IJRET: International Journal of Research in Engineering and Technology

eISSN: 2319-1163 | pISSN: 2321-7308

A SURVEY ON IDENTIFICATION OF RANKING FRAUD FOR

MOBILE APPLICATIONS
Nandini B1, A.Ananda Shankar2
1

MTech Scholar Department of Computer Science & Engineering, Reva ITM, Kattigenahalli, Yelahanka, Bangalore560064
2
Associate Professor Department of Computer Science & Engineering, Reva University, Kattigenahalli, Yelahanka
Bangalore-560064

Abstract
Now a day, mobile App is an exceptionally prevalent and surely understood idea because of the quick progression in the portable
innovation and cell phones. Because of the extensive number of versatile Apps, ranking fraud is the key test before the versatile
App market. In this paper we are proposing a ranking fraud discovery framework for portable Apps. The proposed framework
mines the leading sessions, for example, leading sessions of portable applications to precisely find the ranking fraud. Other than
this, by displaying Apps ranking, rating and review practices utilizing measurable theories tests, we examine three sorts of
confirmations, they are ranking based proofs, rating based proofs and review based confirmations. Proposed an aggregation
method to combine all the proof for fraud detection.

Keywords: Ranking Fraud, Fraud For Mobile, Identification Mobile Fraud

--------------------------------------------------------------------***---------------------------------------------------------------------1.

INTRODUCTION

The mobile applications are growing in a faster rate now a

day, for instance, at the end of 2013 April, there are more
than 1.6 million Apps at Apple's App store and Google Play.
For simulating the growth of the app, many app stores
released leaderboards. This became the most important way
to promote mobile apps. If ranking is high in the
leaderboard, number of downloads and revenue increases.
To increase the download of the app, developers manipulate
the chart ranking, using human water armies and bot-farms.
For example, for a little engineer who's just discharged an
iOS application, a position close to the highest point of
Apple's App Store rankings could mean the contrast
between a million dollars in income and Top Ramen for
supper. So it's nothing unexpected that a few coders attempt
to cheat the framework and that Apple tries to stop it.
Apple got serious about various questionable promoting
firms that utilization programming bots and multitudes of
human clients to download applications as a group, pushing
the titles to prominent situating inside of the App Store's
"Top Free" rankings outline. It isn't clear what number of
applications has been influenced by the bans, yet in any
given month, Apple brings down 5,000 applications for an
assortment of reasons, as per application seek firm
Xyologic. The designers' worries came during a period
when general App Store downloads were strikingly
diminishing. Matthus Krzykowski, the CEO of Xyologic,
said that the volume of downloads on the U.S. Application
Store has fallen 25 percent since January. The diminishing,
he says, can to a limited extent be ascribed to Apple's
crackdown on bots and its choice to battle "incentivized"
introduce. Recently, Apple made expansive move against
outsider promoting administrations firms since proof
demonstrated a few advertisers were controlling the App

Store's top rankings utilizing robotized PC frameworks

that is, software bots and additionally multitudes of
human clients. The advertisers utilized these instruments to
download their customers' iOS applications altogether,
falsely blowing up Apple's store ranking. The ranking fraud
will not happens always, so need to find the exact time when
it happens. This respects to detect the local anomaly.
Detecting of fraud manually is difficult, so have to use the
scalable method to automatically detect the ranking fraud. it
is difficult to recognize the evidence and pledge the
evidences, which leads to discover implicit fraud patterns.
The ranking of the mobile apps are not always high, it will
be only in particular leading events, which leads several
leading sessions. The ranking fraud takes place in these
sessions. Using Mining Leading Sessions will distinguish
the leading events and leading session by scanning the
historical ranking records only once. The ranking of the
fraud apps will differ in each leading session comparing to
the normal apps. The ranking pattern differs in three phases:
rising, maintaining and recession phase. The app which is
fraud will have a rise to peak in short time, but it will not
stay for long in maintaining phase. Using Guassian
approximation and classic maximum-likelihood estimation
(MLE) will find the ranking fraud. To find the fraud not
only ranking is important but also rating and review is
needed. After getting the information, will use the
unsupervised evidence aggregation to integrate the
evidences.

