TWITTER SENTIMENT ANALYSIS

Project report submitted in partial fulfillment of the

requirement for thedegree of

Bachelor of
Technology In
Computer Science and
Engineering By
Robin Singh
(161240) Under the
supervision of

Dr. Hari Singh

To

Department of Computer Science & Engineering and

Information Technology
Jaypee University of Information Technology Waknaghat, Solan
173234, Himachal Pradesh
 Candidate’s Declaration

I hereby declare that the work presented in this report entitled TWITTER
SENTIMENT ANALYSIS in fulfilment of the requirements for the award of the
degree of Bachelor of Technology in Computer Science and Engineering/Information
Technology submitted in the department of Computer Science & Engineering and
Information Technology, Jaypee University of Information Technology, Waknaghat is
an authentic record of my own work carried out over a period from August 2018 to
June 2020 under the supervision of Dr. Hari Singh, Associate Professor, Computer
Science and Engineering/Information Technology.
The matter embodied in the report has not been submitted for the award of any
other degree or diploma.

Robin Singh, 161240

This is to certify that the above statement made by the candidate is true to the best of
my knowledge.

Dr. Hari Singh

Assistant Professor
(Senior Grade)

Computer Science and Engineering / Information Technology

Dated:

i
 ACKNOWLEDGEMENT

I have taken efforts in this project. However, it would not have been possible without
the kind support and help of many individuals and organizations. I would like to
extend our sincere thanks to all of them.

I am highly indebted to Dr. Hari Singh for his guidance and constant supervision as
well as for providing necessary information regarding the project and also for their
support in completing the project.

I would like to express our gratitude towards our parents and Jaypee University of
Information Technology for their kind cooperation and encouragement which helped
us in completion of this project.

Our thanks and appreciations also go to our colleague in developing the project and
people who have willingly helped us out with their abilities.

ii
TABLE OF CONTENTS

Title Page no.

1. Chapter- 1 Introduction
1.1 Introduction 1
1.2 Problem Statement 4
1.3 Objective 4
1.4 Methodology 5
1.5 Organization 6

2. Chapter-2 Literature Survey

2.1 Opinion Mining 7
2.2 Twitter 7
2.3 Microblogging with E-commerce 8
2.4 Social Media 8
2.5 Twitter Sentiment Analysis 9
2.6 Techniques of Sentiment Analysis 12
2.7 Application Programming Interface 14
2.8 Python 14

3. Chapter-3 System Development

3.1 NLP 16
3.2 Platform Used 17
3.3 Python 19
3.4 Modules Used 20

4. Chapter-4 Performance Analysis

4.1 Tweets Extraction 34

iii
 4.2 Making a pandas Data frame 36
4.3 Cleaning tweets 40
4.4 Calculation of Polarity and subjectivity 41
4.5 Visualizing the tweets 42

5. Chapter-5 Conclusion 43

References 44

iv
List of Abbreviations

 NLP: Natural Language Processing

 NLTK: Natural Language Toolkit
 JSON: JavaScript Object notation
 AI: Artificial Intelligence
 ANN: Artificial Neural Network
 SVM: Support Vector Machine
 API: Application Programming Interface
 MDM: Mobile Device management

v
List of Figures

Fig. No Figure Caption

1 Steps of Sentiment Analysis

2 Venn Diagram for NLP

3 Installing Tweepy

4 Accessing twitter API using OAuth

5 Updated status using twitter API

6 Twitter Application settings

7 Accessing user entities

8 Stream Listener methods

9 Tweeting a status

10 Result of on_status()

11 Prints all the hashtag values

12 Similarity between TextBlob and Strings

13 Commands to install WordCloud

14 Word Cloud of hobbies

15 Extracting tweet count from user

vi
16 Extracting 5 recent tweets

17 Result of tweet count and extraction

18 Creating and displaying data frame

19 Accessing information inside single tweet

20 Attributes of tweet

21 Most liked and most retweeted tweet

22 Result of tweet analysis

23 Syntax of sub method

24 Cleaning up the tweets

25 Polarity and Subjectivity methods

26 Generating word cloud

vii
 ABSTRACT

Social media have received more attention nowadays. Public and private opinion about a
wide variety of subjects are expressed and spread continually via numerous social media.
Twitter is one of the social media that is gaining popularity. Twitter offers organizations
a fast and effective way to analyze customers' perspectives toward the critical to success
in the market place. Developing a program for sentiment analysis is an approach to be
used to computationally measure customers' perceptions. This paper reports on the design
of a sentiment analysis, extracting a vast amount of tweets. Python is used in this
development along with various modules such as Tweepy, numpy, pandas and Textblob.
Results classify customers' perspective via tweets into positive and negative, which is
represented in a pie chart and tabular form.

viii
 Chapter 1: Introduction

In this chapter we have discussed about the general idea around this project, specified the
problem statement, objectives, methodology and the general organization of the project.

