Comparison of Compression Algorithms in text data for Data Mining

International Journal of Advanced Engineering, Management and Science (IJAEMS) [Vol-6, Issue-6, Jun-2020]
https://dx.doi.org/10.22161/ijaems.66.2 ISSN: 2454-1311
www.ijaems.com Page | 231
Comparison of Compression Algorithms in text
data for Data Mining
Roya Mahmoudi*
, Mansoureh Zare
Department of Computer and Information Technology Engineering, Sabzevar branch, Islamic Azad University, Sabzevar, Iran.
*Corresponding Author
Abstract— Text data compression is a process of reducing text size by encrypting it. In our daily lives, we
sometimes come to the point where the physical limitations of a text or data sent to work with it need to be
addressed more quickly. Ways to create text compression should be adaptive and challenging.
Compressing text data without losing the original context is important because it significantly reduces
storage and communication costs. In this study, the focus is on different techniques of encoding text files
and comparing them in data processing. Some efficient memory encryption programs are analyzed and
executed in this work. These include: Shannon Fano coding, Hoffman coding, Hoffman comparative
coding, LZW coding and Run-length coding. These analyzes show how these coding techniques work, how
much compression for these coding techniques. Writing, the amount of memory required for each
technique, a comparison between these techniques is possible to find out which technique is better in which
conditions. Experiments have shown that Hoffman's comparative coding shows a higher compression ratio.
In addition, the improved RLE encoding suggests a higher performance than the typical example in
compressing data in text.
Keywords— data compression, data mining, encryption, text data.
I. INTRODUCTION
In computer science and information theory, using
encryption techniques encrypts and compresses data using
fewer bits or symbols than the original representation. Data
compression is useful because it helps reduce the
consumption of expensive resources, such as hard disk
space or transmission bandwidth [3]. In the present
situation where we are faced with the problem of data
abundance and data disruption, data mining and data
refinement and compression are used for this purpose.
However, the problem is that the pressure of large
compression may cause data loss [2]. That data is one of
the valuable assets of any organization that the loss of
some information can lead to major problems in the
organization's decisions and strategies. The purpose here is
to present and implement some data compression
algorithms efficiently and compare them to the methods
used for data compression in data mining. Calculating
some of the important compression factors for each of
these algorithms, comparing different encryption
techniques and improving the performance of different data
compression techniques, and selecting a suitable
encryption method for data mining is the textual data
considered.
Data mining, which has become very popular in recent
years, refers to the use of data analysis tools to discover
valid patterns and relationships that have been unknown
until now [13]. These tools may include statistical models,
mathematical algorithms, and learning methods (Machine
Learning Methods) that are automated based on experience
gained through neural networks or decision-making trees.
Improves. Data mining is not limited to data collection and
management, but also involves the analysis of data and the
refinement and compression of data. Applications that
explore data by examining text or multimedia files for data
warehouses that have become popular [10]. This type of
warehouse provides a variety of data types from text,
image to storage and processing for use in organizational
strategies and better understanding of data as information.
Their advantage over databases is the ability to refine data
and compress data to preserve important data and delete
trivial data [5]. Each database has its query and query
language. A multi-year problem in data analysis is
increasing the size of the data set. This process
demonstrates the need for more efficient compression
programs and also for performing analytical operations that
work directly on compressed data. Efficient compression
schemes can be designed based on the misuse of patterns