2. LITERATURE SURVEY
While there are some associated work regarding ranking
fraud, such as:

_______________________________________________________________________________________
Volume: 05 Issue: 03 | Mar-2016, Available @ http://www.ijret.org

473

IJRET: International Journal of Research in Engineering and Technology

2.1 Web Ranking Spam Detection

Vyas Krishna Maheshchandra, and Prof. Ankit P. Vaishnav
done an overview on an eminent source of assembling the
analysis on particular item where people will write their
reviews based on the item. Some people will mislead by
writing the wrong comments. This leads to the review spam.
So, they used the diverse methods acquainted with recognize
the Review spam with their outcome, methods such as
Vector Space, SVM, SLM, LM and I-match. Some of the
time individuals might run over the off-base conclusions
called as survey spam [1].
Ms. Meenal M. Shingare, and Prof. S. R. Chaudhary worked
on the recognition of the spam in web which makes trick or
deceive to web search tool. Clients have a harder time
finding the data they need, and internet searchers need to
adapt to a swelled corpus, which thus causes their expense
per question to increment. In this manner, web crawlers
have a solid impetus to get rid of spam website pages from
their file. They used a Language model for an effective
detection of web spam which merges latest link-based
features build on a classifier. For executing this SVMs
calculation is utilized which go for scanning for a hyper
plane that isolates two classes of information with the
biggest edge [16].
Shrijina Sreenivasan and B.Lakshmipathi proposed the work
which identifies with the correlation of web spam location
utilizing three unsupervised learning strategies, SOM, HMM
and ART as opposed to the directed methods SVM. The
regulated procedures experiences the disadvantage that it
functions admirably just with vast datasets and is not
expected to be a genuine time application. After a near
study, the proposed strategy is found to yield a higher
execution than the current supervised learning techniques
[21].

2.2 Online Review Spam Detection

Michael Crawford*, Taghi M. Khoshgoftaar, Joseph D.
Prusa, Aaron N. Richter and Hamzah Al Najada did the
survey on Online spam recognition to give a solid and
thorough near investigation of momentum exploration on
distinguishing survey spam utilizing different machine
learning strategies and to devise approach for leading further
examination. They used the noticeable machine learning
methods that have been proposed to take care of the issue of
audit spam recognition and the execution of various
methodologies for grouping and recognition of audit spam
[2].
Sushant Kokate, and Bharat Tidke worked on finding the
online trumped up feedback for the movies. For doing this
they used a classifier called J48, it will produce ARFF from
the unmistakable components to identifying the untruthful
audits. As a feature of future work, they fuse survey
spammer discovery into the survey location and the other
way around. Investigating approaches to learn conduct
designs identified with that spamming in order to enhance
the precision of the present relapse model is moreover [3].

eISSN: 2319-1163 | pISSN: 2321-7308

Amir Karam, and Bin Zhou proposed to utilize

classifications of lexical semantic and linguistic features in
the recognition of online spam audits. The examination
results appeared that by consolidating numerous linguistic
features of surveys, the recognition execution of spam audits
can be incredibly enhanced, contrasting and the best in class
strategies [4].

