Reviews on swarm intelligence algorithms for text document clustering

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 04 | Apr 2022 www.irjet.net p-ISSN: 2395-0072
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 682
Reviews on swarm intelligence algorithms for text document
clustering
S DHANALAKSHMI1, S SATHIYABAMA2
1,2 Department of Computer Science
1,2 Thiruvalluvar Government Arts College, Rasipuram, Tamilnadu, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Text clustering is an unsupervised learning
technique that divides a large numberoftextdocumentsintoa
small number of clusters. Each cluster contains similar
documents, while the clusters contain dissimilar text
documents. Various optimization problems, including text
document clustering challenges, have been effectively solved
using swarm intelligence (SI) optimization algorithms. This
paper reviews all of the relevant literature on SI-based text
document clustering applications including many variants,
basic, enhanced, and hybrid methods. The main procedure of
text clustering, distance and similarity function, and
theoretical discussion are also discussed.
Key Words: text mining, text clustering, swarm intelligence,
optimization algorithms, data mining
1. INTRODUCTION
Clustering is a general text mining technique for
representing a dataset using a limited set of clusters,
sometimes with a fixed number of clusters, based on
similarities between its elements [1, 2]. The partitioning
clustering technique is widely applied to solve real-world
applications including data clustering, image clustering,
marketing, and bio-informatics. The goal oftextclustering is
to create optimal clusters that contain related documents.
Clustering is the process of dividing a large number of
documents into a set of related groups.
Each group contains many similar objects, but different
groups contain different objects [3-5]. Theoverall technique
for the text clustering problem as an optimization problem,
its formulation, mathematical notations, preprocessing
stages, document format, clustering problem solution
representation, and the fitness function are all described in
this part. This can aid future studies in obtainingclear broad
information about the issue [6, 7].
Extracting meaningful data from documents is a difficult
operation that necessitates the use of rapid and high-quality
document clustering techniques. TheK-meansalgorithmisa
straightforward, quick, and unsupervised partitioning
algorithm that produces results that are both parallelized
and comprehensible [8, 9]. However,ithasmanydrawbacks
like local optima, low accuracy, and failure to achieve global
solutions. Hence, the swarm intelligence (SI) optimization
algorithm can overcome the limitations of the k-means
algorithm and obtain a global optimum value.
SI is the collective behavior of self-organized and
decentralized systems, which includes both intelligent and
non-intelligent individuals who follow simple rules or
behaviors to do very complicated tasks with limited local
information. Observing natural or artificial behaviors such
as bird flocks, fish schools, and ant food foraging led to the
development of SI algorithms. Particle swarm optimization
(PSO), bat optimization (BA), grey wolf optimization(GWO),
firefly optimization (FFA), ant colony optimization (ACO),
artificial fish swarm algorithm (AFSA), and artificial bee
colony optimization (ABC) are examples of SI algorithms
[10].
By comparing the many SI optimization algorithms
accessible for text document clustering,thisresearchaimsto
give the reader an accurate overview of the numerous SI
optimizationalgorithmsavailable.Thiswork investigates the
accessibility of each class and the implementation of a
suitable optimization algorithm for each. As a result, this
research will help academics and clinicians choose methods
and algorithms that are appropriate for a wide range of text
clustering applications. As a result of these studies, the goal
of this study was to investigate the field of SI clustering
methods and achieve the following goals:
 The present paper discusses the pre-processingstepsof
text clustering
 Various related workshavebeendiscussedusingswarm
intelligence algorithms.
 To provide a comprehensive classification of clustering
evaluation criteria that can be used in experimental
research.
 Conduct a theoretical examination of each class's best
representative SI optimization techniques.
The following is a list of the main sections of this study.
In Section 2, the main procedures of the text clustering are
presented. The variants of SI algorithms that have been
employed to solve text clustering problems are shown in
Section 3. Section 4 discusses the evaluationcriteria utilized
in text clustering applications. Section 5 contains a
discussion and theoretical analysis. Finally, Section 6
presents the survey's findings as well as future research
possibilities.

