You are on page 1of 7

Proceedings of the Fourth International Conference on Smart Systems and Inventive Technology (ICSSIT-2022)

IEEE Xplore Part Number: CFP22P17-ART; ISBN: 978-1-6654-0118-0

HATE SPEECH & OFFENSIVE LANGUAGE DETECTION USING ML &NLP

Geetha Harshini Panchala V V S Sasank


Student, Department of CSE, Koneru Lakshmaiah Assistant Professor, Department of CSE, Koneru
Education Foundation, Vaddeswaram, AP, India, 522502. Lakshmaiah Education Foundation, Vaddeswaram, AP,
India, 522502.

Dory Ratna Harshitha Adidela Pachipala Yellamma


Student, Department of CSE, Koneru Lakshmaiah Associate Professor, Department of CSE, Koneru
Education Foundation, Vaddeswaram, AP, India, 522502. Lakshmaiah Education Foundation, Vaddeswaram, AP,
India, 522502
2022 4th International Conference on Smart Systems and Inventive Technology (ICSSIT) | 978-1-6654-0118-0/22/$31.00 ©2022 IEEE | DOI: 10.1109/ICSSIT53264.2022.9716417

Dr. K Ashesh
Associate Professor, Department of CSE, Koneru Chitturi Prasad
Lakshmaiah Education Foundation, Vaddeswaram, AP, Student, Department of CSE, Koneru Lakshmaiah
India, 522502. Education Foundation, Vaddeswaram, AP, India, 522502
Email:cprasad@kluniversity.in.

Abstract— To restore peace and harmony in this cross-cultural users from using it. Hate speech is a form of damaging online
Internet era, it is of utmost importance for every citizen to content that targets a group or an individual member based on
behave and spread brotherhood. Under the given circumstances their real or perceived characteristics of identification, such as
of 5G evolution citizens have taken their role onto the internet race, religion, or sexual orientation. With the increase of online
very seriously thereby most of the netizens spend their time hate speech, automated detection as a natural language
condemning, judging, and trolling other netizens, public figures processing job is gaining traction. However, it was only
for that matter. Because of the consequences in an unprejudiced recently discovered that current models do not generalize well
society involving race, gender, or religion, the challenge of to unknown data.
automatically detecting hate speech and objectionable language
in social media material is critical. However, existing research
in this field is mostly focused on several languages, which
limits its relevance to certain groups. The use of harsh language
on social media platforms, as well as the consequences that this
has, has become a serious problem in modern culture.
Automatic ways to recognize and deal with this sort of content
are necessary due to the large volume of content produced every
day. Machine Learning & Natural Language processing has
cutting-edge algorithms and classifiers that have benefitted
mankind in impossible ways. Hence, our effort in this project is
to make use of this impeccable technology to create an efficient
system that automatically detects hate speech and offensive Figure 1: No Swearing picture [Source: Franklin Law]
language from the Twitter dataset.

Keywords— Twitter Data, Hate, Speech, Language, Offensive, There is no overall lawful meaning of hate speech, and the idea
Machine Learning, Natural Language Processing, Classifiers, of what is thought of as "scornful" is begging to be proven
Naïve Bayes, Random Forest, English. wrong. Hate Speech is characterized in this record as any type
of correspondence, regardless of whether oral, composed, or
I. INTRODUCTION physical, that objectives or utilizations censorious or oppressive
language concerning an individual or a gathering dependent on
We are all aware that if social media platforms are not handled
what their identity is, like their religion, identity, ethnicity, race,
correctly, they may cause global turmoil. The use of hate speech
shading, plummet, sexual orientation, or another personality
and offensive language is one of the issues that these platforms
factor. This is as often as possible dependent on and makes
confront. The use of such language frequently leads to
bigotry and antagonism, and perhaps corrupting and
confrontations, crimes, and, in the worst-case scenario, riots. As
troublesome specifically circumstances.
humans are unable to monitor such vast amounts of data, we
may rely on AI to detect the usage of such language and restrict

978-1-6654-0118-0/22/$31.00 ©2022 IEEE 1262


Authorized licensed use limited to: Dayananda Sagar University. Downloaded on November 02,2023 at 03:48:09 UTC from IEEE Xplore. Restrictions apply.
Proceedings of the Fourth International Conference on Smart Systems and Inventive Technology (ICSSIT-2022)
IEEE Xplore Part Number: CFP22P17-ART; ISBN: 978-1-6654-0118-0

Abusive language includes profanity, racial, ethnic, sexist The goal of researching automatic hate speech & offensive
insults, or slurs based on color, religion, or national origin, and language identification on Twitter is to make it easier to reduce
includes harsh, violent, vulgar, or disparaging words that would the harm caused by online hate speech. Hate speech detection
diminish an individual's dignity. algorithms must be able to deal with hate speech's continual
development and change. Hence in our project, we have utilized
the ML algorithms & classifiers of SVM, RF, and multinomial
NB, XG Boost, and Logistic regression with the help of NLP
modules for Pre-Processing like Vectorization, Bag of Words.

