International Journal of Computer Applications (0975 – 8887)
Volume 160 – No 7, February 2017
11
Machine Learning: A Review on Binary Classification
Roshan Kumari
M.Tech. Scholar
Department Of Computer
Science & Engineering
ABES Engineering College,
Ghaziabad
Saurabh Kr. Srivastava
Sr. Asst. Professor
Department Of Computer
Science & Engineering
ABES Engineering College,
Ghaziabad
ABSTRACT
In the field of information extraction and retrieval, binary
classification is the process of classifying given
document/account on the basis of predefined classes.
Sockpuppet detection is based on binary, in which given
accounts are detected either sockpuppet or non-sockpuppet.
Sockpuppets has become significant issues, in which one can
have fake identity for some specific purpose or malicious use.
Text categorization is also performed with binary
classification. This research synthesizes binary classification
in which various approaches for binary classification are
discussed.
Keywords
Sockpuppets, Non sockpuppets, multiple identity deception,
text categorization, NB, SVM, Random Forest, Ensemble
methods and Binary Classification
1. INTRODUCTION
Sockpuppets are some fake IDs or accounts, which are created
for some specific malicious use. So sockpuppet detection is
based on binary classification, in which classification is done
on each account to assign a class i.e., sockpuppet or non-
sockpuppet. Anyone can have new account with the help of
less information. So it's necessary to have some method to
find out these sockpuppet cases or suspicious cases because
its violates privacy. Wikipedia does not provide any specific
facility to detect such malicious accounts. So the current
process are done manually which is time consuming and cost
effective. So identify such accounts as sockpuppet in
Wikipedia, is significant issue. Multiple identity is also an
example of binary classification, in which one person has
more than one account for malicious use. With the help of
multiple account, one can try to alter senders contents for
some specific purpose. So multiple identity deception is also a
big issues on social media. Text categorization is also done by
performing binary classification. Text categorization plays an
important role on information retrieval for classification of
different documents.
Sockpuppet detection becomes a significant problem in
social media environment. Thamar Solorio et.al.[1] has
contributed his work towards sockpuppet detection. They
have done their work with small case study of automated
detection of sockpuppets based on Authorship Attributes
where the task consists of analyzing a written document to
predict the true author. Some features of authorship attribution
are collected and examined. Each user comment is a
"document". There are two steps taken for classification
process, in initial step, predictions from the classifier on each
comment has taken. Then in second step, predictions for each
comments and combine them in majority voting schema to
assign final decisions to each account. Michail Tsikerdekis
et.al. [2] has proposed a novel approach for use of non- verbal
behavior to detect multiple account identity deception on
social media. New accounts initiated by blocked users are
called sockpuppetry. In social media, identity deception is a
major issue. Nonverbal communication(user activity or
movement) are more powerful than Verbal
communication(speech or text). Identity deception focuses on
manipulating the senders information and is divided in three
categories-identity concealment, identity theft and identity
forgery. Major issue with identity deception in social media is
the presence of multiple identities by one user. They have
used Logs of blocked users on Wikipedia during the period
since February 2004 until October 2013 as dataset and used
SVM, Random forest and Adaptive Boosting(ADA) method
for classification of sockpuppetry or not. They found that
Adaptive Boosting provides the best balance between Recall
and Precision, and achieved highest Accuracy among all used
classification techniques.