2. PROBLEME FORMULATION
A collection of documents D is groupedintoa predefined
number of K clusters. D can be demonstratedasanobject's
vectors  
1 2
, ,..., n
D d d d
 . Here, n is the total number of
documents. Each group has a cluster centroid, which is
denoted by k
c and is represented as a vector of word term
weights. k
c presents the th
K cluster center. The similarity of
documents is determined using distance measures, which
are used to cluster each object to the cluster centroid that is
closest to it. The fundamental goal of clustering is to create
groups based on the intrinsic contents of the objects. The
text requires model pre-processing actions for constructing
clusters in this era. Tokenizing, stop word removing,
stemming, item weighting and document representationare
pre-processing steps in text clustering. The pre-processing
procedures are briefly discussed in the next sub-section.
2.1 Tokenizing
Tokenization is the method of breaking words into bits
(or tokens) that are probably missing individual letters at
the same time, such as punctuation. These tokens are
usually linked to words or items, but it's important to
distinguish between types/tokens. A token is a sequence of
letters that is structured as a functional semantic unit in a
text. A sort is a collection of all tokens that have the same
letter chain. A word is an example of a term found in the
search method's vocabulary.
2.2 Stop words removal
Stop words include terms like “which”, “the”,”our”, “is”,
"an,", “me”, "that," and others, as well as important phrases
in the text that are extremely often used and small beneficial
words. These terms should be removed from the text since
they are frequently repeated, reducing the effectiveness of
clustering approaches. The total number of terms on the
stop-word list is above 500.
2.3 Stemming
Stemming is the procedure of shortening current terms
to their root/stem. The stem technique differs from theroot
morphological method in that it uses the same stem to
outline words, although it is not a true root.
2.4 Term weighting
The frequency of each phrase in the document is used to
provide a term weighting to each term or feature. In
weightingprocedures,thetermfrequency-inversedocument
frequency (TF-IDF) is commonly used. The following
equation shows how each document is represented as a
vector of term weights.
,1 ,2 ,
( , ,..., )
i i i i t
d w w w
 (1)
The term weight in the document i for the feature j is
calculated as follows,
, ( , ) ( , ) ( , ) log( )
( ))
i j
n
w tf i j idf i j ft i j
df j
    (2)
Where, ,
i j
w -denotes the weight of document i and term
j . ( , )
tf i j -denotes the occurrence of term j in document
i . ( , )
idf i j - denotes the inverse document frequency.
n represents the number of all documents in the datasets.
( ))
df j -represents the number of documents that contain
features j . In most text mining applications, the vector
space model (VSM) is utilized to describe documentfeatures
as a vector (row) of weights. The following equation is the
VSM's most frequent format, whichconsistsof n documents
with t words in each.
(3)
Figure 1 : Overviews of text document clustering
 

 
 
 
 
 
 
 

  

  

  
   