Figure 2: NO Abuse Sign Board [Source: The couple


land times]
Hate Speech is a type of harming on the web content that Figure 3: Screenshot of abusive Tweets [Source: WIKI]
objectives a gathering or a singular part dependent on their
genuine or saw qualities of distinguishing proof, like race, Objectives
religion, or sexual direction. With the expansion of online Hate
Speech, robotized identification as a characteristic language Our objectives in this project are as follows-
preparing position is acquiring footing. Nonetheless, it was as  To understand the research carried out by various
of late found that current models don't sum up well to obscure other researchers in the same field and their
information. Even though many types of harmful and offending implementation methodologies.
language are firmly connected, there are vital qualifications to  To understand the pre-requisite study and to get hands-
be made. In the space of robotized discovery examination, on experience in executing ML & NLP classifiers.
hostile and harmful language are both utilized as classes for  To build an efficient system the identifies hate speech
harming the material. While the models incorporate terms like and abusive language so that it serves a noble purpose
"incredibly upsetting, discourteous," and "possible utilization
for mankind.
of irreverence," oppressive language includes a significant
degree of deliberateness. Therefore, hostile language has a
more extensive definition, and Hate Speech fits under both of
II. LITERATURE REVIEW
these classes. Most investigations on Hate Speech location include parallel
grouping errands and fine-grained arrangement of unmistakable
Hate Speech is unmistakable from different types of hostile types of Hate Speech, as indicated by connected exploration.
language on account of the depiction given previously. Hate Speech ID, notwithstanding these assignments, may
Individual attacks, for instance, are characterized by being foresee the substance depends on the measure of contempt. This
pointed against an individual and are not generally propelled by guarantees that main the individuals who are antagonistic
the objective's personality. Hate Speech is likewise particular should be advocated as taboo, instead of a severe split among
from cyberbullying, which is executed on weak casualties who disdain and non-Hate Speech. A few exploration centers around
can't guard themselves more than once and over the long run. the ternary arrangement of Hate Speech, hostile language, and
Even though exploration that incorporates both Hate Speech not one or the other, even though the sum isn't pretty much as
and other questionable words is tended to, this review focuses high as other location undertakings.
on Hate Speech and Hate Speech datasets.
Utilizing n-grams, skip-grams, and grouping-based word
As a result of the remarkable development in the utilization of portrayal as qualities, Malmasi and Zampieri et al have
the web by people, all things considered, and instructive separated commonly offensive language from Hate Speech in
foundations, the harmful internet-based substance has turned web-based media. When contrasted with the straight SVM, the
into a genuine worry in the present society. In mechanized RBF portion SVM was demonstrated to be better suitable for
recognizable proof of perilous text material, recognizing Hate information with fewer components.
Speech and hostile language is a significant issue. We propose As attributes, a few levels of surface n-grams, words and
in this review a strategy for consequently grouping messages on characters, and word skip grams are utilized. The creators
Twitter into two classes. We need to get incredible precision utilized a straight SVM classifier to attempt to separate between
while evaluating the model on test information in the wake of Hate Speech and hostile language utilizing an openly available
tweaking it to give the best outcomes. Davidson dataset. The review utilized element choice to