2. REVIEWED PAPERS IN THIS
DIRECTION
Thamar Solorio et.al.[3] has described a corpus of sockpuppet
cases from Wikipedia. A Corpus provides a real world dataset
of short messages from malicious users. Sockpuppet
investigations in Wikipedia(SPI) are identified using support
vector machine. Author has tried to detect SPI, to decide
whether to mentioned the editors as belonging to the same
person or not on the basis of binary classification. These
results are based on observations from comments made by
each user. Used complete list of features can be found at the
following link:http://docsig.cis.uab.edu/media/2014/03/list-of-
features.pdf. Xueling Zheng et.al.[4] has proposed algorithm
for detecting sockpuppet pair in one forum and two different
forum. Two different online accounts but belong to the same
person are referred as sockpuppets pairs. In this paper there
are two methods proposed for detecting sockpuppets. The first
one is designed for detecting those sock puppets pairs in the
same discussion forum while the second one is for detecting
sockpuppets pairs that appear in two different forum. Authors
has used dataset from Uwants and HK discuss during the
period of March 2010 to May 2010 . On the basis of Detection
Score, they have tried to find out similar keywords used by
different people. Sadia Afroz et.al.[5] Author has proposed a
method to detect stylistic deception in written document. It is
mentioned in this paper that with the help of large feature set,
it is possible to distinguish regular documents from deceptive
documents. To detect adversarial writing, it is necessary to
identify a set of discriminating features that distinguish
deceptive writing from regular writing. After determining
these features supervised learning techniques are used to
classify new writing styles. Three feature sets are used: Write
print feature set, Lying-detection feature set and 9-feature
set(Authorship attribution features). SVM and DT techniques
are used for analysis. SVM classifier works best with the
write prints feature and DT performed well with the Lying
International Journal of Computer Applications (0975 – 8887)
Volume 160 – No 7, February 2017
12
detection features. Dhanyasree P et.al.[6] has contributed their
work for detection of identity deception on social networking
sites. On social networking sites, one person creates multiple
account for malicious use. So this become a very big issue on
social sites. So on the basis of verbal and non verbal behavior
it can be detect such types of account. So authors has tried to
detect such accounts on the basis of verbal and non verbal
behavior. They have used algorithms, Calculation of non-
verbal variables and model testing using Random Forest
method and Identification of time window using PSO. They
found that Detecting multiple accounts through nonverbal
behavior has more accuracy. The automated system to detect
multiple accounts gives good performance . Both the verbal
and nonverbal behavior can be combined and used for
sockpuppets detection, in which binary classification are done
to detect sockpuppet or non sockpuppet cases. M BalaaNand
et.al. [7] has proposed a method to detect multiple account
and fake identity on social media like WIKIPEDIA using non-
verbal behavior(User activity and User Movement). Authors
has worked for time independent based non verbal behavior.
In which they has used data from Wikipedia and SVM,RF and
ADA techniques for binary classification. They found
Adaptive Boosting gives the best balance between recall and
precision with high accuracy. Sheetal Antony et.al.[8] has
proposed a system that can use verbal and non verbal
behavioral patterns to detect identity deception. There is an
Admin who manages each account for users. The details and
activities of the user are analyzed and detect if there is some
deception. The details are verified in database. If it detects
that there is some deception then there are some security
questions that are asked to users. Zaher Yamak et.al.[9] has
proposed a detection method in which following steps are
taken: first of all data are crawled from Wikipedia, then detect
sockpuppet accounts, after that create a set of non-verbal
behavior features and then calculate the values of the
proposed features and finally used machine learning algorithm
for classification. SVM, RF, Naive Bayes, K nearest neighbor,
Bayesian Network and Adaptive Boosting are taken for result
comparison. Best accuracy given by Random Forest(99.8%)
and Bayesian Network(99.6%) for sockpuppet detection.
Malware detection is an important issues to save our computer
system and communication infrastructure. So, Anti-virus
technology is a key player in tackling malware files, based on
two methods: signature based and heuristic-based method.
Asaf Shabtai et.al.[10] has addressed different challenges i.e.,
files representation method, feature selection method and
classification algorithm. Some additional issues are also
mentioned in this paper such as: weighting
algorithm(ensembles),imbalance problem ,active learning and
chronological evaluation. Authors has proposed a framework
for detecting new malicious code in executable files can be
designed to achieve very high accuracy while maintaining low
false positives. Antu Mary et.al.[11] has proposed a method
for detecting identity deception by a single user is based on
using Nonverbal behavior. Non verbal behavior explains
activities done by each user separately such as Some
Wikipedia users create multiple accounts and use them for
various malicious purposes such as Number of articles
generates, Number of searches done for same articles,
Number of bytes added and also removed, Number of times
same spelling mistakes carryout constantly, Time taken
between each revision, creating fraudulent articles, damaging
existing article text etc. So these deceptions cannot easily
detected by any authority. Numerous methods have been
proposed that can help in detecting multiple accounts owned
by the same persons. Using verbal and nonverbal behavior of
user can easily detect the sockpuppet with limited amount of
time. Ashkan Sami et.al.[12] has provide a framework for
analyzing and classifying PE files based on data mining
techniques. Windows Application programming
interface(API) can be used to extract knowledge describing
behavior of executables .Each API call is used as a feature.
