A STUDY ON TWITTER SENTIMENT ANALYSIS USING DEEP LEARNING

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 10 Issue: 02 | Feb 2023 www.irjet.net p-ISSN: 2395-0072
© 2023, IRJET | Impact Factor value: 8.226 | ISO 9001:2008 Certified Journal | Page 799
A STUDY ON TWITTER SENTIMENT ANALYSIS USING DEEP LEARNING
Sathya M1, Thulasimani K2, DevaSurithi S3
1,3 P. G. Student, Department of Computer Engineering, GCE, Tirunelveli, TamilNadu, India
2 Professor, Department of Computer Engineering, GCE, Tirunelveli, TamilNadu, India
---------------------------------------------------------------------***---------------------------------------------------------
Abstract - Sentiment analysis is a branch of research that
examines feelings, attitudes, and reviews from many public
spheres. Now-a-days, people share their thoughts and
insights on a wide range of issues and topics via social
media. Recently social networking sites like Twitter and
Face book have become popular because users can able to
express their opinions with other internet users through
micro blogging. Today Twitter is among the most widely
used blogging sites. But the disrespectful, insensitive, or
unfair remarks that occasionally appear in online forums
drive many people away. The majority of websites are
unable to promote productive discourse, thus either heavily
restrict or fully disable user comments. Insightful data about
what is stated on Twitter is provided when sentiment
analysis is combined with Twitter. This study analyzed with
various deep-learning techniques for the classification of
negative and positive elements. Data set SemEval-2017 from
Twitter is used to train the final models and will be useful to
identify the model which produces the most accurate results.
Key Words: Sentiment analysis, social media, Twitter,
tweets, positive, negative, neutral, deep learning
1. INTRODUCTION
Twitter had developed to turn into a wellspring of
fluctuated sort of data because of the nature of small-scale
writes on which individuals post continuous messages
about their suppositions on an assortment of themes,
examines current issues, whine, and express positive
assessment for items used in day-by-day life. The primary
goal of this study is to conduct sentiment analysis on
tweets utilizing different deep learning algorithms that
classifies the tweets into the positive or negative category.
If a tweet contains both positive and negative elements,
the final message should be chosen based on which
element is more prominent. Emojis, usernames, and
hashtags in the tweets must be analyzed and converted
into a standard format.
However analyzing the sentiment expressed is
not an easy task. There are several problems in terms of
tonality, polarity, lexicon, and tweet grammar. It seems to
be highly informal and pseudo-grammatical and it's going
to be hard to grasp their background. In contrast, the
regular use of slang words, acronyms, and vocabulary
words are very popular when posted online. The
categorization of such terms by polarity is difficult for the
natural processors involved. The identification of negative,
neutral, and positive tweets are obtained using
Bidirectional- Attention-based LSTMs, CNNs, and fine-
tuning Google's pre-trained BERT architecture, which has
generally performed as a state of art for most NLP tasks.
2. RELATED WORK
Numerous studies have been conducted on fully
automated systems that extract features from datasets
devoid of human involvement [2] Utilizing novel features
like DAL scores and n-grams, among others, the sentiment
analysis for categorization was done at the phrasal level.
Syntactic details' polarity [4, 5] was employed as a feature.
However, this approach need a precise expression border
to capture the intended mood. Due to the difference in
how words are produced using DAL, which is not a
component of speech, it also cannot manage polysemy.
VADER is a straightforward rule-based model for
broad sentiment analysis[6], and contrasted its
performance with well-known state-of-the-art
benchmarks like SentiWordNet, LIWC, ANEW, and
machine learning methods like Naive Bayes and Support
Vector Machine (SVM) algorithms. Express and emphasize
sentiment, VADER blends the lexical elements with
general grammatical and syntactical norms.
Analyzing the tweets in written English that come from
various KSA telecommunications firms[7] and used
supervised machine learning techniques for classification
to execute opinion mining on them. Gauging the tweet,
how essential a word is used in a particular tweet, and also
employed with TF-IDF (Term Frequency-Inverse
Document Frequency).
Sentiment analysis approaches embedded in public
Arabic tweets and Face book comments[8]. Supervised
machine learning algorithms such as Support Vector
Machine (SVM) and Naïve Bayes, are Binary Model(BM)
and TF-IDF returns the effect of several terms weighting
functions on the accuracy of sentiment analysis. Using
natural language analysis for Arabic language text[9]
sentiment analysis is applied on the Twitter dataset of
4700 for Saudi dialect sentiment analysis with (k=0.807).

