Classification of Sentiment Analysis on Tweets Based on Techniques from Machine Learning

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 10 Issue: 01 | Jan 2023 www.irjet.net p-ISSN: 2395-0072
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 761
Classification of Sentiment Analysis on Tweets Based on Techniques
from Machine Learning
Anjali Singh1, Mr.Manish Kumar Soni2
1M.Tech, Computer Science Engineering, Bansal Institute of Engineering & Technology, Lucknow, India
2 Assistant Professor, Department of Computer Science Engineering, Bansal Institute of Engineering & Technology,
Lucknow, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - This trend is expected to continue growing in
popularity. When it comes to making specific choices, the
public and the stakeholders might benefit from hearing the
thoughts of the people. The process of retrieving information
via the use of search engines, online blogs, micro-blogs,
Twitter, and social networks is referred to as opinion mining.
The user-generated material on Twitter provides a wealth of
opportunities to collect the perspectives of people. It is
challenging to physically delineate the information due to the
enormous volume of tweets, which are in the form of
unstructured text. Therefore, to mine and condensethetweets
from the corpora, one has to be knowledgeable in
computational techniques and have an understanding of
phrases that convey emotions. The extraction of emotionfrom
unstructured text may be accomplished by the use of a wide
variety of computer models, methodologies, and algorithms.
The vast majority of them are based on machine-learning
strategies, with the Bag-of-Words (BoW) representation
serving as the primary foundation. In this investigation, we
used a lexicon-based technique to automatically identify
sentiment for tweets that were gathered from the public
domain of Twitter.
Key Words: Bag-of-Words (BoW), Lexicon, Machine
Learning Algorithms, Laplace Smoothing, Part-of-Speech
(POS)
1. INTRODUCTION
In addition to this, an increasing number of people are
turning to the internet to make it possibleforpeoplelivingin
other countries to have access to their point of view.
According to the findings of two surveys that were
conducted on more than 2000 adults in the United States,
each eighty percent of Internet users (or sixty percent of
Americans) have conducted research on a product online at
least once, and twenty percent (fifteen percent of
Americans) prefer it on a specific day. These statistics were
derived from the findings of two surveys that were carried
out in the United States. The polls were conducted in the
United States by two different organizations that are
independent of one another. The United States of America
served as the location for all of the polling and interviewing
that took place.
On the other hand, researchers working in the sector found
the fact that around eighty thousand new blogs and two
million new articles are generated every single day. Because
of the widespread use of the internet and the continually
shifting behaviors of customers, many of the conventional
techniques of monitoring are becoming more useless. As a
direct result of this, there is a desire for cutting-edge
technology that is associated with product photographs for
quality control. Therefore, in addition to individuals, a
separate audience group for systems that can automatically
analyze consumer sentiment,suchasisdescribedinnosmall
part in online forums, are businesses that are eager to
comprehend how their products and services are being
perceived by customers in the market.
1.1. Principle of Sentiment Analysis
The source of the opinion, the target of the view, and the
evaluative remarksorcommentsmadebytheopinionholder
are the primary components that make up a SA problem.
These components are not listed in any particular sequence.
According to Liu [6,] the definition of a SA problem is as
follows: "Given a set of evaluative text documents D that
contain opinions (or sentiments) about an object, Sentiment
Analysis aims to extract attributes and components of the
object that have been commented on ineachdocumentd €D
and to figure out whether the comments are positive,
negative, or neutral."
1.2. Tracking down the Origins of Opinions
The person or medium that is responsible for the
transmission of a feeling is considered to be the feeling's
"carrier." The genesis of a feeling is the person or medium
that is responsible for its transmission. The genesis of a
feeling is of the utmost importance when it comes to
validating and categorizing that feeling since it givescontext
to the emotion in the issue. This is because the origin of a
position has a significant impact, both positive and negative,
on the credibility and dependabilityofthatviewpoint. Thisis
because the origin of a perspective hasa majorimpactonthe
quality of a feeling, which is expressed bythatviewpoint. For
instance, the weight of an expert's perspective may be
compared to the weight of the opinionoftheaverage person,
and an opinion is considered to be credible when it is
received from a respectable source.

