each question
agreeing to user- indicated security requirements. Relying
on the definition of two conflicting metrics, specifically
personalization utility and security risk, at that point again
different leveled customer profile, we define the issue of
protection saving modified look as Risk Prerecords
Generalization, with its NP hardness proved. It has
fundamental be that as it may compelling speculation
calculation specifically Greedy IL to support runtime
profiling. While the previous tries to expand the separating
power (DP), the last endeavors to minimize the data
misfortune (IL). By exploiting a number of heuristics,
GreedyIL beats Covetous DP significantly. We give an
reasonable instrument at that point again the customer to
decide whether to customize a question in UPS. This choice
can be made some time as of late each runtime profiling to
upgrade the steadiness of the look results while avoid the
pointless presentation of the profile. The PP-PWS
framework at that point again UPS upgrades the steadiness
of the look quality and moves forward the security
assurance against diverse sort of attacks. It too maintains a
strategic distance from pointless presentation of the
customer pre-record and gives runtime profiling.
II. BACKGROUND
The existing profile-based Customized Web Look does not
support runtime profiling. A customer pre-record is
commonly summed up at that point again just once offline,
and used to customize all questions from a same customer
indiscriminatingly. Such “one pre-record fits all” method
certainly has disadvantages given the mixed bag of queries.
One proof reported in is that profile-based personalization
might not indeed help to make strides the look quality at
that point again some commercial hoc queries, though
uncovering customer pre-record to a server has put the
user’s security at risk. The existing procedures do not take
into account the customization of security requirements.
This presumably makes some customer security to be
overprotected while others insufficiently protected. At that
point again example, in, all the touchy points are
distinguished utilizing an outright metric called surprisal
based on the data theory, expecting that the intrigues with
less customer record support are more sensitive. However,
this supposition can be doubted with a fundamental
counterexample: In the occasion that a customer has a huge
number of reports about “status,” the surprise of this subject
might commercial to a conclusion that “status” is
exceptionally general and not sensitive, in spite of the truth
which is opposite. Unfortunately, little print that point again
work can effectively address person security needs amid the
generalization.

The figure (1) gives an outline of the entirety system. A
calculation is given at that point again the customer to
consequently assemble a different leveled customer pre-
record that speaks to the user’s understood individual
interests. General intrigues are put on a higher level and
specific intrigues are put on a lower level. Only some
portions of the customer pre-record will be exposed to the
look motor in accordance with a user’s own security
settings. A look motor wrapper is created on the server side
to consolidate a fractional customer pre-record with the
results returned from a look engine. Rankings from both
fractional customer prerecords and look motor results are
combined. The modified results are delivered to the
customer by the wrapper.
Existing Algorithm
The existing plan calculation comprises of two stages
specifically called Split process and Development process.
The taking after steps depict the Split process of Client
profile:
Step 1: The customer sends a question and the fractional
customer prorecord to the look motor wrapper, where the
fractional customer prorecord is spoken to by a set of <t, wt
> pairs.
Step 2: The List of customer prorecord entries is ordered
utilizing ascending at that point again descending based on
the esteem of the user.
Step 3: The wrapper calls the look motor to retrieve the
look result from the web. Each result comprises of a set of
links related to the query, where each join is given a rank
from search, called Look Rank. These links are passed to
the fractional customer profile.
Step 4: At that point again each of the returned join l, a
score called UP Score is figured by the fractional customer
prorecord as follows: (Σ×=tttfwlUPScore) where t is any
term in the fractional customer profile, and tf is the
recurrence of the term t in the webpage of the join l. An UP
Rank is relegated to each join agreeing to its UP Score, and
the join with the highest UP Score will be positioned first.
Step 5: The closeness of customer terms can be identified
and that covers the record sets with overlap of the customer
profile.
Step 6: The specific terms regularly appear together with
general terms of the customer pre-record and it can be split
based on the rank of the customer list.
Step 7: Re-positioning results by joining positions from
both MSN look and the fractional customer profile. The last
rank, PP Rank (Privacy-enhancing Customized Rank), is
figured as PP Rank = α*UP Rank + (1∈- α)*MSN Rank,
where the parameter α shows the weight relegated to the
rank from the fractional customer profile. In the occasion
that α=0, the customer pre-record is ignored, and the last
rank is decided by the customer pre-record in commercial of
the look motor at the point when α=1.
The taking after step describes the Development process of
Client profile:
Step 1: “interest” and “term” are indistinguish capable in
the content of the customer profile. The support of an
interest at that point again a term t is Sup (t), and S (t)
speaks to all the supporting reports at that point again term
t.
Step 2: ΣSup (t) =|D| is at that point again all terms t on the
leave node, where |D| speaks to the complete number of
supports received from individual data.
Step 3: According to likelihood theories, the possibility of
one interest (at that point again a term) can be figured as P
(t) =Sup (t)/|D the disadvantages in the existing plan are
• The existing profile-based PWS do not support
runtime profiling.
• The existing procedures do not take into account
the customization of security requirements.
• The existing framework endures from the modified
security policy maintenance.
• Security assurance range requires iterative
customer communications at that point again
personalization. This created in compelling results.
• Failed to secure data from successive and
foundation attackers.
III. REVIEWS OF EXISTING WORK
Look personalization is based on the fact that person clients
tend to have diverse preferences and that knowing the user’s
inclination can be used to make strides the criticalness of
the results the look motor returns. There have been
numerous endeavors to customize web search. These
endeavors for the most part differ in
• How to infer the customer preference, whether
explicitly by needing the customer to show data
about herself at that point again implicitly from the
user’s interactions,
• What kind of data is used to infer the user’s
preference?
• Where this data is gathered at that point again
stored, whether on the customer side at that point
again the server side, and