978-1-6654-0118-0/22/$31.00 ©2022 IEEE 1263


Authorized licensed use limited to: Dayananda Sagar University. Downloaded on November 02,2023 at 03:48:09 UTC from IEEE Xplore. Restrictions apply.
Proceedings of the Fourth International Conference on Smart Systems and Inventive Technology (ICSSIT-2022)
IEEE Xplore Part Number: CFP22P17-ART; ISBN: 978-1-6654-0118-0

decrease the number of qualities and tracked down that these


elements were incapable of segregating among scorn and Step 1: Tokenize a sentence first.
hostile language. Step 2: Tokenize all of the sentences.
Gaydhani et al. utilized three distinctive datasets to make a Step 3: Create a vocabulary and vectors.
more adjusted Hate Speech class and got a good outcome for
the LR classifier with L2 standardization. The creators utilized Vectorization is a jargon term for a traditional method of
TF-IDF esteems to gauge every one of the n-gram attributes transforming raw data (text) into vectors of real numbers, which
recovered from the tweets. The creator of H Watanabe et al. is the format that ML models allow. This method has been
proposed an even-minded method to separating unigram, around since the dawn of computing, has proven to be effective
opinion highlight, semantic element, and example include in in a variety of fields, and is currently being utilized in NLP.
their paper. Composed reliance attributes were proposed by Vectorization is a stage in feature extraction in Machine
Madukwe and Gao et al. for recognizing Hate Speech and Learning. By translating text to numerical vectors, the goal is to
questionable language. When utilizing inserted include extract some distinguishing characteristics from the text for the
determination, great precision might be achieved with a model to learn from.
diminished list of capabilities.
IV. RESEARCH METHODOLOGY
Here, we display the overview approach of our project in
III. CONCEPTUAL BACKGROUND systematic, decisive ways that portray the flow of the classifier
AI is a sort of information examination that robotizes the system
formation of logical models. It's a field of AI dependent on the
idea that PCs can gain from information, perceive examples,
and settle on decisions with next to zero human information. In
our undertaking, we will use the classifiers of SVM, Logistic
Regression, Naïve Bayes, Random Forest, and XG Boost
models. The procedure of speculating the class of given
information focuses is known as a grouping. Targets, marks,
and classifications are altogether terms used to depict classes.
The work of assessing a planning capacity (f) from input factors
(X) to discrete yield factors is known as grouping prescient
displaying (y).

There are a few order calculations accessible today, and it is Figure 5: Flowchart of the project overview approach
difficult to decide one is better than the others. It is subject to
the application and the sort of information assortment gave. Dataset Description-The dataset we shall be using in our
Order is a sort of regulated learning wherein the information is project has been obtained from the Kaggle website entitled
additionally provided to the targets. The grouping has a few Twitter Hate Speech. The author of this dataset is Rohit
uses in an assortment of fields, including credit endorsement, Agarwal who has uploaded it 3 years ago, which has become
clinical finding, and target showcasing. The most vital stage in very popular recently with three thousand plus downloads and
the wake of preparing the model is to assess the classifier to 12 unique contributors. The file download size is 5 MB with
guarantee its materialness. Accuracy alludes to the extent of public accessibility. It contains two CSV files with a test file
pertinent models found among the recovered occasions, while and a trained file. Each file has three columns of id, label, and
review alludes to the extent of significant cases found among tweet collected from Twitter just as shown in the below figure.
the general number of application examples. Accuracy and The dataset can be found in this URL link -
review are utilized to survey the significance of the information. https://www.kaggle.com/vkrahul/twitter-hate-speech

Natural Language Processing :


Human language is separated into pieces in regular language
handling with the goal that the linguistic design of sentences
and the significance of words might be analyzed and
appreciated in the setting. This empowers PCs to peruse and
grasp spoken and composed material in the very way that
individuals do. The Bow model is utilized in man-made
reasoning for PC vision, normal language preparing, Bayesian
spam channels, record arrangement, and data recovery. An
assemblage of messages, like a sentence or an archive, is
viewed as a pack of words in a Bow.
Figure 6: Screenshot of the CSV file compact list of the dataset
[ Source: Kaggle]

978-1-6654-0118-0/22/$31.00 ©2022 IEEE 1264


Authorized licensed use limited to: Dayananda Sagar University. Downloaded on November 02,2023 at 03:48:09 UTC from IEEE Xplore. Restrictions apply.
Proceedings of the Fourth International Conference on Smart Systems and Inventive Technology (ICSSIT-2022)
IEEE Xplore Part Number: CFP22P17-ART; ISBN: 978-1-6654-0118-0

Dataset Cleansing with NLP- In the data cleansing phase we


intend to purify and chastise the data drop or add a column or  As our next step is concerned, we intend to preprocess
two basing on our needs. We intend to modify the misspelled the tweets by using the stop word functionality and
words here, comprehend the input word text meaning, and if the adding the column to the in-process dataset file with
meaning contributes to hate and abusive language, we dedicate the “Processed Tweets” label where we also remove
it to the percentage count of hate speech. This shall be
certain special characters which don’t contribute
elaborated further in the latter sections. Here we have to
understand that Normal Tweets are labeled as Class 0 and hate directly to our project like the symbol hashtag,
tweets are labeled as Class 1, Post this re-labeling, we attempt punctuation marks emoticons, etc.
to import and generate a word cloud to the given dataset. We
import the word cloud function, Image Color Generator, and
Stop Words function too. We define the font size, color of the
background, plot size too with bilinear interpolation. Details of
what those functions give are detailed below-