Egyptian dialect using a corpus such as tweets, and
products review data for sentiment analysis[10]. Natural
language processing technique is applied to understand
the Egyptian dialect and additionally, used a lexicon-based
classification method to classify the data. Using Support
Vector Machine for emotion analysis[11], collected
optimistic, negative, and neutral tweets from several
sources, including the Sentiment140 database. The
features are extracted from each message, such as
characters n-grams, number of hash-tags, emoticons, etc.,
for classification and the F1 score of 69.02 is obtained in
three-way classification. Similar studies in the assessment
of document classification techniques in [12], Bi-LSTM
model returns the strongest result for tweet three-way
classification and F-score value is 68.5 using SemEval
dataset.
In the course of a few years, sentimental computing
has set foot in the area of machine learning as social media
have been used by different abusers. Also, these are
required abilities for many Human-computer interaction
applications. Lot of studies have been done on sentimental
and content analysis in a combined manner for identifying
and interpreting human emotional messages[13]. Apart
from the traditional parsers that were based on normal
searching, a greedy parsing technique named Transition
based dependency is used for the classification. This
parser will give more accuracy of the classification results
3. SYSTEM ARCHITECTURE
Pre-processing of training data uses natural language
processing which is a technique accustomed to perceiving
computer information and handling human interactions.
The text comments that are given to the model are
additionally being pre-processed. Both pieces of
information are passed to the sentiment library where the
feature extraction of the pre-processed information is
being done. From this, we tend to get the trained model for
the knowledge sets. The system architecture is depicted in
Fig.1. The classification is finished using LSTM, which is
another version of the Recurrent Neural Network.
The classifier takes the input and then classifies them
as positive and negative emotions. These data are then
passed onto another classifier for further classification of
positive and negative emotions. The positive features are
then classified as enthusiasm, fun, love, happiness, neutral,
relief, and surprise. The negative features are classified
into anger, boredom, hate, emptiness, sadness, and worry.
The classifier then predicts the output of the test input,
which provides the results of the model. The text
comments from the tweets undergo pre-processing since
it contains URL id. Since we tend not to think about any
address, we eliminate all the URLs and avoid all unwanted
areas. These processes are done in pre–processing stage.
4. METHODOLOGY
4.1 Dataset
In social networking service, Twitter is a real time
messages that lets its users to post called tweets. Tweets
have many unique characteristics. Twitter, with nearly
600 million users and over 250 million messages per day,
has rapidly turned into a gold mine for organizations to
monitor their reputation and brands by extracting and
analyzing the sentiment of the Tweets posted by the public
about their remarks, markets, and other contenders.
Performing Sentiment Analysis on Twitter is complicated
than doing it for large reviews. This is because the tweets
are very short and mostly contain slangs, emoticons, hash
tags and other twitter language [16].The data consists of a
large number of tweets collected from the Kaggle
repository and Twitter. The Twitter API is used to create a
Twitter application and get authorization from tweets. The
collected tweet data is in the form of positive as well as
negative. Training and Testing dataset consist of both
positive and negative tweets.
Fig.1: Sentiment Classification Technique[1]
4.2 Preprocessing
Messages from Twitter are too informal and have
different styles of using tweets based on the nationality,
origin, age, and gender of the user. Therefore raw Twitter
information must be standardized and prepare a formal
dataset that can be effectively learned by different
classifiers. Most of the preprocessing techniques are