VSM
n t
n n n t
n n n t
t t
,( 1)
,1 ,2 ,
( 1),1 ( 1),2 ( 1),
1,1 1,2 1,( 1) 1,

3. RELATED WORKS FOR TEXT CLUSTERING
PROBLEME
Text document clustering is a method of separating
documents into parent groups,may managedocumentswith
unlabeled or unauthorized clusters, which is unusual. The
determination of the similarity measure between the
documents in each cluster has an impact on the text
clustering process. Text clustering is a technique for
improving the consistency of text document organization
and providing a user-friendly view of documents based on
an evaluation criterion that can be represented as an
objective function. Text clustering-based methods, in
particular, have aided users in dealing with a huge number
of documents and organizing them withbetterprecisionand
simplicity than if they did not utilize a text clustering-based
model. Because of its minimal computational cost, the
partitional clustering approach is now well-suited for
clustering a huge collection of document data sets [50].
Furthermore, becausemostpartitional clusteringtechniques
are essentially linear in time, they are commonly used.
Figure 1 shows the overviews of text document clustering
process.
The PSO is a well-known SI method that is applied to
solve various real-world applications including text
document clustering [11]. For example, Karol et al. (2013)
[12] developed a new method which is the use of a soft
computing strategy as an intelligent hybrid approach. The
hybrid approach combined with PSO and K-Means and FCM
(fuzzy c-means). Cagnina, L et al. (2014) [13] developed a
new efficient text clustering method based on PSO
(CLUDIPSO) for short text news. With small corpora,
CLUDIPSO produced high-quality results, but with bigger
corpora, performance deteriorated significantly. Song et al.
(2015) developed a new hybrid text clustering algorithm
based on Quantum-behaved PSO (QPSO) and genetic
algorithm (GA) [14]. T[he created methodisbeing evaluated
to normalize the search range of particles in a sound
environment usinga newpositionupdatingstrategy.Suchan
approach can increase the likpoelihood of finding the best
option. Pamba et al. (2017) [15] developed a document ring
method using a frequent pattern growth strategy based on
fuzzy PSO (FPFPSO). The goal is to assess the proposed
methodology for clustering online documents with varying
degrees of membership in all of their respective clusters.
The approach reduces search space by preserving
relationships with the search context and automatically
generates cluster centroids and particles.
Chouhan et al. (2018) [16] introduces a method for
document clustering basedonthePSOandk-means.The PSO
method is used before K-means to locate the best points in
the search space, and these points are utilized as initial
cluster centroids in the K-means to find final document
clusters. Janani et al. (2019) [17] introduce a novel spectral
clustering algorithm with PSO (SCPSO) to improve text
document clustering. The randomization is done using the
initial population using global and local optimization
functions. To deal with the massive amount of text
documents, the goal is to combine spectral clustering with
swarm optimization. The final findings reveal that the
suggested SCPSO algorithm outperforms other clustering
algorithms in terms of clustering accuracy. Pamba et al.
(2020) [18] developed a new document clustering method
usingSelf-adaptiveFrequent PatternGrowth-BasedDynamic
Fuzzy PSO (FPDFPSO) for Web Document Clustering. The
FPDFPSO outperforms existing conventional approaches in
terms of convergence speed and diversity preservation.
Alam et al. ( 2016) [19], a web clustering technique, is
presented to cluster web content effectively. Both K-means
and the Artificial Bee Colony (ABC)clusteringalgorithms are
used in the suggested technique. The ABC method is utilized
as the global search optimizerinthispaper,whileK-meansis
used to refine the solutions. As a result, the cluster's quality
has improved. Hijazi et al. (2021) [20] introduces a feature
selection method based on the combination of chi-square
and an artificial bee colony (ABC). The resultsshowedthata
smaller number of features surpassed the original features
set in terms of classification accuracy. Furthermore, when
compared to the chi-square approachandtheABCalgorithm
as a feature selection method, the proposed method
performed better. Mohammed et a. (2015) developeda new
text clustering method based weight-based fireflyalgorithm
(WFA) [21]. This finding method achieved better clustering
by search solution's exploitationisimproved. Mohammedet
al. (2016) [22] introduced an improvedsearchmethodofthe
ABC and embed two local search paradigmsnamelygradient
and chaotic local search to improve its exploitation
capability.
The SSO algorithm is based on a simulation of social-
spider cooperative behaviour. Individuals in the proposed
algorithm imitate a group of spiders that interact with one
another depending on the cooperative colony's biological
laws [23]. Chandran et al. (2017) [24] developed a novel
text clustering algorithm based on SSO. Chandran, T.R. et al.
(2018) developed a new document clustering method based
on [25] SSO. The use of a single cluster implementation for
each spider is defined in this study as a new and more
effective implementation of SSO (ESSOSC) for solving the
text document clustering problem. Rashaideh et al. (2020)
developed a new text clustering algorithm-based grey wolf
optimizer (GWO) [26]. The GWO is a successful andeffective
swarm-based algorithm that was recently presented to
simulate the hunting behaviour of grey wolves. GWO has
enhanced the allocation of documents to the bestclusters,as
well as assisted in the transition from local optimum
solutions to the best global solutions.
Krill herd (KH) is a revolutionary swarm-based
optimization system that mimics the behaviour of a KH in
real-time. This algorithm has a lot of potential because it
outperforms other optimization methods and balances the
exploration and exploitation process by combining the
strengths of local nearby searching and global wide-range
searching. The KH has been applied to solve the document
clustering problem. For example, Abualigah et al. (2016)
developed a new innovative text clusteringmethodbased on
the KH to improve the clusteringofwebtextdocuments[27].
The basic KH algorithm with all of its operatorsisusedinthe