 Stop Words- Any term in a stop list that is filtered out


before or after natural language data processing is
Figure 8: Screenshot of Dataset Header after Preprocessing
referred to as a stop word. For example, words like
"mcg," "dr.", and "patient" appear in nearly every Over Sampling & Classifier Build- The usage of social media
paper you come across in clinical books. As a result, is growing every day, and hate speech is growing in lockstep
these keywords might be used as stop words in clinical with the number of users. As a result, corporations face a
text mining and retrieval. Similarly, phrases like "#," difficult task in monitoring every user's tweet. As a result, we're
"RT," and "@username" might be considered stop working on a machine learning model to automatically identify
words in tweets. In any human language, there are hate speech tweets, which will save them a lot of time and
plenty of stop words. We eliminate the low-level money. Because the data set provided in this challenge contains
information from our text by eliminating these terms, just 25% of hate tweets, when we create a model using it, the
allowing us to focus more on the vital information. model performs better on regular tweets, which does not
address our problem. As a result, we must oversample the hate
tweets for the two classes to be equal for our model to operate
 Word cloud- is a cloud of words, as the name properly.
indicates. It is a text data visualization approach in
which each word is shown concerning its relevance in Arbitrary oversampling is the course of haphazardly
the context of frequency. This is an extremely useful picking and supplanting tests from the minority class and
tool for deciphering the gist of today's news or the adding them to the preparation dataset. Arbitrary
content of any channel. undersampling is the course of haphazardly picking and
eliminating occurrences from the greater part class from the
preparation dataset. As the given dataset is profoundly
imbalanced, we have decided to over-example it.

V. EMPIRICAL EVALUATION
The After the Data purifying stage, we would now be able to
continue with the test train split capacity to apply our classifiers
Figure 7: Word cloud of our dataset to the pre-handled datasets. Before this progression, we utilize
the vectorization capacity of NLP where it serves us. Word
 As we handle the distributions of the word statistics, vectorization is an NLP strategy for planning words or
we enabled a column with percentage for the count expressions from a dictionary to a coordinating with the vector
corresponding the Class 0 and 1 i.e., Normal & Hate of genuine numbers, which may then be utilized to decide word
Speech respectively whose details are thus obtained- expectations and semantics. Vectorization is the most common
way of transforming words into numbers.

Random Forest Classifier:


An arbitrary woods is a meta assessor that uses averaging to
improve forecast precision and command over-fitting by fitting
a few choice tree classifiers on different sub-examples of the
Table 1: Hate Vs Normal Tweet word count/ Percentage dataset. If bootstrap=True (default), the sub-example size is
managed by the maximum example's contention; in any case,

978-1-6654-0118-0/22/$31.00 ©2022 IEEE 1265


Authorized licensed use limited to: Dayananda Sagar University. Downloaded on November 02,2023 at 03:48:09 UTC from IEEE Xplore. Restrictions apply.
Proceedings of the Fourth International Conference on Smart Systems and Inventive Technology (ICSSIT-2022)
IEEE Xplore Part Number: CFP22P17-ART; ISBN: 978-1-6654-0118-0

the entire dataset is used to make each tree. We fixed the Logistic Regression is a Machine Learning strategy that is
assessors to a worth of 500 here, underneath is the order report utilized to address characterization issues. It is a prescient
of our dataset with RF application. insightful procedure that depends on the likelihood of thought.
The calculated relapse theory recommends that the expense
work be restricted to a worth somewhere in the range of 0 and
1.

Output 1- Classification Report of RF classifier to Twitter


Dataset

XG Boost Algorithm:
Output 4- Classification Report of logistic regression for our
The inclination helped trees method is carried out in XGBoost, Twitter dataset
a well-known and effective open-source execution. Inclination
boosting is an administered learning method that consolidates Multi NB Classifier:
the evaluations of an assortment of more modest, more fragile
The multinomial Naive Bayes classifier is useful for discrete
models to endeavor to precisely foresee an objective variable.
elements like word includes in-text arrangement. Number
element counts are needed for the multinomial dispersion.
Fragmentary counts like tf-idf, then again, may work practically
speaking.