How this customer inclination is used to improve. Lidan
Shou, et.al, 2014, presented a client-side security assurance
framework called UPS at that point again modified web
search. UPS could possibly be received by any PWS that
catches customer pro-records in a different leveled
taxonomy. The framework allowed clients to indicate
modified security necessities by means of the different
leveled profiles. In addition, UPS too performed online
speculation on customer pro-records to secure the individual
security without compromising the look quality. It too tells
about where the data is gathered at that point again stored,
whether on the customer side at that point again the server
side.
Customized look is a promising way to make strides the
exactness of web look, and has been attracting much
attention recently. However, compelling modified look
requires collecting and aggregating customer information,
which regularly raises serious concerns of security
infringement at that point again numerous users. Indeed,
these concerns have ended up one of the main barriers at
that point again deploying modified look applications, and
how to do protection saving personalization is a great
challenge. Here we systematically examine the issue of
security preservation in modified search. We distinguish the
four levels of security protection, and investigate different
software architectures at that point again modified search.
We show that client-side personalization has focal points
over the existing server-side modified look administrations
in preserving privacy, and envision conceivable future
procedures to completely secure customer privacy.
Merits
• It catches customer pre-records in different leveled
taxonomy.
• It protects individual security without
compromising look quality.
• Improves exactness of web search.
Demerits
• Acts as barrier at that point again deploying
modified look applications
• It regularly raise serious concerns of security
infringement at that point again numerous clients.
Zhicheng, et.al, 2007, proposed modified look has been
used at that point again numerous years and numerous
personalization procedures have been investigated, it is still
unclear whether personalization is reliably compelling on
diverse questions at that point again diverse users, and
under diverse look contexts. The paper studies the issue and
gives some preliminary conclusions. We present a
substantial scale assessment framework at that point again
modified look based on question logs, and at that point
assess five modified look procedures (counting two click-
based and three profile-based ones) utilizing 12-day MSN
question logs. By breaking down the results, we uncover
that modified look has critical change over fundamental
web look on some questions be that as it may it has little
effect on other questions (e.g., questions with little click
entropy). It indeed harms look exactness under some
situations. Furthermore, we show that straightforward click-
based personalization procedures per structure reliably and
fundamentally well, while profile-based ones are unstable in
our experiments. We too uncover that both long-term and
short-term settings are exceptionally critical in improving
look execution at that point again profile-based modified
look strategies.
Merits
• It is reliably compelling on diverse questions at
that point again diverse users, and under diverse
look contexts.
• Customized Look has critical change over
fundamental web search.
Demerits
• It harms look exactness under some situations.
• Profile-based ones are unstable under these
experiments.
Susan T. Dumais, et.al, 2005, proposed look calculations
that consider a user’s pirate that point again
communications with a wide mixed bag of content to
customize that user’s current Web search. Rather than
relying on the unrealistic supposition that individuals will
precisely indicate their intent at the point when searching, it
pursues procedures that leverage understood data about the
user’s interests. This data is used to re-rank web look results
inside a criticalness feedback framework. It explore rich
models of customer intrigues built from both search-related
data such as beforehand issued questions and beforehand
visited web pages and other data about the customer such as
reports and email the customer has commercial and created.
The relook suggests that rich representations of the
customer and the corpus are critical at that point again
personalization be that as it may that it is conceivable to
approximate these representations. Web look motors (e.g.
Google, Yahoo, Microsoft Live Search, etc.) are generally
used to find certain data among a huge sum of data in a
insignificant sum of time. However, these valuable
instruments too posture a security danger to the users: web
look motors pre-record their clients by storing and breaking
down past searches submitted by them. To address this
security threat, current arrangements propose new
mechanisms that introduce a high cost in terms of reckoning
and communication. In this paper we present a novel
convention specially composed to secure the users’ security
in front of web look profiling. Our framework gives a
distorted customer pre-record to the web look engine. We
offer implementation details and computational and