first technique, while the genetic operators in the basic KH
algorithm are ignored in a second way. The proposed KH
algorithms' performanceisevaluatedandcomparedtothek-
mean algorithm.
Abualigah, L.M et al. (2018) [28] developed a new hybrid
text clustering method based new objective functionand KH
algorithm. In comparison to the other comparative
algorithms, the suggested modified KH with hybrid function
acquired almost all of the best outcomes for all datasets.
More, P.S et al. (2021) [29] developed a new text clustering
method based on the KH. Pre-processing is used to remove
noise and artifacts from the data, which is done using the
stemming and stop word removal methods. Furthermore,
the term frequency-inverse document frequency (TF–IDF)
method is used to extract notable features. TF–IDF features
are used to cluster documents, while the KH approach is
used to cluster text documents. The fitness function was
created from scratch, taking into account mean squareerror
and Jaccard similarity.
Feature selection is a key unsupervised learningstrategy
for selecting a fresh subset of useful text characteristics to
improve text clustering performance and reduce
computational time. To improve the performance and
computing efficiency of text clustering algorithms, a variety
of feature selection methods are used. For example,
Abualigah et al. (2017) [30] developed a new feature
selection method using a combination of PSO and genetic
algorithm for enhancing the performance of text clustering
problems. Asif, M et al. (2021) [31] introduces a self-inertia
weight adaptive PSO (SIW-APSO) based feature selection
approach to improve the performance of text classification
systems. Because of its high search competency and ability
to efficiently locate feature subsets, SIW-APSO has a fast
convergence phenomenon.
Singh et al. (2021) [32] presents a new hybrid feature
selection technique by utilizing the combination of ACO and
GA. The feature selection method has been improved,
resulting in a reduction in the dimensionality of the feature
space and, as a result, an increase in performance. Janani et
al. (2020) [33] introduceda newautomatic textclassification
approach based on machine learning. There are two stages
to the field study. The relevant features for classificationare
chosen in the first phase, and textdocumentsareclassifiedin
the second phase.
4. EVALUATION METRICS
The text clustering technique offers two types of
evaluation metrics: internal and exterior metrics. Details of
these measures are discussed in the following sections,
4.1 Internal metrics
The underlying base for developing the clustering
optimization algorithms processes is similarityanddistance
metrics. When it comes to qualitative results, the distance
measure is preferred when dealing with qualitative data.
Internal metrics are employed to determine a group of text
documents to be delivered to their typical cluster without
regard to the text class mark.
4.1.1 Distance measures
Clustering necessitates the construction of a distance
measure, which lends a numerical value to the extent of
difference between two documents and is used by the
clustering algorithm to divide a dataset into groups.Thereis
no single distance measure that works well in all situations,
therefore choosing one is a matter of determining which
distance measure best captures the essence of essential
distinguishing traits for the particulardocumentset. Table1
shows some mostly used distance measures in literature.
4.1.2 Similarity Measures
In-text mining, determining the degree of similarity
between two documents is a difficult task. Mostly used
similarity measures are reviewed in collected articles and
demonstrated in Table 2 [34, 35].
4.2 External measures
External measurementsare usedtoexaminetheaccuracy
(correctness) of the gathered clusters based on the class
labels in the dataset for the current article. The external
evaluation measures used to examine the output of the
clustering algorithms are described in the subsections
below. Accuracy, purity, entropy, precision, recall, and F-
measures are the most commonlyused evaluationmetricsin
the text clustering domain.
4.2.1 Accuracy
To determine the proper documents assigned to all
categories in the given dataset, accuracy is used. This
accuracy metric is calculated as follows:
,
1
1 k
i i
i
AC n
n 
  (4)
Where, ,
i i
n is the total number of correct candidates for
class i in cluster. n is the total number of given documents,
and K is the total number of given clusters in the dataset.
4.2.2 Purity
In a large class, the purity is used to compute each
cluster's percentage. This metric assigns each group to the
most common classification. Because of the percentage of
large class sizes in each group, which is compared according
to its size, an acceptable purity value is close to 1. As a
result, the purity value in the 1
,1
k
 