Output 2- Classification Report of XGB Classifier to our


Twitter dataset

Linear SVM Model: Output 5- Classification Report of Multi NB classifier for our
Twitter dataset
Direct SVM is a classifier that is utilized for straightly
detachable information, which suggests that if a dataset can be Evaluation Metrics- The metrics that we have chosen for the
sorted into two classes utilizing a solitary straight line, it is evaluation of our model are accuracy score, classification
called straightly distinct information, and the classifier is called reports, and confusion matrix. Now we test the sample data. As
Linear SVM classifier. we have posted the classification reports of the 5 different
classifiers that we built in this project we can now understand
studying it.
 Accuracy Score- The number of correct predictions
divided by the total number of input samples is the
ratio. It only works when there is an equal number of
samples in each class.

 Classification Report- In machine learning, a


classification report is a performance evaluation
Output 3- Classification Report of Linear SVM classifier to our
Twitter dataset statistic. It's used to demonstrate your trained
classification model's accuracy, recall, F1 Score, and
Logistic Regression: support.

978-1-6654-0118-0/22/$31.00 ©2022 IEEE 1266


Authorized licensed use limited to: Dayananda Sagar University. Downloaded on November 02,2023 at 03:48:09 UTC from IEEE Xplore. Restrictions apply.
Proceedings of the Fourth International Conference on Smart Systems and Inventive Technology (ICSSIT-2022)
IEEE Xplore Part Number: CFP22P17-ART; ISBN: 978-1-6654-0118-0

Existing hate speech recognition methods perform badly when


 Confusion Matrix - A Confusion Matrix is a applied to fresh, previously unknown datasets. The behavior of
rundown of arrangement issue forecast results. social media users, particularly haters, puts existing NLP
The quantity of precise and wrong forecasts is approaches to the test. Deep learning models are prone to
added up to and split somewhere around class overfitting in small datasets, and dataset biases are transferred
utilizing count esteems. The disarray grid's key is to models. More work in both domains may be done to improve
generalizability in two key directions: data and models. At the
this. The disarray framework portrays the
same time, the larger context and impact should be taken into.
different manners by which your order model may
be misjudged.
REFERENCES

[1] Pandian, A. Pasumpon. “Performance Evaluation and


Comparison using Deep Learning Techniques in Sentiment
Analysis.” Journal of Soft Computing Paradigm (JSCP) 3,
no. 02 (2021): 123-134.
[2] Manoharan, J. Samuel. “Study of Variants of Extreme
Learning Machine (ELM) Brands and its Performance
Measure on Classification Algorithm.” Journal of Soft
Computing Paradigm (JSCP) 3, no. 02 (2021): 83-95.
[3] Ranganathan, G. “A Study to Find Facts Behind
Preprocessing on Deep Learning Algorithms.” Journal of
Innovative Image Processing (JIIP) 3, no. 01 (2021): 66-
74.
[4] Gaydhani, V. Doma, S. Kendre, and L. Bhagwat,
“Detecting Hate Speech and Offensive Language on
Twitter using Machine Learning: An N-gram and TFIDF
Figure 5: Confusion Matric for our Classifiers based Approach,” 2019.
[5] Akhter, M. P., Jiangbin, Z., Naqvi, I. R., Abdelmajeed, M.,
Mehmood, A., & Sadiq, M. T. (2020). Document-level text
In this matrix, we can observe that there are the predictions classification using single-layer multisize filters
correlated to their actual predictions and their correct prediction convolutional neural network. IEEE Access, 8, 42689-
samples are classified according to their labels defined these 42707.
matrices will be common for the random forest, logistic- [6] K. J. Madukwe and X. Gao, “The Thin Line Between Hate
regression, XG-Boost. and Profanity,” in Australasian Joint Conference on
Artificial Intelligence, 2019, pp. 344–356.
[7] Beeravolu, A. R., Azam, S., Jonkman, M., Shanmugam, B.,
VI. CONCLUSIONS AND FUTURE WORK Kannoorpatti, K., & Anwar, A. (2021). Preprocessing of
In The problem of generalizability affects every aspect of Breast Cancer Images to Create Datasets for Deep-CNN.
hate speech detection, including dataset creation, model IEEE Access, 9, 33438-33463.
training and evaluation, and application. As a result, the [8] Chen, Z., Zhou, L. J., Da Li, X., Zhang, J. N., & Huo, W.
challenges of detecting generalizable hate speech are largely J. (2020). The Lao text classification method based on
KNN. Procedia Computer Science, 166, 523-528.
intertwined. Uncertain and differing definitions of hate speech
[9] Diker, A., Avci, E., Tanyildizi, E., & Gedikpinar, M.
result in discrepancies in the literature, as well as sample and
annotation biases and a discrepancy between datasets, reducing (2020). A novel ECG signal classification method using
the generalizability of models trained on such data. In the order DEA-ELM. Medical hypotheses, 136, 109515.
of performance evaluation, we can conclude that SVM has [10] Heidari, M., Mirniaharikandehei, S., Khuzani, A. Z.,
performed the best with 98 percent accuracy immediately Danala, G., Qiu, Y., & Zheng, B. (2020). Improving the
followed by both random forest and logistic regression performance of CNN to predict the likelihood of COVID-
classifiers. NB classifier has yet proved to be the most 19 using chest X-ray images with preprocessing
competitive classifier in detecting the two classes of 0 & 1 algorithms. International journal of medical informatics,
corresponding to both hate and non-hate Twitter tweets data 144, 104284.
with 95 percent accuracy. XB Boost which is the latest decisive [11] Poloni, K. M., de Oliveira, I. A. D., Tam, R., Ferrari, R. J.,
algorithm in machine learning has obtained a score of 91 & Alzheimer’s Disease Neuroimaging Initiative. (2021).
percent. Brain MR image classification for Alzheimer’s disease
diagnosis using structural hippocampal asymmetrical