communication results that show that the proposed
convention moves forward the existing arrangements in
terms of question delay. Our plan gives an affordable
overhead commercial while offering security benefits to the
users.
Merits
• It is used to find certain data among a huge sum of
data in a insignificant sum of time.
• Proposed convention moves forward the existing
arrangements in terms of question delay.
Demerits
• Some instruments they posture a security danger to
the clients
• This framework gives a distorted customer
prorecord to the web look motor.
IV. PROPOSED WORK
The proposed plan contains Protection Preserving
Customized Web Look framework UPS, which can sum up
prerecords at that point again each question agreeing to
user-indicated security requirements. Relying on the
definition of two conflicting metrics, specifically
personalization utility and security risk, at that point again
different leveled customer profile, we define the issue of
protection saving modified look as Risk Pre-record
Generalization, with its NP hardness proved. It has
fundamental and compelling speculation calculation
specifically Greedy IL, to support runtime profiling. While
the previous tries to expand the separating power (DP), the
last endeavors to minimize the data misfortune (IL). By
exploiting a number of heuristics, Greedy IL beats Covetous
DP significantly. We give a reasonable instrument at that
point again the customer to decide whether to customize a
question in UPS. This choice can be made some time as of
late runtime profiling to
Figure 2: Protection Preserving Customized Web Look
framework
UPS upgrade the steadiness of look results while avoid the
pointless presentation of the profile. A) Security assurance in PWS
System We propose a PWS framework called UPS that can sum
up prerecords in at that point again each question agreeing to user-
indicated security requirements. Two prescient measurements are
proposed to assess the security breach hazard and the question
utility at that point again different leveled customer profile. We
create two fundamental be that as it may compelling speculation
calculations at that point again customer prerecords allowing at
that point again query-level customization utilizing our proposed
metrics. We too give an online forecast instrument based on
question utility at that point again deciding whether to customize a
question in UPS. Extensive tests demonstrate the proficiency and
adequacy of our framework. (See Figure 2) B) Generating Client
Profile The speculation process has to meet specific prerequisites
to handle the customer profile. This is achieved by handling the
customer profile. At first, the process initializes the customer pre-
record by taking the indicated parent customer pre-record into
account. The process adds the inherited properties to the properties
of the neighborhood customer profile. Thereafter the process loads
the data at that point again the foreground and the foundation of
the map agreeing to the depicted selection in the customer profile.
Additionally, utilizing references empowers caching and is
accommodating at the point when considering an implementation
in a production environment. The reference to the customer pre-
record can be used as an identifier at that point again already
processed customer profiles. It allows performing the
customization process once, be that as it may reutilizing the result
various times.
However, it has to be made sure, that an update of the customer
pre-record is too propagated to the speculation process. This
requires specific update strategies, which check after a specific
timeout at that point again a specific event, in the occasion that the
customer pre-record has not changed yet. Additionally, as the
speculation process involves remote data services, which might be
updated frequently, the cached speculation results might ended up
outdated. Subsequently selecting a specific caching method
requires careful analysis. (See Figure 3) C) Coding and Encoding
in Security Insurance Technique the encoding and decoding
process of the cryptography framework is illustrated below:
Quantization is the method of preparing something from a
moderately huge at that point again continuous set of values (such
as the genuine numbers) to a moderately little discrete set (such as
the integers). The discrete cosine Tran structure (DCT) helps
separate the content into parts (at that point again spectral sub-
bands) of differing criticalness (with respect to the image's visual
quality). The DCT is comparative to the discrete Fourier trans
structure be that as it may utilizing just genuine numbers. There
are eight standard DCT variants, of which four are common. The
most fundamental variant of discrete cosine trans structure is the
type-II DCT, which is regularly called essentially "the DCT"; its
inverse, the type-III DCT, is correspondingly regularly called
essentially "the inverse DCT" at that point again "the IDCT". Two
related changes are the discrete sine changes (DST), which is
equivalent to a DFT of genuine and odd functions, and the
modified