2.3 Mobile App Recommendation

Xiao Xia, Xiaodong Wang, and Xingming Zhou proposed a
novel method recommendation by making us of global
details about applications, where the dangerous development
of portable applications offers meet people's high
expectations of application revelation. They also produced
suggestions by both examining the metadata and measuring
the closeness between applications, utilizing the Latent
Semantic Index strategy. They also proposed an assorted
qualities measurementbased advancement structure for the
improvement of versatile application recommender
frameworks. To execute the structure, they assist show the
framework advancement as a multi-criteria streamlining
issue and plan a rank accumulation scheme to settle it [22].
Hengshu Zhu, Hui Xiong,Yong Ge and Enhong Chen
developed a phony application recommender system by
using the privacy and security alertness. They composed a
versatile and programmed approach for evaluating the
security dangers of Mobile Apps. To examine both Apps
'reputation and clients' security inclinations for suggestions,
they presented an adaptable App suggestion strategy in light
of the advanced portfolio hypothesis. Especially, they too
built up an App hash tree to effectively turn upward Apps in
suggestion [17].
Bin Liu, Deguang Kong, Lei Cen, Neil Zhenqiang Gong,
Hongxia Jin and Hui Xiong presented the first methodical
study
on
consolidating
both
interest-usefulness
collaborations and clients' security inclinations to perform
customized Application suggestions. They recommended a
method for taking the trade off among functionality &
clients privacy inclination [5]
Donghwan Bae, Keejun Han ; Park, J. ; Yi, M.Y., proposed
AppTrends, which consolidates a chart based strategy for
application suggestion in the Android OS environment.
Their trial results got from the field utilization record of
more than 4 million applications unmistakably demonstrate
that the proposed chart based suggestion model is more
precise than the Slope One Model [6]

2.4 Internet Water Armies

Kun Wang, Yang Xiao, and Zhen Xiao quantified the
Internet Water Army's conduct from numerous
measurements. At that point they chose a few successful
components as the preparation model and utilize machine
learning strategies for order. In view of the conduct of
clients review the remark, they proposed a model to quantify
the impact of Internet Water Army. With a specific end goal
to lessen the impact coefficient, they proposed another direct
time many-sided quality online calculations named

_______________________________________________________________________________________
Volume: 05 Issue: 03 | Mar-2016, Available @ http://www.ijret.org

474

IJRET: International Journal of Research in Engineering and Technology

MEIWA. The new calculation results demonstrated that the

impact is decreased to one 6th of the succession procedure
which is utilized as a matter of course with guaranteeing
clients' survey remarks propensities [18].
Wang Xiang, Zhang Zhilin ; Yu Xiang ; Jia Yan ; Zhou
Bin ; Li Shasha consider the individual and gathering
qualities of sorted out notices. A classifier is developed
taking into account the individual and gathering attributes to
recognize them. Broad test results on three genuine datasets
show that our strategy in light of individual and gathering
qualities utilizing SVM model (IGCSVM) is compelling in
recognizing sorted out publications and superior to anything
existing techniques. They examine finding the promoters
taking into account the distinguished sorted out publications
of our IGCSVM technique. Their analyses demonstrate that
it is viable in identifying promoters [7].
Survey on some important challenges for detecting fraud in
mobile application is as follows:
Positioning extortion does not continuously happen in the
entire life cycle of an App, so we require identifying the
time when misrepresentation happens. Such test can be
viewed as identifying the local anomaly rather than global
anomaly of portable Apps.

2.5 Automatically Detecting of Ranking

Vaishali Date, Dipali Dongare, Pooja Jadhav, Tejal Wayal,
and Asmita Mali did the survey for finding out the
positioning scam founded in the mobile applications. They
proposed a system which collects all the positioning details
along this feedback and rating results are collected. Then
they used the aggregation procedure to do the hypotheses for
finding the actual position of the application. They will
recover framework with information gathered structure
application particle play store for long stretch of time [8].
Tejaswini B. Gade, Prof. Nilesh G. Pardeshi takes a look on
different existing procedure for online and web spam based
on the positioning of the apps. They find out the local
anomaly in the leading sessions and based on these they find
out the ranking of the app correctly in the leader board.
Finally combined all the evidences and optimized the results
[9].
Raghuveer Dagade, Prof. Lomesh Ahire reviewed the
detection of ranking for the mobile. They first find that
positioning misrepresentation happen in driving sessions
and gave a technique to digging driving sessions for each
App from its chronicled positioning records. Moreover, they
proposed an improvement based accumulation strategy to
coordinate all the confirmations for assessing the
dependability of driving sessions from portable Apps. That
every one of the confirmations can be displayed by factual
speculation tests for the novel viewpoint of this
methodology, in this way it is anything but difficult to be
stretched out with other confirmations from space learning
to identify positioning extortion [10].