FISHER SCORE based feature selection process is used. Top
4 categories by Fisher's Score are :File Management, Process
and Thread, Console and Registry. 34820 PE files where
31,869 were malicious and 2951 were benign windows PE
files. RF,NB and DT techniques are used. Random Forest
gives good performance. G.Ganesh Sundarkumar et.al. [13]
has done text mining for feature selection. Then Mutual
Information is used to extract most influential features. Then
data mining models such as Decision tree, Neural network
model, SVM , Probabilistic neural network and group method
of data handling(GMDH) is used. On the basis of Accuracy,
Sensitivity. Specificity all 59 models are compared. DT,
SVM, PNN, NN and GMDH techniques are used for
comparison. Then again the dataset are balanced using
Oversampling and again tested the model .After balancing
sensitivity/accuracy improved. Prasha Shrestha et.al.[14] has
explained Malware Family Identification process using string
information. Classification of malware into correct family is
an important task for antivirus vendor. Using term-frequency
and inverse document frequency(tf-idf) and using prominent
strings extraction classification work are done in this paper.
To check accuracy-way vendor agreement are compared with
accuracy achieved by used algorithm or techniques. Exact
match: Global vocabulary, exact matches: Prominent strings,
Prominent strings set and Absence of prominent string are
techniques used for this purpose. Data are used from
University's malware database(1504 malware files). On the
basis of above mentioned experiments it can be easy to detect
malware family files. Exact Match: Global vocabulary gives
the best result. Michael Bailey et.al.[15] has explained that
anti-virus is incomplete in that it fails to detect or provide
labels of the malware samples. Authors explained that when
these systems do provide labels, theses labels do not have
consistent meaning across families and variants within a
single naming convention as well as across multiple vendors.
Finally they demonstrated that these system lack conciseness
in that they provide some little information or sometime too
much information about a specific piece of malware. Authors
has proposed a novel technique to overcome these problems.
On the basis of behavioral fingerprints of malware's activity,
automated malware classification are done. To compare and
combine these fingerprints, single-linkage hierarchical
clustering approach are applied. Gaston L’Huillier et.al.[16]
has explained phishing mail classification. Phishing email
fraud is to attempt to gain personal/sensitive information such
as username, passwords and credit cards details. Algorithm
like Support vector machines, naïve Bayes, Random forest
algorithm are used for classification of phishing emails. The
classification of phishing emails is extension of text mining.
In this paper feature extraction methodology for fishing
emails are enhances by using latent semantic analysis features
and keyword extraction techniques. SVMs ,the naïve Bayes
model and the logistic regression method are used in Weka
tool to improve accuracy. Rafiqul Islam et.al.[17] has tried to
classify malware on the use of static and dynamic features.
There are some drawback in static techniques for malware
classification. So it focuses to detect some dynamic features
which is very useful in classification process. For static
features there are two information needed: function length
frequency and printable strings information. For dynamic
features API functions name are used. SVM,DT,RF and Naive
Bayes techniques are used in WEKA tools with 10 fold cross
International Journal of Computer Applications (0975 – 8887)
Volume 160 – No 7, February 2017
13
validation for classification. Random forest gives the highest
accuracy with TP,FN and Accuracy parameters. . Ali Danesh
et.al. [18] proposed a classifier fusion method to improve text
classification. Proposed approach combined Naive Bayes, K-
NN and Rocchio methods by Voting algorithms methods and
achieve a better classification rate which experimental results
shows that the classification error decreases by 15%. 2000
documents from 20 different newsgroups has taken for
experiment. Aytug Onan et.al.[19] proposed ensemble
approach such as Adaboost, Bagging, Dagging, Random
Subspaces and majority Voting. Two way ANOVA test
conducted. The experimental analysis shows that the bagging
ensemble of Random Forest with the most frequent based
keyword extraction method yields promising results for text
classification. The experimental result shows that the
utilization of keyword based representation of text documents
in conjunction with ensemble learning can enhance the
predictive performance and scalability of text classification
schemes. Baoxun Xu et.al.[20] has proposed an improved
Random forest classifier for text categorization. They
proposed improved random forest methods with both feature
weighting and tree selection methods(WTRF), Breiman's
Random forest (BRF) and the random forest with only tree
selection method(TRF).Comparisons are based on accuracy
and F-measures. All these three improved random forest
methods are compared with other widely used text
categorization methods i.e., support vector
machines(SVM),Naive Bayesian(NB),and K-
NearestNeighbor(KNN). M.Sivakumar et.al. [21] proposed a
hybrid text classification Approach using KNN and SVM.