and intrinsic structures in the data [9]. Data rotation is one
of these features that can significantly enhance
compression.
Text compression is a process to reduce text size by
encoding more information. It uses this number of bits and
bytes to store data which is somewhat reduced [1]. It
requires a hassle-free technique that will not lose any data
when compressing a text file. This approach returns the
text file to its original state [4]. Text compression and
decompression techniques are intended for natural
languages such as large data sets of high redundancy
English and other data with a similar sequential structure
such as the source code of a program. However, this
method can be used for any data type to achieve some
compression [11]. The importance of text compression
involves reducing storage hardware, data transfer time, and
bandwidth, which is the use of file Compression can lead
to a significant reduction in the cost of a drive or solid-state
drive and improve network bandwidth, cost savings [7].
Textual data of various sizes are used in English to apply
compression and decompression techniques. Text
compression reduces data storage space. By reducing or
eliminating redundancies, data compression is done. The
encryption technique uses fewer bits than the original
version and eliminates unnecessary information in this
case. This is a harmless way in which the number of bits
and bytes is cut off to store information.
Previous investigations
There are various methods for compressing data that take
about forty years to develop. Shannon Fano and Hoffman
developed and developed compression algorithms in the
late 40s [10]. In 1949, Shannon and Fano devised a method
for compression using a systematic method of assigning
code words based on the probability of blocks called
Shannon Fano coding. Another method of compressing
data was found in 1951 by Huffman, known as the
Huffman algorithm. Since then, the Huffman algorithm has
been used for compression.
It has been shown in a review [13], that text data
compression using the Shannon-Fano algorithm performs
the same function as the Huffman algorithm, when all
characters are repeated in a string, and when the expression
is short and only one character in Text repeats, has the
same functionality. When input data that have long text
and data text has a more combinatorial character in the
string or word, the Huffman algorithm has a greater impact
on compression.
In another review [2], we tried compression ratio,
compression time and decompression time for RLE
encryption algorithms, Huffman encryption algorithm,
Shannon Fan algorithm, Adaptive Hoffman encryption
algorithm, Arithmetic encryption algorithm, and Lempel
Zev Wel encryption. Were compared using random text
files. The results show that as the text file size increases,
the compression time increases. Medium-value encryption
was less time consuming than the other algorithms for both
Huffman and Shannon-Fano encryption approaches. In the
LZW algorithm, it only worked fine for small files. The
compression time of the Huffman algorithm was
comparable to the encoding size at the highest time. The
decompression time of all algorithms was less than
500,000 milliseconds except for the Adaptive Huffman and
LZW algorithms. The compression ratio was similar for
small size files except for RLE encoding, but LZW
encoding worked best.
In another study [9], a comparison was made between
RLE, Huffman, LZW computational encoding, first LZW
then Hoffman and finally RLE, on the random doc, txt,
BMP, tiff, gif and jpg files. , Which studies showed that
LZW and Hoffman give almost the same results when used
for compressing text files.
In another study [4], different methods of data compression
algorithms such as LZW, Hoffman, fixed-length code
(FLC) and Hoffman after using fixed-length code (HFLC)
on English text files in terms of size compression, The
ratio, time (speed) and entropy were examined. The best
algorithms on all LZW encryption compression scales,
then Hoffman, Hoffman, after using a fixed-length code
(HFLC) and a fixed-length code (FLC), with entropy were
4.719, 4.855, 5.014 and 6.889, respectively.
In a similar study [5] that analyzed Hoffmann's algorithm
and compared it with other common compression
techniques such as Arithmetic, LZW, and RLE based on
their use in different programs and its benefits. This is very
efficient coding, and in RLE coding, when the sequence of
pixels with fewer bits is more frequent, it significantly
reduces the file size. The LZW algorithm is mostly used in
TIFF, GIF and text files, which is a fast, harmless
algorithm that is very easy to use, while the Hoffmann
algorithm is used in JPEG compression, which produces
optimal and low-volume code but is relatively slow.
In another paper [7], Shannon Fano coding, Hoffman
coding, Adaptive Huffman coding, RLE, arithmetic
coding, LZ77, LZ78, and LZW were tested using the
Calgary body. In statistical compression techniques,
accounting coding has been more effective than other
methods. In another entropy review, the English text file
for coding Shannon Fano, Hoffmann coding, Run-length
(RLE), Lempel-Ziv-Welch (LZW) coding was calculated.
The compression ratio for encoding Shannon Fano and