2. COLLECTION AND PREPROCESSING OF DATA
The rapid development of communication and information
technology has resulted in the emergence of an entirely new
set of challenges, which has greatly increased the level of
difficulty associated with the process of information
dissemination. One cannot resist emphasizing the relevance
of social networks such as Facebook, Twitter, and MySpace
while discussing the regulations that regulate the
transmission of information and the collection of business
intelligence. Examples of such networks are Facebook,
Twitter, and MySpace. Tweets from Twitter were used to
ensure that the behavior that was the subject of the study
could be maintained. Twitter is unique among social
networking sites in that each messagemaybenolongerthan
140 characters. This character limit sets Twitter apart from
other social networking services. Despite this limitation,the
information that is included in tweets is of immense value
since it is feasible to extract a significant amount of data
from such a restricted amount of space. Additionally, the
images, videos, and transcripts of the presentations may all
be seen together in one location, making it feasible to
comprehend the whole narrative in a single sitting. Theease
of access to the data is yet another crucial factor to take into
account while making this decision.Inthepast,we wereonly
able to put our model through its paces using a training set
consisting of hundreds of objects. However, now that we
have access to the Twitter API, we can gather millions of
tweets to utilize as data for training our model. In the past,
we were only able to put our model through its pacesusinga
training set that had hundreds of different objects.
2.1. The Twitter API
When using the Twitter platform, users have access to two
different kindsofapplication programminginterfaces(APIs).
These APIs are referred to as REST and Streaming
respectively. In addition, the REST API is comprised of two
more APIs, namely the REST API and the Search API (whose
difference is due to their history of upgradation). The
Streaming API, in contrast, to REST APIs, keeps an active
connection to the server at all times and transfers data in a
manner that is quite close to being in real-time. This
differentiates it from REST APIs in a significant way. On the
other hand, connections to REST APIs will only be permitted
if they are active for a certain period and adhere to specified
rate limitations (one can download a restricted amount of
data per day). Users can access the data that Twitter has to
offer at any time of the day or night thanks to the REST APIs.
2.2. From Twitter using Third Party API(Twython)
It is essential to collect only tweetsthatarespecificallyabout
that product or movie given that the purpose of the thesis is
to determine the sentiment (positive,negative,orneutral)of
tweets about a specific product or a movie, and it is the
purpose of the thesis to determine whethertweetsaboutthe
product or movie are positive, negative, or neutral. This is
because the objective of the thesis is to analyze the
sentiments expressed in tweets that are related to the
product or movie in question. This is a result of the fact that
the purpose of the thesis is to analyze the feelings that are
expressed in tweets that are relevant to the product or
movie that is under discussion. Having said that, puttingthis
idea into effect won't be an easy task by any stretch of the
imagination. It would seem that no method canbeemployed
to use to collect all of the tweets that have been made about
a certain subject that has been discussed.
Figure 1: Twitter data snapshot using a third-party
API
Figure 2: A snapshot of tweet volume relative to the
period for "rage"