Figure 3: Enhanced Security Insurance Framework
Discrete cosine changes (MDCT), which is based on a DCT of
overlapping data. D) Calculation of Proposed Design the Greedy
IL calculation moves forward the proficiency of the speculation
utilizing heuristics based on a few findings. One critical finding is
that any prune-leaf operation reduces the separating power of the
profile. In other words, the DP displays monotony-city by prune-
leaf. The benefits of making the above runtime choice are, it
upgrades the steadiness of the look quality and it maintains a
strategic distance from the pointless presentation of the customer
profile. Therefore, Greedy IL is anticipated to fundamentally out
structure Covetous DP. The steps at that point again Greedy IL
calculation are
Step 1: In the occasion that G’ is a pre-record acquired by
applying a prune leaf operation on G, at that point DP(q; G) ≥
DP(q, G’).
Step 2: Specifically, each hopeful operated that point again in the
queue is a tuple like op = (t, IL (t, Gi)), where t is the leaf to be
pruned by op and IL (t, Gi), shows the IL incurred by pruning t
from Gi.
Step 3: The iterative process can terminate whenever ϑ-hazard is
satisfied.
Step 4: The second term (TS (q, G) remains unchanged at that
point again any pruning operations until a single leaf is cleared out
(in such case the just choice at that point again pruning is the
single leaf itself).
Step 5: In C1, t is a hub with no siblings, and In C2, t is a hub with
siblings. The case C1 is simple to handle. However, the
assessment of IL in case C2 requires introducing a shadow kin of t.
Step 6: Each time in the occasion that we attempt to prune t, we
actually merge t into shadow to get a new shadow leaf shadow0,
together with the inclination of t,
Step 7: Prune-leaf just operates on a single subject t. Thus,
it does not impact the IL of other hopeful administrators in
Q. While in case C2, pruning t incurs re-reckoning of the
inclination values of its kin nodes.
Step 8: Once a leaf subject t is pruned, just the hopeful
administrators pruning t’s kin points need to be updated in
Q. In general, Greedy IL traces the data misfortune in
commercial of the separating power. This saves a parcel of
computational cost.
The focal points Enhanced Security Insurance Framework is
as follows:
• It upgrades the steadiness of the look quality
• Improves the security assurance against diverse
sort of attacks
• It maintains a strategic distance from the pointless
presentation of the customer pre-record.
• It gives runtime profiling.
V. RESULTS AND DISCUSSION
The taking after execution parameters are just used in
security assurance system evaluation. The existing approach
is looked at with proposed approach utilizing these
assessment parameters. The framework is assessed in terms
of Precision, Recall, F-measure, Computational Delay and
Distortion. Accuracy - It is measure of correctly anticipated
reports by the framework among all the anticipated
documents. It is characterized as the number of material
reports retrieved by a look divided by the complete number
of reports retrieved by that search.
Accuracy = number of right results/ number of all returned
results
No. of Precision
Categories User Profiles Existing Proposed
20 NG 412 75% 98%
Sports 300 61% 96%
Health 669 58% 90%
Society 442 68% 91%
Nearby News 254 68% 73%
Accuracy Comparative Evaluation of Accuracy utilizing
Greedy IL Algorithm: The proposed approach exactness
level is high at the point when looked at with the existing
one. Frequency-Measure: F-measure combines exactness
and review and is the harmonic mean of exactness and
recall. F-measure=2*(precision*recall/exactness + recall)
F-Measure Result analysis
No. Of User F-Measure
Categories Profiles Existing Proposed
20 NG 412 66.67% 75%
Sports 300 69.22% 82.35%
Health 669 68.28% 86.35%
Society 442 75.15% 87.89%
Local News 254 77.83% 81.40%

