Sentiment Analysis of Tweets Using Natural Language Processing (#1130188) - 2484168
Sentiment Analysis of Tweets Using Natural Language Processing (#1130188) - 2484168
Sentiment Analysis of Tweets Using Natural Language Processing (#1130188) - 2484168
Technologies
e-ISSN: 2602-4888 2022, 6 (1), 58 -60
dergipark.org.tr/en/pub/ijmsit DOI: DOI: 10.36287/ijmsit.6.1.58
Research Article Received: May 20, 2022; Accepted: June 25, 2022
[email protected]
[email protected]
Abstract – Millions of people use Twitter and other social media sites to share their everyday thoughts in the form of tweets. It is a
short and straightforward way of expressing oneself, which is a hallmark of tweeting. As a result, we concentrated on sentiment
analysis of Twitter data in our research. Sentiment Analysis is a subset of natural language processing and text data mining. It is
feasible to investigate sentiment analysis using Twitter data. performed in a number of different circumstances The technique of
finding valuable patterns from textual data is referred to as sentiment analysis. Using particular analysis tools, these valuable
patterns include evaluating and categorizing feelings as neutral, positive, or negative.
The study's authors look at a range of information processing approaches, including sentiment analysis, Twitter's network
structure, event dispersion across the network, and impact identification. There have been several ways described for exploring
semantics for sentiment analysis, which can be classified into contextual semantic and conceptual semantic approaches. One of the
most important disciplines of natural language processing is sentiment analysis. The technique of finding valuable patterns from
textual data is referred to as sentiment analysis. According to this research, sentiment analysis applications will continue to develop
in the future, and sentiment analytical approaches will become more standardized across systems and services. Because of the vast
amount of data available, Twitter is one of the best virtual environments for tracking and monitoring information.
Citation: Shihab, F., Ekmekci, D. (2022). Sentiment Analysis Of Tweets Using Natural Language Processing . International Journal of
Multidisciplinary Studies and Innovative Technologies, 6(1): 58-60
58
International Journal of Multidisciplinary Studies and Innovative Technologies, 2022, 6(1): 58 – 60
the distribution or frequency of sentiment classes in the worth noting that when using these dictionaries, the problem
dataset they're looking at. Mandal, to study and forecast a of different emotions in words arises, especially that humans
situation, a based classification algorithm is used. can express their emotions in different ways.
The polarity of a user's feelings. On words, it used An example of this is: ―Oh, peace!‖ It can have a meaning of
positive, comparative, and superlative degrees of joy or a negative meaning, such as sarcasm. Sentiment
comparison.[5] companies can respond to their needs, and analysis systems rely on a number of natural language
improve their services and products to suit those needs.[6] processing algorithms and methods, and we will mention
For example, a company used sentiment analysis to analyze them below:
more than 3,000 opinions about a particular product of its
1- rule-based methodologies: rule-based methodologies.
own, and the company discovered that those opinions were
These methodologies rely on manually defining a set of rules
satisfied with the price of that product but in return complain
to determine the polarity and target entity (eg product name).
about the company's customer service. What this information
Some of the techniques used to build these bases: Stemming,
helps us not only understand the current context of a
tokenization, and parsing.Dictionaries (a group of words and
particular topic, but also enables us to predict the future, in
expressions with corresponding feelings).Example: Create
addition to the possibility of using that information contained
two strings of words that directly identify polarity (negative
in the texts to calculate the positive/negative index, which is
words, such as: bad, failed..., and positive words, such as:
an important indicator in the decision-making process. Case
good, great...).
in point: some governments use sentiment analysis results
during their election campaigns. Count the words with positive feelings and words with
Despite the fact that sentiment analysis and emotional negative feelings, that are in a particular context.[9]rule-
intelligence are sometimes employed interchangeably, they based methodologies: rule-based methodologies. These
are not the same thing. While sentiment analysis relies on methodologies rely on manually defining a set of rules to
data to classify sentiments as positive, negative, or neutral, determine the polarity and target entity (eg product name).
EQ digs deeper into the nuance of the emotions expressed in Some of the techniques used to build these bases: Stemming,
the remarks. EQ is far more challenging and complex than tokenization, and parsing.
sentiment analysis.[7]
Sentiment analysis, for example, will determine if a given 2-Methodologies based on machine learning techniques.
statement is good, negative, or neutral, but emotional Based methodologies. Unlike the previous methodology,
intelligence will determine whether the comment causes these methodologies do not rely on hand-written rules, but
grief, discontent, or sarcasm if it is determined to be negative. rather on machine learning algorithms. The issue of sentiment
analysis is a classification problem. The input to this
classifier is textual data, and its output is one of the kinds of
III. MATERIAL AND METHOD sentiment (for example: positive, negative, neutral). The
following figure shows how the sentiment analysis classifier
Sentiment analysis is based on several models (according works:
to the goal), starting with the models that are concerned with
polarity only (positive, negative, neutral), and passing
through the models capable of determining emotion (anger,
love, happiness…), ending with those that are concerned with
revealing intentions (interested, not). interested).[8]
2.1 Fine-grained Sentiment Analysis:
In some cases, polarity accuracy is required, and then the
classes of polarity are expanded as follows: very positive,
positive, neutral, negative, very negative Items in this model
can also be viewed as a numerical rating, such as those based
on five stars. very positive =very negative.