 
 
interval is used to
derive the purity value of the cluster j :
,
1
( ) max
j i j
j
j
P c n
n
 (5)
Where, max j - denotes the size of the large class in
group j . ij
n - denotes the number of all exact of the class
label i in cluster j . ij
n - denotes the complete number of
members of cluster j .

The purity for all clusters is determined as follows,
1
( )
k
j
j
j
n
p p c
n

  (6)
4.2.3 Entropy
In each group, the entropy measures the partitioning of
class labels. This metric focusesonhow well differentcluster
classes are contained. A good sample has zero entropy,
indicating a low entropy state for an optimal document
clustering solution. The following equation can be used to
calculate the entropy rate of a cluster j based on the size of
each group.
, ,
( ) log
j i j i j
i
E c p p
  (7)
Where ij
p is the probability value of classi members
that belong to group j . The entropy for all groups is
determined as follows,
1
( )
K j
j
j
n
E E c
n

  (8)
4.2.4 Precision
For each group, the precision is determined based onthe
specified class label in the main datasets. In all groups, the
precision test is the ratio of documents to the total number
of documents. The precision is determined for class i and
class j as follows,
( , ) ij
j
n
p i j
n
 (9)
Where, ij
n -denotes thenumber ofcorrectoftheclass i in
the j group. j
n -denotes the total number of objects in the
j group.
4.2.5 Recall
Based on the assigned class label, the recall (R) value for
each cluster is calculated. The recall value is calculated by
dividing the total number of relevant items in the dataset by
the number of key documents in each category. The
following equation is used to calculate the recall value for
class i and cluster j .
( , )
ij
i
n
p i j
n
 (10)
Where, ij
n -denotes numberofcorrectoftheclass i inthe
j group. j
n -denotes total number ofobjectsinthe j group.
4.2.6 F-Measures
The F-measures aim to evaluate clusters of the analyzed
partition clusters at the class label partitionclusterswiththe
largest match. Based on the aggregation of precision and
recall, this metric is a prominent evaluation criterion in the
cluster domain. The following equation is calculated for f-
measures:
2 ( , ) ( , )
( )
( , ) ( , )
P i j R i j
F j
P i j R i j
 


(11)
Table 2: Similarity measures
Name Formula
Cosine similarity
Jaccard similarity
Sorensen similarity
Dice similarity
Table 1: Distance measurements
Name Formula
Minkowski
distance
Standardized
Euclidean distance
Cosine distance
Pearson
correlation
distance
Mahalanobis
distance
.
cos( , )
A B
A B
A B


( , )
A B
J A B
A B



( , ) 2
A B
Dice A B
A B

 

2
( , )
A B
S A B
A B
 


2
1/2
1
)
d li lj
l
x x
sl

 

 

 
 