VI. CONCLUSION AND FUTURE WORK
The remarkable development of data on the Web has forced
new challenges at that point again the construction of
compelling look engines. The proposed work gives data on
customer customizable security preserving look framework-
UPS at that point again Customized Web Search. UPS
could possibly be received by any PWS that catches
customer records in a different leveled taxonomy. The
framework allowed clients to indicate modified security
necessities by means of the different leveled profiles.
Another critical conclusion we revealed in this proposed
work is that personalization does not work equally well
under different situations. The click entropy is used to
measure mixed bag in data needs of clients under a query.
Experimental results showed that modified Web look yields
critical improvements over generic Web look at that point
again questions with a high click entropy. At that point
again the questions with low click entropy, personalization
procedures performed similarly at that point again indeed
worse than generic search. As modified look commercial
diverse adequacy at that point again diverse kinds of
queries, we argued that questions should not be handled in
the same way with regard to personalization. The proposed
click entropy can be used as a fundamental measurement on
whether a question should be personalized. At that point
again future work, we try to resist adversaries with request
foundation information counting exclusiveness, sequentially
and so on at that point again the capacity to catch a
arrangement of questions from the victim.
ACKNOWLEDGEMENT
1
R.SUMATHI, M.Phil Research Scholar, PG and Research
Department of Computer Science, STET Women’s College,
Mannargudi.
2
Mrs. V. GEETHA M.Sc., M.Phil., B.Ed., Head, PG and
Research Department of Computer Science, STET
Women’s College, Mannargudi.
REFERENCES
[1] I. F. Akyildiz, X. Wang, and W. Wang, “Wireless Mesh
Networks: A Survey”, Computer Networks and ISDN
Systems, Vol.47, Issue-2, 2005, pp.445-487.
[2] I. F. Akyildiz, and X. Wang, “A Survey on Wireless Mesh
Networks”, IEEE Radio Communications, Vol.43, Issue-3,
2005, pp.23-30.
[3] M. Lee et al., “Emerging Standards for Wireless Mesh
Technology”, IEEE Wireless Communications, Vol.13,
Issue-4, 2006, pp.56-63.
[4] N.B. Salem, and J-P Hubaux, “Securing Wireless Mesh
Networks”, IEEE Wireless Communications, Vol.13, Issue-
2, 2006, pp.50-55.
[5] S. Han, E. Chang, L. Gao, T. Dillon, T., Taxonomy of
Attacks on Wireless Sensor Networks, in the Proceedings of
the 1st European Conference on Computer Network
Defence (EC2ND), University of Glamorgan, UK, Springer
Press, SpringerLink Date: December 2007.
[6] C. Karlof and D. Wagner, “Secure routing in wireless sensor
networks: attacks and countermeasures,” Ad Hoc Networks
1, 2003, pp. 293-315.
[7] Y. Yang, Y. Gu, X. Tan and L. Ma, “A New Wireless Mesh
Networks Authentication Scheme Based on Threshold
Method,” 9th
International Conference for Young Computer
Scientists (ICYCS-2008), 2008, pp. 2260-2265
[8] He Bai ; Ke Chen ; Gang Chen,” Supporting Privacy
Protection in Personalized Web Search” Published
in:Knowledge and Data Engineering, IEEE Transactions on
(Volume:26 , Issue: 2 )Page(s):453 – 467, 2013
[9] Wanying Ding ; Mengwen Liu ; Xiaoli Song more authors,”
Scalable user intent mining using a multimodal Restricted
Boltzmann Machine”. Published in:Computing, Networking
and Communications (ICNC), 2015 International
Conference of Date of Conference 16-19 Feb. 2015
Page(s):618 – 624.
[10] Lina Zhou ; Dongsong Zhang,” An integrated method for
hierarchy construction of domain-specific terms” Published
in:Computer and Information Science (ICIS), 2014
IEEE/ACIS 13th International Conference onDate of
Conference:4-6 June 2014 Page(s):485 – 490
[11] Shekhar, S. ; Jalal, A.S.,” Semantic Based Image Retrieval
using multi-agent model by searching and filtering
replicated web images”. Published in:Information and
Communication Technologies (WICT), 2012 World
Congress ofDate of f Conference:Oct. 30 2012-Nov. 2 2012
Page(s):817 - 821

29 ijcse-01238-7 sumathi

Recommended

Recommended

More Related Content

Viewers also liked

Viewers also liked (7)

Similar to 29 ijcse-01238-7 sumathi

Similar to 29 ijcse-01238-7 sumathi (20)

More from Shivlal Mewada

More from Shivlal Mewada (20)

Recently uploaded

Recently uploaded (20)

29 ijcse-01238-7 sumathi