Hoffman was almost the same, and the two algorithms can
save 54.7% of space. The compression ratio of Lempel-
Ziv-Welch algorithms is low compared to the Hoffmann
and Shannon Fano algorithms, and it has been concluded
that the Hoffmann encryption algorithm is the best result
for text files.
In another study [12], data was first compressed by each
code based on the length of the run, such as Golomb code,
FDR code, EFDR, MFDR, SAFDR, and OLEL encryption,
and then another compressed data with Hoffman code was
compressed. The double compression using the Hoffman
code was 50.8% compression ratio. Better results were
obtained for the data set with incremental data.
In a study [1] on execution time, compression ratio and
compression efficiency in a distributed customer-server
environment using four compression algorithms: Hoffmann
algorithm, Shannon Fano algorithm, LZW algorithm, and
Run-length encryption algorithm Was analyzed. A
customer's data is distributed across multiple
processors/servers and subsequently compressed by servers
in remote locations and sent to the customer. The Sigrid
framework was used, and the results showed that the LZ
algorithm achieves better efficiency/scalability and
compression ratio, but is slower than other algorithms.
Hoffman coding, LZW coding, LZW-based Hoffman
coding were compared for multiple and single
compression. This showed that Hoffman-based LZW
encryption can compress data more than any other three,
while in normal mode the LZW compression ratio is based
on Hoffmann 4.41, where the maximum compression ratio
is maximal in the case of compression. The LZW build is
4.17. Hoffman-based LZW compression is in some cases
better than LZW compression.
II. METHOD
As a test, five compression and stress relief techniques
were used: Shannon-Fano coding, Hoffman coding,
Hoffman comparative coding, and Lempel Ziv Welch
coding, and improved RLE and RLE coding. Different
sizes of text files have been used as data sets, and the
compression ratio has been measured to determine which
compression method is best.
III. RESULTS
The compression ratio depends on the size of the output
file. The more compressed the output file is, the lower the
compression ratio. When the compression ratio exceeds 1,
the output file size is larger than the original input file size.
Here we can see that in most modified RLE encrypted
input files the compression ratio is greater than 1, meaning
that these algorithms expand the original files rather than
compress them. RLE encoding works best for files with
sequential duplicates, but typically duplicate data files do
not. For this reason, the compression ratio is greater than 1.
The results also show that Shannon Fano encoding is better
for most inputs than Huffman encoding since for some
input files the symbol code length in Huffman encoding is
so large that it increases the output files over the input
files. Becomes original. It can be said that Repeated
Huffman encoding shows the best compression ratio for
most files.
If the code is shorter, the algorithms have less memory.
Now if we consider the average length of the code, we will
see that for Runlength encoding, we have no value for the
average code length because no encoding is created in Run
length. But for other techniques, Shannon Fano encodes
less code than Huffman coding and modified Run-length
coding. Here, too, Huffman coding shows the best average
code length for most files. Again, if the standard deviations
are smaller, the algorithms occupy less memory. Here, too,
for run-length coding, we have no value for standard
deviation since no coding is created in Run-length coding.
But for other techniques, Shannon Fano has less standard
deviation than Huffman coding and modified Run-length
coding. Again, Huffman coding provides the best standard
deviation for most files.
From the above analysis, we can say that Repeated
Huffman coding works best for most cases. It has a smaller
compression ratio, average code length, and standard
deviation. So, it seems the best algorithm among them.
This ultimately results in less memory than others for the
output file after the final lapse.
If extra runtime is provided for Hoffman's frequent
programming, then Shannon Fano's programming at lower
runtime would work quite well. RLE coding for files with
consecutive duplicates can be very effective. Modified
RLE coding can be very useful if the temporary file
intermediary created after applying the Huffman encoding
to the input file has 0 and 1 consecutive iterations. Finally,
each compression technique has its pros and cons. It
depends on the content of the input files on how much
better compression can be achieved. Table 1 shows the
four input files after compression and resizing algorithms.