1.1 Introduction:

Sentiment Analysis is a machinery based method of interpreting text and crucial the
feelings of the text into good, bad or neutral. Performing arts Sentiment Analysis on
Twitter knowledge will facilitate corporations acquire qualitative insights to know
however folks are talking concerning their whole. With over thirty million active users,
causation daily average of five hundred million tweets, Twitter has become one in every
of the highest social media platforms for news, data, and interaction with brands and
cogent figures round the world. Therefore, it's no surprise that corporations contemplate
this micro blogging platform a necessary channel for his or her selling strategy and to
supply client service. Twitter permits businesses to succeed in a broad audience and
connect with customers while not intermediaries. On the drawback, it is damage a
whole’s name if negative content concerning the brand suddenly goes infectious agent –
you would possibly find yourself with associate surprising PR crisis on your hands. this is
often one in every of the explanations why social listening that's observation spoken
communication and feedback in social media ― has become an important method in
social media selling.

Monitoring Twitter permits corporations to know their audience, keep it up prime of

what's being said concerning their whole and their competitors, and see new trends within
the trade. However, once it involves analyzing Twitter knowledge, quantitative metrics
just like the range of mentions or retweets don't seem to be enough to induce a full image
of a scenario. What counts is having the ability to understand the significance of these

1
mentions. Are users talking completely or negatively a few product or some topic? And
that’s precisely what sentiment analysis determines. It provides qualitative insights on
what's being aforesaid a few topic or whole.

In this project, we'll take a more in-depth check up on sentiment analysis and the way
you'll be able to use it to research Twitter knowledge. We'll have a depth practice of the
complete method, from recommending some tools to gather information from Twitter, to
obtaining you up with the steps involved and running with Twitter sentiment analysis
program.

Definition:

Sentiment analysis (a.k.a opinion mining) is that the machine-controlled method of

distinctive and extracting the subjective data that written language. this could be either an
opinion, a judgment, or a sense a few specific topic or subject. the foremost common
variety of sentiment analysis is termed ‘polarity detection’ and consists in classifying a
press release as ‘positive’, ‘negative’ or ‘neutral’.

For example, allow us to take this sentence: “I don’t realize the app useful: it’s extremely
slow and perpetually crashing”. A sentiment analysis model would mechanically tag this
as Negative. A sub-field of tongue process, sentiment analysis has been obtaining plenty
of attention in recent years thanks to its several exciting apps in a very sort of fields,
starting from business to political studies.

Natural Language Processing: natural language processing could be a field in machine

learning with the flexibility of a pc to know, analyze, manipulate, and doubtless generate
human language.

2
Some options of NLP:

• Data Retrieval (Google finds relevant and similar results).

• Computational linguistics (Google Translate interprets language from one language to

another).

• Sentiment Analysis (Hater news provides U.S. the sentiment of the user).

• Spam Filter (Gmail filters spam emails separately).

• Auto-Predict (Google Search predicts user search results).

• Auto-Correct (Google Keyboard and grammatically correct words otherwise spelled

wrong).

• Speech Recognition.

(Natural Language Toolkit) NLTK: NLTK could be a well-liked ASCII text file package
in Python. NLTK provides all common natural language processing Tasks.

Thousands of text documents are processed for sentiment (and different options as well
as named entities, topics, themes, etc.) in seconds, compared to the hours it might take a
team of individuals to manually complete a similar task.

Thanks to sentiment analysis, corporations will perceive the name of their whole. By
analyzing social media posts, product reviews, client feedback, or NPS responses (among
different sources of unstructured business data), they'll bear in mind of however their
customers feel concerning their product. they'll additionally track specific topics and find
relevant insights on however folks are talking concerning those topics.