eISSN: 2319-1163 | pISSN: 2321-7308

Prof. Amruta Gadekar, Rajani Gupta, Mamta Kumari,

Monika Munswamy worked on fraud detection from users
recommendation. A novel point of perspective of the
procedure is that each one of the verifications can be shown
by quantifiable hypothesis test; along these lines it is
definitely not hard to be connected with various affirmations
from space data to recognize situating deception. They did
intensification of situating distortion area methodology is
performed with other convenient App related organizations
[11]
Prajakta Gayke and prof. Sanjay Thakre add to a situating
blackmail disclosure structure for versatile Apps. They
proposed a change based aggregate framework to
consolidate each one of the verifications for evaluating the
legitimacy of driving sessions from compact Apps. A novel
perspective of this approach is that each one of the
verifications can be shown by quantifiable hypothesis tests;
along these lines it is definitely not hard to be connected
with various affirmations from space data to recognize
situating deception. They plan to think more practical
deception affirms and dismember the unmoving relationship
among rating, overview and rankings. They want to enhance
their situating distortion area approach with other
convenient App related organizations [12].
Catarina Moreira, Bruno Martins, Pavel Calado contended
that unsupervised rank total techniques give a sound
methodology for consolidating different estimators of
mastery, got from the printed substance, from the diagram
structure of the group of specialists, and from master profile
data. Probes a dataset of scholastic productions indicate
extremely aggressive results in terms of P@5 and MAP,
bearing witness to for the ampleness of the proposed
approaches. This is especially fascinating to the application
area of scholastic master hunt; subsequent to the
significance judgments required by directed methodologies
are just barely accessible [13].
V.Mural Krishnan, J.Joseph Elango, T.Mohanraj,
Ms.Pushpalatha tend to diagram up an arranging confusion
conspicuous verification framework for adaptable Apps.
They tend to saw arranging based by and large
certifications, rating affirmations and audit based
attestations for various arranging compulsion moreover,
they tend to foreseen accomplice change based generally
accumulation procedure to sort out every one of the proofs
for assessing the way of driving sessions from moveable
Apps [14].

2.6 Latent Dirichlet Allocation (LDA)

Ostrowski, D.A. investigate subject considering so as to
demonstrate the systems of Latent Dirichlet Allocation
which is a generative probabilistic model for a gathering of
discrete information. They assess this procedure from the
point of view of grouping and also distinguishing proof of
foremost subjects as it is connected to a separated
accumulation of Twitter messages. Tests demonstrate that
these strategies are powerful for the recognizable proof of

_______________________________________________________________________________________
Volume: 05 Issue: 03 | Mar-2016, Available @ http://www.ijret.org

475

IJRET: International Journal of Research in Engineering and Technology

sub-points and in addition to bolster order inside of

substantial scale corpora [15].

eISSN: 2319-1163 | pISSN: 2321-7308

expounded to three capacities to get such positioning from

misrepresentation confirmations.