1 cos
T
i j
i j
x x
x x

  
( , )
1
( ) ( )
i j
i j
cov x x
D x D x
 
1
( ) ( )
T
i j i j
x x S x x

  
 
1/
1
)
n
d n
li lj
l
x x

 


f-measure for all clusters is determined as follows,
(12)
5. THEORETICAL DISCUSSIONS
Clustering analysis is a crucial technique for analyzing
unlabeled data by creating a collection of clusters basedona
predetermined number of clusters. Organizing pre-
processing and algorithm advancement, as well as solution
efficacy and evaluation, are all steps in the clustering
process. Each is inextricably linked to the others and
employs severe challenges in scientific research. SI
algorithms developed by diverse research communities aim
to tackle various clustering processes and have advantages
and disadvantages. Because there are so many intrinsic
conceivable conditions, there are still a lot of unsolved
issues. These issues have already attractedandwill continue
to attract extensive applications from a wide range of fields.
The current review and survey are organized by a list of key
concerns and possible research topics for optimization
cluster algorithms.
 New mechanisms have resulted in more sophisticated
and difficult tasks, necessitating the use of more reliable
clustering methods.
 No generally applicable clustering algorithm can
address all clustering difficulties.
 Local optima, on the other hand, can get imprisoneddue
to their emphasis on exploration (i.e., global search)
rather than exploitation (i.e., local search). This issue
may improve over time if the sets of rules that govern
the operation of various search algorithms become
better understood. The initial cluster centroids and the
number of clusters are two major issues in the text
clustering application.
 The values and settings of the parameters affect the
algorithm's overall performance, therefore parameter
adjustment will be important in future investigations.
 At the post-processing and pre-processing stages,
feature selection, extraction, and cluster validation are
as important as the clustering techniques. The difficulty
of succeeding schemes can be greatly reduced by
selecting relevant and necessary qualities, and result
evaluations reflect the level of trust we can place in the
produced clusters. Regrettably, neither strategy has
widespread leadership.
 Finally, the applications themselves are still responsible
for balancing numerous criteria and processes.
6. CONCLUSIONS
This survey report examined over 25 research papers to
determine the resilience and weaknesses of SI optimization
REFERENCES
[1] K. Tamilarisi, M. Gogulkumar, and K. Velusamy,
"Data clustering using bacterial colonyoptimization
with particle swarm optimization," in 2021 Fourth
International Conference on Electrical, Computer
and Communication Technologies (ICECCT), 2021:
IEEE, pp. 1-5.
[2] K. Tamilarasi, M. Gogulkumar, and K. Velusamy,
"Enhancing the performance of social spider
optimization with neighbourhood attraction
algorithm," in Journal of Physics: ConferenceSeries,
2021, vol. 1767, no. 1: IOP Publishing, p. 012017.
[3] D. Mustafi, A. Mustafi, and G. Sahoo, "A novel
approach to text clustering using genetic algorithm
based on the nearest neighbour heuristic,"
International Journal of Computers and
Applications, vol. 44, no. 3, pp. 291-303, 2022.
[4] K. Velusamy and R. Manavalan, "Performance
analysis of unsupervised classification based on
optimization," International Journal of Computer
Applications, vol. 975, p. 8887, 2012.
[5] S. S. Babu and K. Jayasudha, "A survey of nature-
inspired algorithm for partitional data clustering,"
in Journal of Physics: Conference Series, 2020, vol.
1706, no. 1: IOP Publishing, p. 012163.
[6] A. A. Jalal and B. H. Ali, "Text documents clustering
using data miningtechniques,"International Journal
of Electrical & Computer Engineering (2088-8708),
vol. 11, no. 1, 2021.
[7] S. Dhanalakshmi and S. Sathiyabama, "Survey on
Information Mining Procedures Utilized in
Healthcare Services."
[8] T.-H. Jo, "Inverted index based modified version of
k-means algorithm for text clustering," Journal of
Information Processing Systems, vol. 4, no. 2, pp.
67-76, 2008.