Sentiment analysis is especially helpful for social media observation as a result of it goes
on the far side metrics that concentrate on the quantity of likes or retweets, and provides a
qualitative purpose of read. Let’s say a corporation has simply launched a replacement

3
product feature and you notice a pointy increase in mentions on Twitter. However,
receiving plenty of mentions doesn't essentially mean a decent factor. Are customers
tweeting additional as a result of their expressing good items concerning this new product
feature? Or, are customers really whiney concerning the feature having many bugs?
Performing arts Twitter sentiment analysis is a wonderful thanks to perceive the tone of
these mentions and procure period of time insights on however users are perceiving your
new product.

1.2 Problem Statement:

The main goal of the project is to perform sentimental analysis on the tweets of a
particular user and i.e. determine whether the sentiments/feelings associated with a
particular tweet are positive, negative or neutral. Also, to perform various kinds of
graphical analysis in the data i.e. subjectivity, no, of likes, retweets etc.

1.3 Objectives:

The main objectives of this project embody crucial the feelings related to the varied
tweets and obtaining and basic graphical analysis of assorted tweet attributes over an
amount of your time. This could be useful in crucial the opinion of an outsized quantity
of individuals.

4
1.4 Methodology:

Methods of Sentiment Analysis:

• Information assortment

Consumers sometimes specific their sentiments on public forums just like the blogs,
discussion boards, product reviews also as on their personal logs – Social network sites
like Facebook and Twitter. Sentiments are asserted in numerous method, with totally
contrasting terminology and meaning of writing, creating the info immense and
scrambled. Manual analysis of sentiment information is nearly not possible. Therefore,
special programming languages like ‘R’ are accustomed method and analyze the info.

• Text Preparation

Preparing the text is however anyway separating the extricated data before examination.
It includes differentiating and deleting non-text content and content that is extraneous to
the world of study from the information that is present.

1. Analysing the sentiment

At this stage, every sentence of the review is checked for sound judgment basis.
Sentences with subjective expressions are kept and tweets that conveys objective
expressions are discarded.

2. Classification

Sentiments are broadly speaking classified into 2 things i.e positive and negative. At
this stage of sentiment analysis methodology, every tweet detected is assessed into
positive, negative and neutral.

5
 3. Results

The main plan of sentiment analysis is to convert JSON text into a pandas data frame.
On completing the project, we will visualize the data using matplotlib and will display
the sentiments.

1.5 Organization:

“Sentiment Analysis is outlined as a scientific analysis of on-line expressions “. There are

five steps to research sentiment information and here’s the graphical illustration of the
methodology to try and do constant.

Fig 1: Steps of sentiment analysis

6
 Chapter 2: Literature Survey

Here, in this part, we have examined about the different segments utilized in our venture
and we have referenced the separate research paper from where the content has been
cited. The different segments incorporate Opinion mining, Twitter, smaller scale
blogging, python, online life and ways to deal with wistful examination.

2.1 Opinion Mining

Assessment mining alludes to the expansive space of language process, content mining,
etymology, that includes the procedure investigation of estimations, sentiments and
feelings communicated in content [8]. In another view, read or point upheld feeling as
opposed to reason is normally conversationally named as a notion [8]. Subsequently,
transfer to steady for assessment mining or supposition examination. [9] express that
supposition mining has a few application areas just as bookkeeping, law, inquire about,
redirection, training, innovation, legislative issues, and selling. In prior days a few web
based life have provided internet clients path to distribute their position and feelings [10].

2.2 Twitter

Twitter might be a widely used ongoing social website that grants clients to express their
views alluded to as tweet that are restricted to one hundred forty characters [11]. Clients
compose tweets to communicate their sentiments about numerous things in regards to
their everyday life. Twitter is a stage for the retrieval of broad feelings on explicit point
[9,10]. Many of the tweets is utilized in light of the fact that the essential corpus for slant
investigation, that alludes to the use of assessment mining or language process [1].
Twitter, with enormous clients and million messages for each day, has immediately
turned into an important quality for links to investigate their name and types by removal

7
and tearing down the slope of the tweets by the entire group of people concerning their
product. [2] showed that, from the internet based life results in sentiments with the large
progress of the globe wide internet, big volume of conclusion messages inside the style of
tweets, audits, web journal or any talk group and gatherings are available for
consideration, hence making the globe wide net the fastest.