generic and can be used in various applications except
Sentiment Analysis. Several pre-processing methods such
as removal of URL, usernames, hashtags, character
normalization, special character removal like
punctuations, numbers, special character, lower casing
steps have been applied to standardize the dataset and
decrease its size.
4.3 Feature Extraction
Terms Presence and Frequency, Parts of Speech
(POS), Opinion words and phrases, Negations are some the
features used for feature extraction process. Individual
words like specific unigrams, bigrams and n-grams words
with their frequency counts are either uses the term
frequency weights or gives binary weighting to the words.
The important indicators of opinions of POS are used for
finding descriptive words or adjectives from the content.
Opinion and phrases words generally used to express
opinions like good or bad or hate. Some phrases are
express opinions without using opinion words. Generally a
document or sentence expresses one opinion orientation
like positive or negative about its subject matter. The
presence of negative words might change the opinion
orientation like not beautiful is equivalent to ugly.
4.4 Classification
Sentiment Classification is the binary
classification which deals with a small number of classes.
Sentiment classification is one of the simple tasks
compared to text auto-categorization. While Opinion
mining presents various extra tasks other than sentiment
polarity detection. The sentiment training set consists of
raw tweets labeled positive, negative and neutral. In an
effort of making sentiment classifiers, different methods
are compared.
4.4.1 Decision Tree
A decision tree is a type of classifier in which each
hub or node refers to a test on an attribute of a dataset and
its off-spring refer to the outcomes[18]. Decision tree
model is applied on test data for node test information.
The best test condition or choice has P to be made for each
node in the tree. GINI factor is used to select the ideal split.
For a given hub or node t,
GINI(t)=1- ∑j [ P(j|t)] (1)
where (P(j|t)) is the general recurrence or
relative frequency of class j at node t.
4.4.2 Word Embedding
A machine can only understand numbers that
convert text to numbers. But word embedding techniques
is for converting text into vectors. The word embedding
techniques are used to represent words in mathematically.
One Hot Encoding, TF-IDF, Word-to-Vec, Fast Text are
frequently used Word Embedding methods. One of these
techniques in some case is preferred and used according
to the status, size and purpose of processing the data.
The word embedding tool uses both Bag-of-
Words model and the skip-gram model is used for creating
a vector representations of words [19]. Using Neural
Networks and Vectorization of text to numbers, word-to-
vec is a well-liked approach to Natural Language
Processing.
With the help of NLP technology, word
optimization, learning, and data correctness can be
achieved. The BOW model performs more accurately than
the skip-gram model for finding frequent words. In a prior
investigation, it was found that word vectors with
semantic links enhanced the NLP process of information
retrieval and machine translation [20].
4.4.3 Random Forest
The combination of learning algorithm for
classification and regression is called Random Forest.
Based on the combined decisions of those trees, Random
Forest generates a sizable number of decision tree models.
For a large number of tweets with the individual
assessment marks as sentiment labels such as x1, x2,... xn,
packing y1, y2,... yn repeatedly and selects a random sample
like Xb, Yb with substitution. Every arrangement tree fb is
ready to use a different arbitrary example (Xb, Yb), where b
is a number between 1 and B. Finally, a majority vote is
cast on these B-trees' forecasts. Using Scikit-Random
Forest Classifier Learn's from sklearn. ensemble, a random
forest method can be implemented.
4.4.4 Support Vector Machine
Support vector machine is commonly referred to
as binary linear classifiers and non-probabilistic. When the
feature space is large, this method is recommended for
content categorization [21]. The main goal of this classifier
is to identify the maximum-margin hyper-plane that
separates the points with yi = 1 and yi = -1 for a training
set of points (xi, yi) where x is the feature vector and y is
the class. The equation of the hyper-plane is as follows:
(2)