They proposed SVM-KNN approach aims to reduce the
impact of parameters in classification accuracy. The
performance analysis shows the accuracy of SVM-KNN
method remains optimal for even huge values of the
parameters. The accuracy compared to the KNN method is
higher in the SVM-KNN. Unlike the conventional KNN
classification approach, the SVM-KNN approaches has low
impact on the implementation of the parameters. Sundus
Hassan et.al. [22] proposed a method for text categorization in
which they compared Support Vector Machine(SVM) and
Naive Bayes (NB) classifiers. Baseline for the experiment has
setup by removing stopwords and stemmed the dataset by
using Porter Stemmer. They used micro-average and macro
average F-Measure. Experiments shows the improvement in
micro average and macro average F-measure in both method
i.e., SVM and NB.
Table 1: Summary of related work on sockpuppets detection, multiple identity deception detection and text categorization by
performing binary classification
Citation
Dataset used
Classifiers
Measures
Results
Thamar Solorio
et.al.[2013]
Data collected from
Wikipedia
Support Vector
Machine(SVM)
Precision, Recall ,F-
Measure and accuracy
On the basis of authorship
attributes sockpuppet cases
are detected.
Michail
Tsikerdekis
et.al[2014].
Logs of blocked users on
Wikipedia during the period
since February 2004 until
October 2013
SVM, Random forest and
Adaptive Boosting(ADA)
Precision, Recall,
Accuracy, F-Measures, false
positive rate and Matthews
Correlation Coefficient
(MCC)
Higher accuracy
Rate.
Thamar Solorio
et.al[2014]
Wikipedia
Support Vector Machine in
Weka tools
Precision, Recall ,F-
Measure and Accuracy
F-Measure gives the best
result.
Xueling Zheng
et.al[2011]
Uwants and HK discuss
during the period of March
2010 to May 2010
Detection score
On the basis of similarity
and keywords
Sockpuppet pair are detected.
Sadia Afroz
et.al[2011].
(1)Extended -brennan-
Greenstadt corpus
(2)Hemingway-Faulkner
Imitation corpus
(3)Thomas-Amina Hoax
corpus
SVM and DT
Precision, Recall and F-
Measure
SVM classifier works best
with the Write prints feature
and DT performed well with
the Lying detection features.
M BalaaNand.
et.al[2015]
Dataset used from
Wikipedia
SVM, RF and ADA.
Precision, Recall, F-
Measure, Accuracy, MCC
and False Positive rate.
Adaptive Boosting gives the
best balance between recall
and precision with high
accuracy
Sheetal Antony.
et.al[2016]
Not Used
Not Used
Not Used
On the basis of verbal and
non verbal behavioral
patterns, detected identity
deception.
Zaher Yamak
et.al[2016]
Dataset used from
Wikipedia from Feb2004 to
April2015
SVM, RF, Naive Baiyes, K
nearest neighbor, Bayesian
Network and Adaptive
Boosting
TPR, FPR, F-Measure,
Precision and MCC
Best accuracy given by
Random Forest(99.8%) and
Bayesian Network(99.6%).
Asaf Shabtai.
et.al[2009]
Not Used
ANN,DT,KNN,BN,SVM
,OneR,Boosted Algorithm
TPR, FP and, Accuracy
This paper includes aspects of
different challenges for
classifying new malicious
code based on static features
extracted.
International Journal of Computer Applications (0975 – 8887)
Volume 160 – No 7, February 2017
14
Antu Mary
et.al[2015]
Wikipedia
SVM
Recall, Precision and F-
Measure
Identity deception detection is
more accurate using non
verbal behavior in
comparison to verbal
behavior.
Ashkan Sami
et.al[2010]
34820 PE files where
31,869 were malicious and
2951 were benign windows
PE files.
RF, NB and DT
Accuracy, Precision, Recall
and F Arate(False alarm
rate)
Random Forest gives good
performance.
Table 1: Continued...
Citation
Dataset used
Classifiers
Measures
Results
G.Ganesh
Sundarkumar et
al.[2013]
Dataset from CSMINING
group is used
DT,SVM,PNN,NN and
GMDH
Accuracy ,Sensitivity and
Specificity
On the basis of Accuracy
,Sensitivity and Specificity
all 5 models are compared.
Then again the dataset are
balanced using Oversampling
and again testing the model.
After balancing
sensitivity/accuracy
improved.