Figure 3: An overview of the number of tweets related
to the hashtag "sad"
2.3. PREPROCESSING OF DATA
Preprocessing, which is an essential phase, is requiredtoget
rid of data that is either incomplete, noisy, or inconsistent.
This is because preprocessing is a crucial step. This is
necessary due to the need for preprocessing. Before using
any of the data mining capabilities, it is necessary to do
preprocessing in its entirety. This need must be satisfied.
Before using any of the many methods that are available to
us for doing sentiment analysis, we first carried out the
preprocessing stage, which will be discussed in more depth
in the following paragraphs (lexicon-based or machine
learning).
2.3.1. Removal of the hash symbol
There is a certain label that goes by the name of a hashtag,
and you can find it on social networkingsitesaswell asother
kinds of microblogging services. This form of label is used to
categorize posts. Users are now in a position to identify
messages that involve a certain topic matter or content in a
far more expedient manner than was before feasible as a
direct result of these labels. This advancement was
previously impossible. For example, if you do a search using
the hashtag #LOST (or #Lost or #lost as the search is not
case-sensitive), we will get a list of tweets that are related to
the hashtag #lost. These tweets will include references to
#LOST.
3. LEXICON-BASED APPROACH IN SENTIMENT
ANALYSIS
The use of terms that are thought to be indicative of either a
positive or negative bias is taken into consideration to
achieve the detection of subjectivity and the categorization
of attitudes. This is done to accomplish both of these goals.
The way of thinking that drives this technique is dependent
on the idea that the meanings attributed to words may be
understood in terms of the substance of views. This thesis
serves as the foundation for this line of thought. The corpus
of academic work that has been done in this field has
documented the deployment of severalbeneficial techniques
that are based on this strategy. These approaches have been
successful in achieving their goals. For instance, Turney and
colleagues [13] provide a method for the identification of
subjectivity that is based on the use of a list of seed words
that is determined by the proximity measure to other
common phrases.
3.1. SentiWordNet
SentiWordNet is a lexical resource that was developedto aid
with activities requiring opinion mining and sentiment
analysis [14]. Its purpose was to help with these types of
tasks. Its only objective was to assist in the aforementioned
endeavors. An approach that is only semi-automatic is used
to generate SentiWordNet's term-level opinion polarity,
which is provided to users. The WordNet database, which
contains English words and their relationships, isthesource
of the information that was utilized to generate thispolarity.
3.2. WordNet
WordNet is a lexical database for the English language that
was developed at Princeton University to realize the nature
of semantic relations of terms in the English language.
WordNet was developed as a lexical database fortheEnglish
language because of the purpose of realizing these relations.
Realizing the importance of these connections was the
driving force for the creation of WordNet, a lexical database
for the English language. The dawning understanding of the
significance of these associations served as the impetus for
the development of WordNet, a lexical database designed
specifically for the English language.
3.3. Tagging of Speech Elements
The display of the information that can be accessed on
SentiWordNet is in the hands of the POS, which is
responsible for its proper organization. there are significant
shifts that can take place in the amount of objectivity that a
synset may reflect depending on the grammatical function
that it plays. This can be seen as a direct result of the fact
that there is a direct result of this. When classifying a source
text, we need to extract the information on the parts of
speech so that we may correctly apply the scores that we
have acquired from the SentiWordNet database. This is
required so that we can properly categorize the source text.
3.4. PROPOSED MODEL
In the last section of this chapter, the organization of the
SentiWordNet database was dissected in great detail, and
questions were raised about the challenges and constraints

that are involved with the process of accumulating the
required amount of opinion data. By making use of
SentiWordNet, we were able to devise a model for the
organization of features that have the potential to be used in
the course of sentiment analysis. This is the location where
you may find the model. The building of this model was
accomplished while keeping all of these different issues in
mind at the same time. However, the methodology for a list
of features suggested in this section beginsfromtherulethat
the features obtained through SentiWordNet catch different
aspects of tweet sentiment and are best suited to train any
classifier algorithm. This rule serves as the foundation for
the methodology for a list of features suggested in this
section. The technique for a list of characteristics that are
recommended in this part is built on top of this rule, which
acts as the basis.
Figure 4: lexical snapshot produced from the tweet
corpus
Figure 5: some quotes from a review of a film
4. INTERVENTIONAL SETUP
Figure 6 illustrates the component of our system that we
have judged to be of the highest relevance, so go ahead and
have a look at it. As we go ahead, we are going to discuss this
issue in a more in-depth manner, focusing more on the
details. To successfully carry out the implementation, we
have taken use of our data collection, which is made up of
6000 tweets that have been sorted by hand into several
categories. The compilation of tweets includes talks about
businesses and services like Apple, Google, Microsoft, and
Twitter, among others. Positive, negative, neutral, and
irrelevant are the four classifications that may be applied to
tweets. Tweets are considered to be unimportant if they are
written in a language other than English or if there is no
relationship between the tweet and the issue that is the
focus of the conversation that is taking place right now.
Throughout our investigation, we have placed the majority
of our emphasis on three distinctcategories:thepositive,the
negative, and the neutral.
Figure 6: The experiment's intended block diagram