978-1-6654-0118-0/22/$31.00 ©2022 IEEE 1267


Authorized licensed use limited to: Dayananda Sagar University. Downloaded on November 02,2023 at 03:48:09 UTC from IEEE Xplore. Restrictions apply.
Proceedings of the Fourth International Conference on Smart Systems and Inventive Technology (ICSSIT-2022)
IEEE Xplore Part Number: CFP22P17-ART; ISBN: 978-1-6654-0118-0

attributes from directional 3-D logGabor filter responses.


Neurocomputing, 419, 126-135..
[12] Rodrigues, L. F., Naldi, M. C., & Mari, J. F. (2020).
Comparing convolutional neural networks and
preprocessing techniques for HEp-2 cell classification in
immunofluorescence images. Computers in biology and
medicine, 116, 103542.
[13] Vijayakumar, T., & Vinothkanna, R. (2020). Capsule
Network on Font Style Classification. Journal of Artificial
Intelligence, 2(02), 64-76.
[14] Wang, Y., & Shan, S. (2021). Accurate disease detection
quantification of iris based retinal images using random
implication image classifier technique. Microprocessors
and Microsystems, 80, 103350.
[15] Hutto C.J., Gilbert E. Vader: A parsimonious rule-based
model for sentiment analysis of social media text;
Proceedings of the Eighth International AAAI Conference
on Weblogs and Social Media; Ann Arbor, MI, USA. 1–4
June 2014.
[16] G. Sidorov, F. Velasquez, E. Stamatatos, A. Gelbukh, and
L. Chanona Hernández, “Syntactic N-grams as machine
learning features for natural language processing,” Expert
Syst. Appl., vol. 41, no. 3, pp. 853–860, 2014.
[17] Lichouri, M., Abbas, M., Benaziz, B., Zitouni, A., &
Lounnas, K. (2021, April). Preprocessing Solutions for
Detection of Sarcasm and Sentiment for Arabic. In
Proceedings of the Sixth Arabic Natural Language
Processing Workshop (pp. 376-380).
[18] Manoharan, S. (2019). Image detection classification and
recognition for leak detection in automobiles. Journal of
Innovative Image Processing (JIIP), 1(02), 61-70.
[19] Mitra, A. (2020). Sentiment Analysis Using Machine
Learning Approaches (Lexicon based on movie review
dataset). Journal of Ubiquitous Computing and
Communication Technologies (UCCT), 2(03), 145-152.
[20] Jacob, I. J. (2019). Capsule network based biometric
recognition system. Journal of Artificial Intelligence,
1(02), 83-94.

978-1-6654-0118-0/22/$31.00 ©2022 IEEE 1268


Authorized licensed use limited to: Dayananda Sagar University. Downloaded on November 02,2023 at 03:48:09 UTC from IEEE Xplore. Restrictions apply.

You might also like