 
F
K
F
j
K
j
1
( , )
R i j is recall value of class i and group j . Therefore,the
Where, ( , )
P i j is precision value ofclass i andgroup j . techniques. This document comprehensively analyses the
whole literature till the year 2021. The majority of the
papers in this collection discuss optimization techniques
based on SI that have been applied in text clustering
(document) applications. Several algorithm variants are
investigated, including usual, fundamental, enhanced, and
hybrid. Furthermore, various new SI optimization
techniques that can be used to solve clustering problems
have recently been presented. To solve text clustering
challenges, new hybrid and modified methods might be
presented. Furthermore, various new meta-heuristic
optimization techniques that can be used to solve clustering
problems have recently been presented.

[9] K. Vijayakumari and V. Baby Deepa,"FuzzyC-Means
Hybrid with Fuzzy Bacterial Colony Optimization,"
in Advances in Electrical and Computer
Technologies: Springer, 2021, pp. 75-87.
[10] S. Selvaraj and E. Choi, "Swarm Intelligence
Algorithms in Text Document Clustering with
Various Benchmarks," Sensors, vol. 21, no. 9, p.
3196, 2021.
[11] X. Cui, T. E. Potok, and P. Palathingal, "Document
clustering using particle swarm optimization," in
Proceedings 2005 IEEE Swarm Intelligence
Symposium, 2005. SIS 2005., 2005: IEEE, pp. 185-
191.
[12] S. Karol and V. Mangat, "Evaluation of text
document clustering approach based on particle
swarm optimization," Open Computer Science, vol.
3, no. 2, pp. 69-90, 2013.
[13] L. Cagnina, M. Errecalde, D. Ingaramo, and P. Rosso,
"An efficient particle swarm optimization approach
to cluster short texts," Information Sciences, vol.
265, pp. 36-49, 2014.
[14] W. Song, Y. Qiao, S. C. Park, and X. Qian, "A hybrid
evolutionary computation approach with its
application for optimizing text document
clustering," Expert Systems with Applications, vol.
42, no. 5, pp. 2517-2524, 2015.
[15] R. V. Pamba, E. Sherly, and K. Mohan, "Evaluationof
Frequent Pattern Growth Based Fuzzy Particle
Swarm Optimization Approach for Web Document
Clustering," Cham, 2017: Springer International
Publishing, in Computational Science and Its
Applications – ICCSA 2017, pp. 372-384.
[16] R. Chouhan and A. Purohit, "An approach for
document clustering using PSO and K-means
algorithm," in 2018 2nd International Conference
on Inventive Systems and Control (ICISC), 2018:
IEEE, pp. 1380-1384.
[17] R. Janani and S. Vijayarani, "Text document
clustering using spectral clustering algorithm with
particle swarm optimization," Expert Systems with
Applications, vol. 134, pp. 192-200, 2019.
[18] R. V. Pamba, E. Sherly, and K. Mohan, "Self-adaptive
Frequent Pattern Growth-Based Dynamic Fuzzy
Particle Swarm Optimization for Web Document
Clustering,"inComputational Intelligence:Theories,
Applications and Future Directions-Volume II:
Springer, 2019, pp. 15-25.
[19] M. Alam and S. Baulkani, "A hybrid approach for
web document clustering using K-means and
artificial bee colony algorithm," Int J Intell Eng Syst,
vol. 9, no. 4, pp. 11-20, 2016.
[20] M. Hijazi, A. Zeki, and A. Ismail, "Arabic Text
Classification Using Hybrid Feature Selection
Method Using Chi-Square Binary Artificial Bee
Colony Algorithm," Computer Science, vol. 16,no.1,
pp. 213-228, 2021.
[21] A. J. Mohammed, Y. Yusof, and H. Husni, "Document
clustering based on firefly algorithm," Journal of
Computer Science, vol. 11, no. 3, pp. 453-465, 2015.
[22] A. J. Mohammed, Y. Yusof, and H. Husni, "GF-CLUST:
A nature-inspired algorithm for automatic text
clustering," Journal of Information and
Communication Technology, vol. 15, no. 1, pp. 57-
81, 2016.
[23] E. Cuevas, M. Cienfuegos, D. Zaldívar, and M. Pérez-
Cisneros, "A swarmoptimizationalgorithminspired
in the behavior of the social-spider,"ExpertSystems
with Applications, vol. 40, no. 16, pp. 6374-6384,
2013.
[24] T. R. Chandran, A. Reddy, and B. Janet, "Text
clustering quality improvement using a hybrid
social spider optimization," International Journal of
Applied Engineering Research, vol. 12, no. 6, pp.
995-1008, 2017.
[25] T. R. Chandran, A. Reddy, and B. Janet, "An effective
implementation of social spider optimization for
text document clustering using single cluster
approach,"in2018SecondInternational Conference
on Inventive Communication and Computational
Technologies (ICICCT), 2018: IEEE, pp. 508-511.
[26] H. Rashaideh et al., "A grey wolf optimizer for text
document clustering," Journal of Intelligent
Systems, vol. 29, no. 1, pp. 814-830, 2020.
[27] L. M. Abualigah, A. T. Khader, M. A. Al-Betar, and M.
A. Awadallah, "A krill herd algorithm for efficient
text documents clustering," in 2016 IEEE
symposium on computer applications & industrial
electronics (ISCAIE), 2016: IEEE, pp. 67-72.
[28] L. M. Abualigah, A. T. Khader, and E. S. Hanandeh,
"Hybrid clustering analysis using improved krill
herd algorithm," AppliedIntelligence,vol.48,no.11,
pp. 4047-4071, 2018.
[29] P. S. More, B. S. Saini, and K. S. Bhatia, "Krill Herd
(KH) algorithm for text document clustering using
TF–IDF features," in Smart Computing: CRC Press,
2021, pp. 502-512.