5. RESULT AND DISCUSSION
We investigate a wide range of different factors, each of
which is known to have a significant impact on the findings
of sentiment analysis. We made use of N-gram features such
as unigrams (n = 1) and bigrams (n = 2), which are used
often in a variety of text classifications including sentiment
analysis. Specifically, we employed unigrams with n = 1 and
bigrams with n = 2. We have focused specifically on
unigrams and bigrams in our work. During our inquiry, we
experimented with boolean qualitiesbyusingbothunigrams
and bigrams in our work. This allowed us to examine the
relationship between the two. Each n-gram feature has a
boolean value associated with it, which may be turned on or
off according to the user's preferences. This value is only
made to be true under the very precise condition that the
required n-gram can be located in the tweet [12], which is a
criterion that must be met.
Table 1: F1 rating for the MNB classifier
S.No Class
label
Precision(%) Recall(%) F1
score(%)
1 Positive 65.25% 20.51% 31.21%
2 Negative 77.41% 16.05% 27.20%
3 Neutral 80.48% 61.82% 69.93%
Figure 7: ROC curve for the MNB tweet classifier
The findings of our research indicate that the F1 score for
the positive class and the negative class is much lower than
the score for the neutral class. On the other hand, the score
for the class that was seen as neutral was substantially
higher. This is because the great majority of the tweets that
were included in our data collection were manually
categorized as terms that were either irrelevant or neutral.
This is the reason why this is the case. The reasoningforwhy
things are the way they are may be summed up as follows:
As a direct result of this, to use any of the machine learning
methods, we will need a training data set that is free from
any kind of bias and does not call for any kind of human
annotation. This will be necessary forustobeabletouse any
of the machine learning methods.
5.1. EMOTION DATASET
Hashtags are often used by people as a means of
communicating the thoughts and feelings that are currently
going on inside their minds. As a consequence of this, one
may be able to deduce a sufficient number of ideas and
feelings from the utterances that include the hashtag. To
provide our machine learning algorithm with more data, we
have upgraded it to take into consideration these hashtags.
This will allow it to process more information.
Figure 8: Picture of the emotion dataset
6. CONCLUSION
This research is academic in thefieldofsentimentanalysis;it
focuses on the application of lexical resources and machine
learning algorithms to the problem of classifying the
emotional tone of tweets and textmessages,twoexamples of
unstructured data sources. Due to the abundance of
subjective information accessible online, applications of
Sentiment Analysis, which uses an automated approach to