Prasha Shrestha
et al.[2014]
Data are used from
University's malware
database(1504 malware
files).
(1)Exact match: Global
vocabulary
(2) exact matches
(3) Prominent strings set
(4)Absence of prominent
string
Accuracy and correlation
between n-way vendor
agreement with accuracy
Exact match: Global
vocabulary gives the best
accuracy around 91.02%.
Michael Bailey
et al.[2007]
Data collected from data
sources .There are 3 types of
dataset used: Legacy, small
and large.
Hierarchical clustering
algorithm
Consistency, Completeness
and conciseness
It's easy to classification of
malware on the basis of
behavior of fingerprints.
Gaston
L’Huillier et
al.[2013]
Not Used
SVMs ,the naïve Bayes
model and the logistic
regression method were
used in Weka tool
F-Measure
Accuracy
improved
Rafiqul Islam et
al.[2013]
Data are used from
antivirus vendors time
periods 2003-2007 and
2009-2010.
SVM,DT,RF and Naive
Bayes in WEKA tool.
TP,FN and Accuracy
Random forest gives the
highest accuracy.
Ali Danesh et
al.[2007]
100 articles are taken from
20 different newsgroup
NB, KNN, Rocchio ,Voting
and OWA AND DT
Accuracy Rate
Fusion of classifiers gives
better results in comparison to
base classifiers.
Aytug Onan et
al.[2016]
Reuters-21578 dataset used
NB, SVM, LR, RF and
ensemble methods of SVM
and RF.
F-Measure, Accuracy and
AUC Values
This paper represents analysis
of 5 statistical keywords
methods for text
classification.
Baoxun Xu et
al.[2012]
20 different Usenet
newsgroups and contains
18772 documents divided
into 20 different classes
SVM,KNN,NB,BRF,TRF
and WTRF
Micro F-Measures and
Macro F-Measures
It has been observed that
proposed method WTRF
method outperforms among
all other text categorization
methods
M.Sivakumar et
al.[2014]
Reuters-21578 R8 dataset
used.
KNN and SVM
Accuracy
Proposed SVM-KNN method
provides high accuracy.
Sundus Hassan et
al.[2000]
Dataset from 20 Newsgroup
with 1000 documents
NB and SVM
Macro F-Measures and
Micro F-Measures
NB gives better performance
over SVM.
International Journal of Computer Applications (0975 – 8887)
Volume 160 – No 7, February 2017
15
3. CONCLUSION AND FUTURE WORK
This paper presented a taxonomy for binary classification.
Sockpuppet detection is based on binary classification, in
which two classes are predefined i.e. sockpuppet or non-
sockpuppet. And datasets are classified on the basis of
predefined classes. Multiple identity deception is also based
on binary classification in which classification process are
done on given datasets into two groups i.e., sockpuppet and
non-sockpuppet. Text categorization is also done by involving
binary classification. So binary classification implies an
important role in machine learning process. To get better
result, analyze different features sets for binary classification.
With different feature sets, better results can be observed in
terms of precision, recall, F-Measure and accuracy. Different
datasets can be used for experiment with different text
features. These feature sets can be used for multilevel
classification and multiclass classification.
4. REFERENCES
[1] Thamar Solorio, Ragib Hasan and Mainul Mizan, "A Case
Study of Sockpuppet Detection in Wikipedia",
Proceedings of the Workshop on Language in Social
Media(LASM 2013),Pages 59-68,Atlanta,Georgia,June
13 2013.@2013 Association for Computational
Linguistics.
[2] Michail Tsikerdekis and Sherali Zeadally, "Multiple
Account Identity Deception Detection in Social Media
Using Non Verbal Behavior", IEEE Transactions on
Information Forensics and Security, Vol 9, No 8, August
2014.
[3] Thamar Solorio, Ragib Hasan and Mainul Mizan,
"Sockpuppet Detection in Wikipedia :A Corpus of Real-
World Deceptive Writing For Linking Writing",
arXiv:1310.6772v1[cs.CL] 24 Oct 2013.
[4] Xueling Zheng, Yiu Ming Lai, K.P. Chow, Lucas C.K.
Hui and S.M. Yiu, "Detection of Sockpuppets in Online
Discussion Forums", HKU CS Tech Report TR-2011-03.
[5] Sadia Afroz, Michael Brennan and Rachel Greenstadt,
"Detecting Hoaxes Frauds and Deception in Writing
Style Online". 2011.