Bhutada.S, Balaram, V.V.S.S.S. ; Bulusu, V.V. attempted an

endeavor is made to propose a multilevel grouping model
utilizing Latent Dirichlet Allocation (LDA) approach. In
spite of the fact that the presence of Latent Dirichlet
Allocation (LDA) is seen in the writing yet a changed model
for multilevel arrangement is exhibited which is free of any
dialect. Keeping in mind the end goal to accomplish such
model numerous current recommendations were viewed as
like PLSI, which utilizes the Exceptional Maximization
(EM) technique just to prepare the inert classes. The
iterative procedure of Latent Dirichlet Allocation (LDA),
which yields the multilevel order of the corpus. Theme
Modeling is utilized to find the concealed things that swarm
the gathering to clarify the reports as indicated by the new
subjects [19].
Guolong Liu, Xiaofei Xu ; Ying Zhu ; Li Li propose a multicharacteristic inert dirichlet allotment (MA-LDA) model, a
subject examination model in which the time and label
properties of miniaturized scale online journals are
consolidated into LDA model. By presenting a period
variable about the time characteristic, MA-LDA model can
choose whether a word ought to show up in hotly debated
issues or not. Applying label trait permits MA-LDA model
to rank the center words high in results so that the
expressiveness of results can be enhanced over the
customary LDA model. Exact assessment on genuine
information sets show our strategy can identify hotly
debated issues precisely and productively with more terms
connected with each hotly debated issue found. Our study
gives solid proof of the significance of the worldly figure
hotly debated issues extraction [20].

3. EXISTING SYSYTEM
In the literary texts, while there are some associated work,
for example, web ranking spam recognition, online review
spam identification and mobile App suggestion, the issue of
identifying ranking fraud for versatile Apps is still underinvestigated. As a rule, the related works of this study can be
assembled into three classifications. The primary
classification is about web spam detection, second is
identifying online review spam and finally the last
classification incorporates the study on mobile app
suggestion.

4. PROPOSED SYSTEM
With the expansion in the quantity of web Apps, to identify
the fake Apps, we have proposed a basic and powerful
calculation which recognizes the leading sessions of each
Application in light of its chronicled positioning of records.
By examining the ranking behavior of apps, we come across
that the fraud apps frequently has dissimilar patterns for
ranking compared with the normal apps in every leading
sessions. Subsequently, will perceive few extortion
confirmations from applications chronicled records and

Fig 1: System Framework

Further we propose two sorts of fraud evidence taking into
account Apps review and ratings. It mirrors some
peculiarity designs from Apps' authentic rating and survey
records. Fig. 1 shows the structure of our positioning
extortion framework for versatile applications.
The leading sessions of mobile applications are evidence of
interval of popularity, so these driving sessions will include
just positioning control. Subsequently, the issue of
recognizing positioning extortion is to recognize dangerous
driving sessions. Together with the essential errand is to
take out the main sessions of a versatile application from its
chronicled positioning records.
There are two principle stages for identifying the ranking
fraud:
1. Recognizing the leading sessions.
2. Recognizing the evidences of the ranking fraud.
Recognizing the leading sessions:
First and foremost mining driving sessions has two sorts of
ventures to do with portable extortion applications. To start
with, from the Applications authentic positioning records,
divulgence of driving occasions is done and after that
second joining of adjacent driving occasions is done which
appeared for building driving sessions. Completely, some
specific count is appeared from the pseudo code of mining
sessions of given versatile App also, that estimation can
recognize the particular examining so as to drive events and
sessions verifiable records one by one.
Recognizing the confirmation of the fraud in ranking.

Evidence for Ranking

It infers different leading events will be present in a
particular leading session. Subsequently by examination of
key behavior of driving occasions for finding blackmail
affirmations moreover for the application chronicled
positioning records, it is been watched that a specific

_______________________________________________________________________________________
Volume: 05 Issue: 03 | Mar-2016, Available @ http://www.ijret.org

476

IJRET: International Journal of Research in Engineering and Technology

positioning example is always satisfied by application

positioning conduct in a main occasion.

Evidence for Rating

Past ranking found proofs are useful for recognizable proof
reason yet it is definitely not adequate. Determining the
"limit time consumption" matter, misrepresentation
confirmations acknowledgment is arranged because of app
authentic records of rating. While we probably am aware
that rating is been finished in the wake of downloading it by
the client, also, in the event that the rating is high in
leaderboard impressively i.e., pulled in via the greater part
of portable application clients. All of a sudden, the
assessments in the midst of the fundamental session offers
climb to the abnormality plan which happens in the midst of
rating deception. These credible records can be used for
making rating based affirmations.