[30] L. M. Abualigah and A.T.Khader,"Unsupervisedtext
feature selection techniquebasedonhybridparticle
swarm optimization algorithm with genetic
operators for the text clustering," The Journal of
Supercomputing, vol. 73, no. 11, pp. 4773-4795,
2017/11/012017,doi:10.1007/s11227-017-2046-
2.
[31] M. Asif, A. A. Nagra, M. B. Ahmad, and K. Masood,
"Feature Selection Empowered by Self-Inertia
Weight Adaptive Particle Swarm Optimization for
Text Classification," Applied Artificial Intelligence,
pp. 1-16, 2021.
[32] A. Singh and A. Kumar, "Text document
classification using a hybrid approach of ACOGAfor
feature selection," International Journal of
Advanced Intelligence Paradigms, vol. 20, no. 1-2,
pp. 158-170, 2021.
[33] R. Janani and S. Vijayarani, "Automatic text
classification using machine learning and
optimization algorithms," Soft Computing, vol. 25,
no. 2, pp. 1129-1145, 2021.
[34] B. Diallo, J. Hu, T. Li, G. A. Khan, and A. S. Hussein,
"Multi-view document clustering based on
geometrical similarity measurement," International
Journal of Machine Learning and Cybernetics,pp.1-
13, 2021.
[35] Y.-S. Lin, J.-Y. Jiang, and S.-J. Lee, "A similarity
measure for text classification and clustering,"IEEE
transactions on knowledge and data engineering,
vol. 26, no. 7, pp. 1575-1590, 2013.

Reviews on swarm intelligence algorithms for text document clustering

More Related Content

Similar to Reviews on swarm intelligence algorithms for text document clustering (20)

More from IRJET Journal (20)

Recently uploaded (20)

Reviews on swarm intelligence algorithms for text document clustering