2.2 Emotion Detection:
In this model, the goal is to identify emotions, such as:
happiness, fear, anger, etc., and dictionaries are usually used
here (a list of words with their corresponding emotions), and
automatic learning algorithms can also be used. Here it is
worth noting that when using these dictionaries, the problem
of different emotions in words arises, especially that humans
can express their emotions in different ways. An example of
this is: ―Oh, peace!‖ It can have a meaning of joy or a
negative meaning, such as sarcasm.
2.3 Aspect-based Sentiment Analysis:
In this model, the goal is to identify emotions, such as:
happiness, fear, anger, etc., and dictionaries are usually used
here (a list of words with their corresponding emotions), and
automatic learning algorithms can also be used. Here it is Fig. 1 analysis classifier works
59
International Journal of Multidisciplinary Studies and Innovative Technologies, 2022, 6(1): 58 – 60
IV. RESULTS AND DISCUSSION analytics. ACM,2010. B. Hammi, R. Khatoun, S. Zeadally, A. Fayad,
Twitter posts on electronic products were used to create and L. Khoukhi, ―Iot technologies for smart cities,‖ IET Networks,
the dataset. Tweets are short statements that are rife with vol. 7, no. 1, pp. 1–13, 2017.
[4] Gao, Wei, and Fabrizio Sebastiani. "Tweet sentiment: From
slang and misspellings. As a result, we conduct a sentiment classification to quantification." Advances in Social Networks
analysis at the sentence level. This is accomplished in three Analysis and Mining (ASONAM), 2015 IEEE/ACM International
stages. Pre-processing is completed in the first phase. Then, Conference on. IEEE, 2015.
utilizing relevant features, a feature vector is produced. [5] Mandal, Santanu, and Sumit Gupta. "A Lexicon-based text
classificationmodel to analyses and predict sentiments from online
Finally, tweets are divided into positive and negative reviews." Computer, Electrical & Communication Engineering
categories using several classifiers. The final emotion is (ICCECE), 2016 International Conference on. IEEE, 2016.
calculated based on the number of tweets in each class.
HUANG, Lei generated a novel dataset by collecting [6] Wagh, R., & Punde, P. (2018, March). Survey on sentiment analysis
using twitter dataset. In 2018 Second International Conference on
tweets over a period of time lasting from April 2013 to May Electronics, Communication and Aerospace Technology (ICECA)
2013 because conventional Twitter datasets are not accessible (pp. 208-211). IEEE.
for the electronic products industry.[10] [7] C hakraborty, K., Bhatia, S., Bhattacharyya, S., Platos, J., Bag, R., &
The Twitter API is used to automatically collect tweets, Hassanien, A. E. (2020). Sentiment Analysis of COVID-19 tweets by
Deep Learning Classifiers—A study to show how popularity is
which are then manually tagged as good or negative. Using affecting accuracy in social media. Applied Soft Computing, 97,
600 good and 600 negative tweets, a dataset is produced. 106754.
Creation of a Dataset [8] S harma, N., Pabreja, R., Yaqub, U., Atluri, V., Chun, S. A., &
Table 1 shows how the dataset is divided into two sets: Vaidya, J. (2018, May). Web-based application for sentiment analysis
of live tweets. In Proceedings of the 19th Annual International
training and test. Conference on Digital government research: Governance in the data
Age (pp. 1-2).
[9] MEDHAT, Walaa; HASSAN, Ahmed; KORASHY, Hoda. Sentiment
analysis algorithms and applications: A survey. Ain Shams
engineering journal, 2014, 5.4: 1093-1113.
[10] GO, Alec; HUANG, Lei; BHAYANI, Richa. Twitter sentiment
analysis. Entropy, 2009, 17: 252.
V. CONCLUSION
REFERENCES
[1] Boia, Marina, et al. "A:) is worth a thousand words: How people
attachsentiment to emoticons and words in tweets." Social computing
(socialcom),2013
[2] Manuel, K., Kishore Varma Indukuri, and P. Radha Krishna.
"Analyzinginternet slang for sentiment mining." 2010 Second
VaagdeviInternationalConference on Information Technology for
Real World Problems. 2010.
[3] Akcora, Cuneyt Gurcan, et al. "Identifying breakpoints in
publicopinion." Proceedings of the first workshop on social media
60