Evidence of Reviews
We are familiar with the survey which contains a few
printed remarks as audits by application customer and some
time as of late download or using the application customer
generally jump at the chance to escape the surveys given by
a vast bit of the customers. In this way, albeit due to some
past deals with survey spam revelation there still issue on
finding the adjacent quirk of audit in driving sessions. So in
light of uses audit practices, using so as to position extortion
in portable application are identified misrepresentation
confirmations.
These three verifications will be combined by an
unsupervised evidence-aggregation strategy for assessing
the validity of leading sessions from portable apps. To
extract the evidences we are using the statistical hypotheses
tests for mobile applications. The ranking fraud detection
structure is adaptable and can be reached out with other area
created confirmations for ranking fraud detection. Finally,
we will evaluate the proposed structure with certifiable App
data assembled from the Apple's App store for a long time
range, i.e., over two years.

CONCLUSION
This paper surveys different existing strategies utilized for
web spam recognition, which is identified with the
positioning extortion for portable Apps. Additionally, we
have seen references for online survey spam identification
and versatile App suggestion.
By extracting the leading sessions of versatile Apps, we
intend to find the ranking fraud. The leading sessions works
for identifying the nearby inconsistency of App rankings.
The framework expects to distinguish the ranking frauds
taking into account three sorts of confirmations, for
example, ranking, rating and review based proofs. Further,
an optimization based aggregation strategy joins all the three
proofs to distinguish the fraud.

eISSN: 2319-1163 | pISSN: 2321-7308

REFERENCES
[1]. Vyas Krishna Maheshchandra, and Prof. Ankit P.
Vaishnav A Survey on Review Spam Detection
techniques International Journal of Engineering
Research & Technology (IJERT) ISSN: 2278-0181
IJERTV4IS040505 www.ijert.org , Vol. 4 Issue 04,
April-2015.
[2]. Michael Crawford, Taghi M. Khoshgoftaar, Joseph D.
Prusa, Aaron N. Richter and Hamzah Al Najada
Survey of review spam detection using machine
learning techniques Crawford et al. Journal of Big
Data (2015) 2:23 DOI 10.1186/s40537-015-0029-9
[3]. Sushant Kokate, and Bharat Tidke Fake Review and
Brand Spam Detection using J48 Classifier
International Journal of Computer Science and
Information Technologies, Vol. 6 (4) , 2015, 35233526
[4]. Amir Karam, and Bin Zhou Online Review Spam
Detection by New Linguistic Features In iConference
2015 Proceedings.
[5]. Bin Liu, Deguang Kong, Lei Cen, Neil Zhenqiang
Gong, Hongxia Jin and Hui Xiong Personalized
Mobile App Recommendation: Reconciling App
Functionality and User Privacy Preference February
26, 2015, Shanghai, China. Copyright 2015 ACM
978-1-4503-3317-7/15/02 ...$15.00.
[6]. Donghwan Bae, Keejun Han ; Park, J. ; Yi, M.Y. ,
AppTrends:
A
graph-based
mobile
app
recommendation system using usage history Big Data
and Smart Computing (BigComp), 2015 International
Conference.
[7]. Wang Xiang, Zhang Zhilin ; Yu Xiang ; Jia Yan ; Zhou
Bin ; Li Shasha Finding the hidden hands: a case
study of detecting organized posters and promoters in
SINA weibo Communications, China (Volume:12
, Issue: 11 ) November 2015.
[8]. Vaishali Date, Dipali Dongare, Pooja Jadhav, Tejal
Wayal, and Asmita Mali A Survey on Recognize the
Ranking Scam Occurred in Mobile Apps International
Journal of Innovative Research in Computer and
Communication Engineering (An ISO 3297: 2007
Certified Organization) Vol. 3, Issue 12, December
2015
[9]. Tejaswini B. Gade, Prof. Nilesh G. Pardeshi A Survey
on Ranking Fraud Detection Using Opinion Mining for
Mobile Apps International Journal of Advanced
Research
in
Computer
and
Communication
Engineering Vol. 4, Issue 12, December 2015
[10]. Raghuveer Dagade, Prof. Lomesh Ahire Review: A
Ranking Fraud Detection System for Mobile Apps
International Journal of Innovative Research in
Computer and Communication Engineering (An ISO
3297: 2007 Certified Organization) Vol. 3, Issue 11,
November 2015
[11]. Prof. Amruta Gadekar, Rajani Gupta, Mamta Kumari,
Monika Munswamy Detection of Ranking Fraud for
Mobile App and Prevention from Users
Recommendation International Journal of Innovative
Research
in
Computer
and
Communication