[6] Dhanyasree P*, Sajitha Krishnan and Ambikadevi Amma
T, "Deception Detection in Social Media through
Combined Verbal and Non-Verbal Behavior ",
International Journal of Advanced Research in Computer
Science and Software Engineering , Volume 5, Issue 4,
2015.
[7] M Balaanand,R Soumipriya,S Sivaranjani and S Sankari,
"Identifying Fake Users in Social Networks Using Non-
Verbal Behaviour". International Journal of Technology
and Engineering System (IJTES)Vol 7. No.2 2015 Pp.
157-161©gopalax Journals, Singapore.
[8] Sheetal Antony, Prof. B. S. Umashankar, "Identity
Deception Detection and Security in Social Medium,
IJCSMC, Vol. 5, Issue 4, April 2016, pg.499-502.
[9] Zaher Yamak, Julien Saunier and Laurent Vercouter, "
Detection of Multiple Identity Manipulation in
Collaborative Projects", IW3C2, WWW'16 Companion,
April 11-15, 2016, Montreal, Quebec, Canada. ACM
978-1-4503-4144-8/04.
[10] Asaf Shabtai, Robert Moskovitch, Yuval Elovici and
Chanan Glezer, " Detection of malicious code by
applying machine learning classifiers on static features:
A state -of-the-art-survey ", INFORMATION
SECURITY TECHNICAL REPORT 14 (2009) 16-29,
ELSEVIER.
[11] Antu Mary Kuruvilla1 and Saira Varghese2, "A Survey
on detecting Identity Deception in Social Media
Applications", International Journal of Modern Trends in
Engineering and Research (IJMTER) Volume 02, Issue
04, [April – 2015] ISSN (Online):2349–9745 ; ISSN
(Print):2393-8161.
[12] Ashkan Sami, B. Yadegari, N. Peiravian, and S. Hashemi
and A. Hamze, "Malware detection based on mining API
calls", SAC '10: Proceedings of the ACM Symposium on
Applied Computing, pp. 1020-1025, 2010.
[13] G.Ganesh Sundarkumar and Vadlamani Ravi, "Malware
Detection by Text and Data Mining".IEEE2013..
[14] Prasha Shrestha,Suraj Maharajan,Gabriela Ramirez de la
Rosa,Alan Sprague,Thamar Solorio and Gracy Warner,
"Using String Information for Malware Family
Identification" @Springer International Publishing
Switzerland 2014,A.L.C.Bazzan and
K.Pichara(Eds.):IBERAMIA 2014,LNAI 8864,pp.686-
697,2014.DOI:10.1007/978-3-319-12027-0_55
[15] Michael Bailey, Jon Oberheide, Z. Morley Mao, Farnam
Jahanian and Jose Nazario, " Automated Classification
and Analysis of Internet Malware". April 26 2007
[16] Gaston L’Huillier, Alejandro Hevia, Richard Weber and
Sebastian Rios, "Latent Semantic Analysis and Keyword
Extraction for Phishing Classification".IEEE2010.
[17] Rafiqul Islam, Ronghua Tian , Lynn M. Batten and
Steve Versteeg," Classification of malware based on
integrated static and dynamic features". Journal of
Network and Computer Applications 36 (2013) 646–656.
ELSEVIER.
[18] Ali Danesh, Behzad Moshiri and Omid Fatemi, "Improve
Text Classification Accuracy based on Classifier Fusion
Methods".2007 IEEE Xplore.
[19] Aytuğ Onana, Serdar Korukoğlub and Hasan Bulutb, "
Ensemble of keyword extraction methods and classifiers
in text classification". A. Onan et al. / Expert Systems
With Applications 57 (2016) 232–247.
[20] Baoxun Xu, Xiufeng Guo, Yumming Ye and Jiefeng
Cheng, "An Improved Random Forest Classifier for Text
Categorization", [JOURNAL OF COMPUTERS] VOL.
7, NO. 12, DECEMBER 2012.
[21] M. Sivakumar, C. Karthika and P. Renuga, "A Hybrid
Text Classification Approach using KNN and SVM",
[IJIRSET] Volume 3, Special Issue 3, March 2014.
[22] Sundus Hassan, Muhammad Rafi and Muhammad
Shahid Shaikh, "Comparing SVM and Naive Classifiers
for Text categorization with Wikitology as knowledge
enrichment". IEEE Xplore 2012.
IJCA
TM
: www.ijcaonline.org