where w is a vector normal to the hyper-plane and b is an
offset. Support Vector Machine can be run with both
unigram as well as the combination of unigram and
bigram.
4.4.5 BERT
BERT is also known as the Transformers'
Bidirectional Encoder Representations was published at
the end of 2018 [22]. Pre-training language
representations have been created using the BERT model,
which NLP experts can access and use without charge. In
deep learning model, Transformer that is primarily used
by BERT to identify the contextual relationships between
words or sub-words in the text. The basic design of
Transformer's consists of two distinct mechanisms: an
encoder that reads text input and a decoder that creates a
task prediction. BERT only needs the encoder mechanism
because it aims to develop a language model.
Contrary to directional models, which read the
input sequentially, the Transformer encoder reads the full
input sequence only once (right to left or left to right).
Therefore, it is referred to as bidirectional. This feature
enables the model to comprehend a word's meaning based
on all of its neighbors. In some test cases, Transformers
model returns better output than the Google Neural
Machine. The encoding portion is made up of a stack of
encoders. A stack of identical counted decoders makes up
the decoding portion.
The encoders are all comparable in terms of
construction. The inputs of the encoder first travel
through a self-attention layer, which enables the encoder
to view the other terms in the input expression when a
given word is encoded. Each of them is divided into two
sub-layers. The outputs of the self-attention layer serve as
the inputs for the feed-forward network. Since the decoder
contains both levels, the attention layer in between them
enables it to focus on the pertinent passages in the input
paragraph. Self-attention enables to look at specific
locations in the input text for indications that may aid to
contribute to an improved encoding of the word as the
model processes each word. Finding the Value, Key, and
Query matrices comes first. The outputs of the self-
attention layer are computed as:
[
√
] (3)
Where K is the key matrix, Q is the query matrix and d: is
the dimension of the key vector. Pre-trained BERT model
is used from the hugging face transformer library for
PyTorch.
4.4.6 Bidirectional Network
Hochreiter and Schmidhuber first developed
LSTM units in 1997 [23] to address the vanishing gradient
issue. The fundamental idea is to put in place a reliable
gating mechanism that will control how much the LSTM
units retain the derived features of the fresh data input
and maintain the previous state. The input gate and its
subsequent weight matrices Wui, Wvi, Wci, and bi; the forget
gate and its subsequent weight matrices Wuf, Wv, Wc, and
bf; and the output gate and its subsequent weight matrices
Wuo, Wvo, Wco, and bo, where pi1 denotes the current state
of the cell and vi1 state produced by the preceding stage.
The following equations represent how to choose whether
the output state created immediately or later, take inputs,
and forget the stored memory.
4.4.7. Convolutional Neural Network
The process of convolution neural network
involves acquiring input data and selecting a set of
characteristics from it. Data must be first preprocessed
before being converted to a vector format and added to a
convolution layer. Convolution layer output is pooled
using max pooling is the most popular technique. Dropout
is applied after pooling to improve accuracy [24]. Create a
new feature; a convolution algorithm is applied to filter
the input while keeping a window of m words. For
produce feature maps, filter is expanded to any window of
words
The generated feature maps are then put through
a max pooling process where the maximum value of f =
max { f } is used. For each feature map, the most pertinent
feature with the highest importance needs to be retrieved.
Pooling method is naturally addresses sentences with
different lengths. These features are subsequently passed
to a fully connected softmax layer, the penultimate layer,
whose output is non-normalized and corresponds to a
probability distribution across predicted labels. The pre-
trained word vectors can either be left static while the
model's other parameters pick up new information, or can
fine-tune them after a few training epochs once the rest of
the model has picked up some valuable information. With
a restriction on the l2-norms of the weight vector, dropout
is utilized for regularization on the penultimate layer.
5. RESULTS AND EVALUATION
The performance of sentiment classification can
be evaluated by using four indexes calculated as the
following equations:

( ) )
( )
(4)
( )
( )
(5)
( )
( )
(6)
( )
( )
(7)
Comparing the three trained classifiers, the
performance of the classifiers using 1,600,000 tested
tweets from the sentiment140 dataset.
The Precision, Recall, F1-score, and Accuracy of
the BERT, LSTM and CNN model are shown in Table I. In
BERT-base-uncased model for training, with bias weight
decay set to zero, Layer Norm weights and bias set to
0.001, and all parameters set to 0.001 and fine-tuned for
50 epochs with an 8-batch batch size to operate flawlessly.
Table – I : Comparison of BERT, LSTM, CNN
Model Precision Recall F1 Accuracy
BERT 0.64 0.65 0.64 64.50
LSTM 0.60 0.62 0.61 60.05
CNN 0.59 0.61 0.60 59.20
Similarly, In the Bidirectional Attention-based
LSTM's performance using 50-dimensional glove
embeddings and a 50-dimensional hidden dimension, fine-
tune the embeddings after a few epochs when the other
layers start picking up important information. In this
model trained 50 epochs with a batch size of 16 and used
Adam with weight decay and Cross Entropy as the
optimizer and loss function to compute gradients and
perform back-propagation[16].
The performance of the Convolutional Neural
Network-based sentiment classifier with Glove
embeddings of dimension 50 with a window size of [3,4,5],
and 128 filters in addition to the BERT and Bi-Attentive
LSTM classifiers. Adam with weight decay and Cross
Entropy, coupled with fine-tuning of the embedding layer,
serve as the optimizer and loss function, respectively.
6. CONCLUSION
Reduction of data noise and improve accuracy of a
model, three well-known deep learning models are
studied for sentiment analysis of tweets. Pre-trained
Wikipedia language model and the book corpus, which
provide a clearer knowledge of the English language and
BERT perform better than compared to other classifiers. .
REFERENCES
[1] Amit G. Shirbhate, Sachin N and Deshmukh, “ Feature
Extraction for Sentiment Classification on Twitter Data ” in
International Journal of Science and Research Vol. 5 , 2016
[2] A. Agarwal, F. Biadsy, and K. Mckeown, "Contextual
phrase-level polarity analysis using lexical affect scoring
and syntactic n-grams," in Proceedings of the 12th
Conference of the European Chapter of the ACL (EACL
2009), pp. 24–32, 2009.
[3] B. Pang and L. Lee, “A sentimental education:
Sentiment analysis using subjectivity summarization
based on minimum cuts,” arXiv preprint cs/0409058,
2004.
[4] S. A. El Rahman, F. A. AlOtaibi, and W. A. AlShehri,
"Sentiment analysis of Twitter data," in 2019 International
Conference on Computer and Information Sciences
(ICCIS), pp. 1–4, IEEE, 2019.
[5] S.-M. Kim and E. Hovy, “Determining the sentiment of
opinions,” in COLING 2004: Proceedings of the 20th
International Conference on Computational Linguistics,
pp. 1367–1373, 2004.
[6] Hutto C.J., Gilbert E., VADER: A Parsimonious Rule-
based Model for Sentiment Analysis of Social Media Text,
AAAI, 2014.
[7] Qamar, A., Alsuhibany, S. and Ahmed, S. (2017).
Sentiment Classification of Twitter Data Belonging to
Saudi Arabian Telecommunication Companies. (IJACSA)
International Journal of Advanced Computer Science and
Applications, [online]
[8] R. M. Duwairi and I.Qarqaz, "A framework for Arabic
sentiment analysis using supervised classification", Int. J.
Data Mining, Modelling and Management, Vol. 8, No. 4,
pp.369-381, 2016.
[9] Hossam S. Ibrahim, Sherif M. Abdou, Mervat Gheith,
“Sentiment Analysis For Modern Standard Arabic And
Colloquial”, International Journal on Natural Language
Computing (IJNLC), Vol. 4, No.2, pp.95-109, April 2015.
[10] Assiri, A., Emam, A. and Al-Dossari, H. (2016). Saudi
TwitterCorpus for Sentiment Analysis. International
Journal of Computer and Information Engineering, [online]
Available

at:http://waset.org/publications/10003483/saudi-
twitter-corpus-for- sentiment-analysis [Accessed 1 Mar.
2018].
[11] S. M. Mohammad, S. Kiritchenko, and X. Zhu, "Nrc-
Canada: Building the state-of-the-art in sentiment analysis
of tweets," arXiv preprint arXiv:1308.6242, 2013.
[12] J. Barnes, R. Klinger, and S. S. i. Walde, “Assessing
state-of-the-art sentiment models on state-of-the-art
sentiment datasets,” arXiv preprint arXiv:1709.04219,
2017.
[13] S. Niklander and G. Niklander, “Combining
sentimental and content analysis for recognizing and
interpreting human affects,” in International Conference
on Human-Computer Interaction, pp. 465–468, Springer,
2017.
[14] L. M. Rojas-Barahona, “Deep learning for sentiment
analysis,” Language and Linguistics Compass, vol. 10, no.
12, pp. 701–719,2016.
[15] T. Mikolov, I. Sutskever, K. Chen, G. Corrado, and J.
Dean, “Distributed representations of words and phrases
and their compositionality,” in Proceedings of the 27th
Annual Conference on Neural Information Processing
Systems (NIPS ’13), pp. 3111–3119, December 2013.
[16] J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “Bert:
Pre-training of deep bidirectional transformers for
language understanding,” arXiv preprint
arXiv:1810.04805, 2018.
[17] X. Zhang, J. Zhao, and Y. LeCun, “Character-level
convolutional networks for text classification,” in
Advances in neural information processing systems, pp.
649–657, 2015.
[18] H. Kim and Y.-S. Jeong, “Sentiment classification using
convolutional neural networks,” Applied Sciences, vol. 9,
no. 11, p. 2347, 2019.

A STUDY ON TWITTER SENTIMENT ANALYSIS USING DEEP LEARNING

More Related Content

Similar to A STUDY ON TWITTER SENTIMENT ANALYSIS USING DEEP LEARNING (20)

More from IRJET Journal (20)

Recently uploaded (20)

A STUDY ON TWITTER SENTIMENT ANALYSIS USING DEEP LEARNING