_______________________________________________________________________________________
Volume: 05 Issue: 03 | Mar-2016, Available @ http://www.ijret.org

477

IJRET: International Journal of Research in Engineering and Technology

eISSN: 2319-1163 | pISSN: 2321-7308

Engineering (An ISO 3297: 2007 Certified

Organization) Vol. 3, Issue 9, September 2015
[12]. Prajakta Gayke and prof. Sanjay Thakre Detection of
Ranking Fraud for Mobile App IOSR Journal of
Computer Engineering (IOSR-JCE) 2015
[13]. Catarina Moreira, Bruno Martins, Pavel Calado
Using Rank Aggregation for Expert Search in
Academic Digital Libraries arXiv:1501.05140v1
[cs.IR] 21 Jan 2015
[14]. V.Mural Krishnan, J.Joseph Elango, T.Mohanraj,
Ms.Pushpalatha Tracking and Altering Duplicate
Mobile Apps in Website International Journal for
Research in Technological Studies| Vol. 2, Issue 4,
March 2015 | ISSN (online): 2348-1439
[15]. Ostrowski, D.A Using latent dirichlet allocation for
topic modelling in twitter Semantic Computing
(ICSC), 2015 IEEE International Conference
[16]. Ms. Meenal M. Shingare, and Prof. S. R. Chaudhary
Web Spam Recognition through Classification
Algorithms 2014, IJARCSSE
[17]. Hengshu Zhu, Hui Xiong,Yong Ge and Enhong Chen
Mobile App Recommendations with Security and
Privacy Awareness KDD14, August 2427, 2014,
New York, NY, USA. Copyright 2014 ACM 978-14503-2956-9/14/08 ...$15.00
[18]. Kun Wang, Yang Xiao, and Zhen Xiao Detection of
Internet Water Army in Social Network International
Conference on Computer, Communications and
Information Technology (CCIT 2014)
[19]. Bhutada.S, Balaram, .V.S.S.S. ; Bulusu, V.V. Latent
Dirichlet allocation Based multilevel classification
Control,
Instrumentation,
Communication
and
Computational Technologies (ICCICCT), 2014
International Conference
[20]. Guolong Liu, Xiaofei Xu ; Ying Zhu ; Li Li An
Improved Latent Dirichlet Allocation Model for Hot
Topic Extraction Big Data and Cloud Computing
(BdCloud), 2014 IEEE Fourth International
Conference
[21]. ShrijinaSreenivasan
and
B.Lakshmipathi
An
Unsupervised Model to detect Web Spam based on
Qualified Link Analysis and Language Models
International Journal of Computer Applications (0975
8887) Volume 63 No.4, February 2013
[22]. Xiao Xia, Xiaodong Wang, and Xingming Zhou
Evolving Recommender System for Mobile Apps: A
Diversity Measurement Approach Smart Computing
Review,vol.3,no.3,June213.

_______________________________________________________________________________________
Volume: 05 Issue: 03 | Mar-2016, Available @ http://www.ijret.org

478