Table 1- Comparison of compression algorithms
Modified
Runlength
Coding
Run length
Coding
Shannon
Fano Coding
Huffman
Coding
Characteristics
Input File
Size
(KB)
Input File
(.txt)
163 KB291 KB3.89 KB87.5 KBOutput File Size
1561
1.091.960.040.58Compression Ratio
9.83-6.3512.38Avg. Code Length
15.33-0.1912.33Standard Deviation
164 KB245 KB91.8 KB77.3 KBOutput File Size
1172
581 KB388 KB339 KB698 KBOutput File Size
4533
477 KB919 KB387 KB765 KBOutput File Size
4704
IV. CONCLUSION
Text compression is very important in data refinement in
data mining. Because data is an important part of data
mining textual data. Therefore, it requires a no-loss
technique so that no efficient information is lost when
compressing or compressing information. In the
investigations performed in this study on the performance
and computation and comparison of compression ratios,
mean code length and standard deviation for Shannon
Fano coding, Huffman coding, Huffman repeated coding,
Run-length coding, and modified Run-length coding, in
The issue of how to compress was examined by each one
so that, at present, the most effective algorithm can be
used based on the size of the input text file, the type of
content, available memory, and execution time to get the
best results. A new approach to data compression is called
"Run-length Run Algorithm", which offers much better
compression than the run-length algorithm.
REFERENCES
[1] P. Peter, S. Hoffmann, F. Nedwed, L. Hoeltgen, and J.
Weickert, “Evaluating the true potential of diffusion-based
inpainting in a compression context,” Signal Processing:
Image Commun., No. 46,40–53 (2016).
[2] M. N. Huda, “Study on Huffman Coding,” Graduate
Thesis, 2004.
[3] The Canterbury Corpus [Online].
http://corpus.canterbury.ac.nz/resources/cantrbry.zip.
Accessed 30 May (2018)
[4] Habib, A., Rahman, M.S.: Balancing decoding speed and
memory usage for Huffman codes using quaternary tree.
Appl. Inf. 4, 5 (2017). https://doi.org/10.1186/s40535-016-
0032-z
[5] S.R. Kodituwakku and U. S. Amarasinghe, “Comparison of
Lossless Data Compression Algorithms for Text Data,”
Indian Journal of Computer Science and Engineering, vol.
I(4), 2007, pp. 416-426.
[6] C. Lamorahan, B. Pinontoan and N. Nainggolan, “ Data
Compression Using Shannon-Fano Algorithm,” JdC, Vol.
2, No. 2, September, 2013, pp. 10-17.
[7] M. A. Khan, “Evaluation of Basic Data Compression
Algorithms in a Distributed Environment,” Journal of Basic
& Applied Sciences, Vol. 8, 2012, pp. 362-365.
[8] D. Belazzougui, S.J. Puglisi, Range predecessor and
Lempel–Ziv parsing, in: Proc. 27th Annual ACM–SIAM
Symposium on Discrete Algorithms (SODA), 2016, pp.
2053–2071.

[9] P. Bille, M.B. Ettienne, I.L. Gørtz, H.W. Vildhøj, Time-
space trade-offs for Lempel–Ziv compressed indexing,
Theoret. Comput. Sci. 713 (2018) 66–77.
[10] D. Kempa, D. Kosolobov, LZ-End parsing in compressed
space, in: Proc. 27th Data Compression Conference (DCC),
2017, pp. 350–359, Extended version in
https://arxiv.org/pdf/1611.01769.pdf.
[11] G. Navarro, A self-index on Block Trees, in: Proc. 24th
International Symposium on String Processing and
Information Retrieval (SPIRE), in: LNCS, vol. 10508,
2017, pp. 278–289.
[12] T. Nishimoto, T.I.S. Inenaga, H. Bannai, M. Takeda,
Dynamic index and LZ factorization in compressed space,
in: Proc. Prague Stringology Conference (PSC), 2016, pp.
158–170. [44] T. Nishimoto, T.I.S. Inenaga
[13] D. Belazzougui, S.J. Puglisi, Y. Tabei, Access, rank, select
in grammar-compressed strings, in: Proc. 23rd Annual
European Symposium on Algorithms (ESA), in: LNCS,
vol. 9294, 2015, pp. 142–154.
[14] D. Kempa, N. Prezza, At the roots of dictionary
compression: string attractors, in: Proc. 50th Annual ACM
SIGACT Symposium on Theory of Computing (STOC),
2018, pp. 827–840.

Comparison of Compression Algorithms in text data for Data Mining

More Related Content

What's hot

Similar to Comparison of Compression Algorithms in text data for Data Mining

Recently uploaded

Comparison of Compression Algorithms in text data for Data Mining