detect subjective content in a text, may be valuable in many
sectors, including online advertisingandmarketresearch.In
the realm of knowledge management, this kind of opinion
data is a crucial criterion since it is often the determining
factor in major decisions. The reason we're doing this study
is to learn more about the difficulties of Sentiment Analysis
and the many approaches that have been developed to deal
with them. Given the volume and variety of social media
data, extracting its underlying sentiment might be
challenging. Researching which elementsaremostuseful for
Sentiment Analysis, wepickedtweetsfromthepublicstream
and analyzed them. We have tried lexicon-based methods as
well as machine-learning techniques for SA.
REFERENCE
[1] John A Horrigan. Online shopping. Pew Internet &
American Life Project Report, 36, 2008.
[2] Bing Liu. Sentiment analysis and opinion mining.
Synthesis Lectures on Human Language Technologies,
5(1):1–167, 2012.
[3] Bo Pang and Lillian Lee. Opinion mining and sentiment
analysis. Foundations and trends in information retrieval,
2(1-2):1–135, 2008.
[4] Bo Pang and Lillian Lee. A sentimental education:
Sentiment analysis using subjectivity summarization based
on minimum cuts. In Proceedings of the 42nd annual
meeting on Association for Computational Linguistics, page
271. Association for Computational Linguistics, 2004.
[5] Bing Liu. Opinion mining and sentiment analysis. In Web
Data Mining, pages 459–526. Springer, 2011.
[6] Bing Liu. Sentiment analysis and subjectivity. Handbook
of natural language processing, 2:627–666, 2010.
[7] Andr´es Montoyo, Patricio Mart´ıNez-Barco, and
Alexandra Balahur. Subjectivity and sentiment analysis: An
overview of the current state of the area and envisaged
developments. Decision Support Systems, 53(4):675–679,
2012.
[8] Erik Cambria, Bjorn Schuller, Yunqing Xia, and Catherine
Havasi. New avenues in opinion mining and sentiment
analysis. IEEE Intelligent Systems, 28(2):15–21, 2013.
[9] Khairullah Khan, Baharum Baharudin, Aurnagzeb Khan,
and Ashraf Ullah. Mining opinion components from
unstructured reviews: A review. Journal of King Saud
University-Computer and Information Sciences, 26(3):258–
275, 2014.
[10] Ronen Feldman, Moshe Fresko, JacobGoldenberg,Oded
Netzer, and Lyle Ungar. Extracting product comparisons
from discussion boards. In Data Mining, 2007. ICDM 2007.
Seventh IEEE International Conference on, pages 469–474.
IEEE, 2007.
[11] Mohammad Sadegh, Roliana Ibrahim, and Zulaiha Ali
Othman. Opinion mining and sentiment analysis: A survey.
International Journal of Computers &Technology,2(3):171–
178, 2012.
[12] Bo Pang, Lillian Lee, and Shivakumar Vaithyanathan.
Thumbs up? sentiment classificationusingmachinelearning
techniques. In Proceedings of the ACL-02 conference on
Empirical methods in natural language processing-Volume
10, pages 79–86. Association for Computational Linguistics,
2002.
13] Peter D Turney. Thumbs up or thumbs down?: semantic
orientation appliedtounsupervisedclassificationofreviews.
In Proceedings of the 40th annual meetingonassociationfor
computational linguistics, pages 417–424. Association for
Computational Linguistics, 2002.
[14] Andrea Esuli and Fabrizio Sebastiani. Sentiwordnet: A
publicly available lexical resource for opinion mining. In
Proceedings of LREC, volume 6, pages 417–422. Citeseer,
2006.
[15] Alaa Hamouda and Mohamed Rohaim. Reviews
classification using sentiwordnet lexicon.In WorldCongress
on Computer Science and Information Technology, 2011.
[16] Bruno Ohana. Opinion mining with the sentwordnet
lexical resource. 2009.
[17] Mitchell P Marcus, Mary Ann Marcinkiewicz, and
Beatrice Santorini. Building a large annotated corpus of
english: The penn treebank. Computational linguistics,
19(2):313–330, 1993.
[18] Doug Cutting, Julian Kupiec, JanPedersen,andPenelope
Sibun. A practical part-of-speech tagger. In Proceedings of
the third conference onAppliednatural languageprocessing,
pages 133–140. Association for Computational Linguistics,
1992.
[19] Shitanshu Verma and Pushpak Bhattacharyya.
Incorporating semantic knowledge for sentiment analysis.
Proceedings of ICON, 2009.
[20] Kamal Nigam, John La_erty, and Andrew McCallum.
Using maximum entropy for text classification In IJCAI-99
workshop on machine learning for information filtering,
volume 1, pages 61–67, 1999.
[21] Huifeng Tang, Songbo Tan, and Xueqi Cheng. A survey
on sentiment detection of reviews. Expert Systems with
Applications, 36(7):10760–10773, 2009.

[22] Daniel M Bikel and Je_rey Sorensen. If we want your
opinion. In Semantic Computing, 2007. ICSC 2007.
International Conference on, pages 493–500. IEEE, 2007.
[23] Hiroshi Kanayama and Tetsuya Nasukawa. Fully
automatic lexicon expansion for domain-orientedsentiment
analysis. In Proceedingsofthe2006ConferenceonEmpirical
Methods in Natural Language Processing, pages 355–363.
Association for Computational Linguistics, 2006.
[24] Matthias Hagen, Martin Potthast, Michel B¨uchner, and
Benno Stein. Twitter sentiment detection via ensemble
classification using averaged confidence scores.In Advances
in Information Retrieval, pages 741–754. Springer, 2015.
[25] Saif M Mohammad, Svetlana Kiritchenko, and Xiaodan
Zhu. Nrc-canada: Building the state-of-the-art in sentiment
analysis of tweets. 2013.
[26] George Miller, Christiane Fellbaum, Randee Tengi,
PWakefield, H Langone, and BR Haskell. WordNet.MITPress
Cambridge, 1998.
[27] GebrekirstosGebremeskel.Sentimentanalysisoftwitter
posts about news. Sentiment Analysis. Feb, 2011.
[28] I Hemalatha, GP Saradhi Varma, and A Govardhan.
Preprocessing the informal text for e_cient sentiment
analysis. International Journal, 2012.

Classification of Sentiment Analysis on Tweets Based on Techniques from Machine Learning

More Related Content

Similar to Classification of Sentiment Analysis on Tweets Based on Techniques from Machine Learning (20)

More from IRJET Journal (20)

Recently uploaded (20)

Classification of Sentiment Analysis on Tweets Based on Techniques from Machine Learning