2.3 Microblogging with E-commerce

A stage like Twitter is similar to a standard online journal stage just single posts are short
[13]. Twitter has limited for a little scope of words that are meant for the short
transmission of info. or trade of supposition [7]. In any case, less business or large
association are commencement to the capability of microblogging as an internet business
selling device [3]. However, microblogging stage possesses been created numerous years'
energy for advancing remote exchange site by utilizing an outside microblogging stage as
Twitter selling [3].The moment of sharing, intelligent, network situated alternatives are
hole A web based business, propelled a {brand new} brilliant recognize that it are
regularly indicated that microblogging stage has empowered firms do mark picture, item
imperative deals channel, improve item deals, visit customer for a fair connection and
diverse business exercises concerned [2,3] [14]. [13] stated, truth be told, the
organizations delivering such product have begun to study big online journals to start a
method for regular interpretation for a product. Constantly these organizations render
user responses and answer to users on social sites [14].

2.4 Social Media

[15] Mentions an online life as an association of net-put together applications that make
with respect to the thoughtful and innovative establishments of Web2.0 that is permitted
to make and trade of client created substance. In an enormously exchange of Internet
beginning, realizes that a number of web clients is growing and kept on expanding with

8
online life by the whole time spent on smartphones and net based life in the America
across Personal computer hyperbolic by 37% to one hundred twenty-one billion minutes
in 2012, contrasted with eighty-eight billion min. in 2011. On the other side,
organizations utilize person to person communication locales to search out and speak
with buyers, business are frequently incontestable mischief to profitability brought about
by interpersonal interaction [17]. As media are frequently report so just to the general
population, it will harm individual information to showed inside the social world [11].
Despite what might be expected, [18] referenced that the upsides of participating in
online life have gone on the far side only social sharing to make association's name and
produce in profession openings and budgetary monetary profit. also, [15] referenced that
the web-based social networking is furthermore being utilized for advertising by firms for
advancements, experts for watching out, selecting, social learning on-line and electronic
trade. Electronic trade or E-business alludes to the procurement and closeout of items or
administrations on-line which may through online networking, such has Twitter that is
helpful in view of its 24-hours accessibility, simple customer administration and world
reach [19].
Among the clarifications of why business will in general utilize a great deal of internet
based life is for acquiring knowledge into customer conduct propensities, showcase
insight and blessing an opportunity to find out about customer survey and discernments.

2.5 Twitter Sentiment Analysis

The notions are frequently found inside the remarks or tweet to deliver supportive
pointers for a few very surprising capacities [20]. Additionally, [12] and [36] unequivocal
that a conclusion is frequently characterized into 2 groups, that is negative and positive
words. Assumption Analysis might be a language procedure strategy to measure a
communicated conclusion or sentiment inside an assortment of tweets [8]. Feeling
examination alludes to the last system to remove extremity and judgment from etymology
direction that alludes to the quality of words and extremity content or expressions [19].

9
There have been 2 fundamental methodologies for separating feeling precisely that are
the lexicon methods and Artificial Intelligence methods [19-23].

1. Lexicon based

Lexicon method assemble utilisation of already mentioned rundown of texts any place
each text is identified with a specific slant. The vocab ways are different as per the
settings during which they were made and include hard direction for an archive from the
etymology direction of writings or expressions inside the reports [19]. Also, [24] express
that a dictionary-supposition is to discover word-conveying conclusion inside the set thus
to foresee assessment communicated in the content. [20] has demonstrated the vocabulary
ways that have a fundamental worldview which are:

i. Pre-interpret each tweet, post by remove accentuation

ii. Introduce a total extreme score (s) equivalent zero - > s=0

iii. Check in the event that token is blessing in a very wordbook, at that point

If that token is sure, s will be certain (+)

Else if it is negative, s will be less than zero (-)

iv. Approve the entire polar value of twitter tweet.

On the off chance that s > edge, tweet post as positive

In the event that s < edge, tweet post as negative

1
Be that as it may, [21] featured one preferred position of inclining based philosophy, is
that it's the ability to adjust and make prepared models for explicit capacities and settings.
In differentiation, partner degree openness of named information and accordingly the low
importance of the methodology of most recent information that is cause naming
information might be costly or perhaps safeguard for a couple of errands [21].

2. AI based Approach

AI ways ordinarily regard administered arrangement approaches any place feeling

recognition is surrounded as a twofold that are sure and negative [24]. This approach
needs marked information to mentor classifiers [21]. This methodology, it becomes
evident that parts of the local setting of a word should be mulled over like negative (for
example Not wonderful) and increase (for example frightfully delightful) [19].

In any case, [20] demonstrated a fundamental worldview for produce an element vector
is:

i. Affix a locale of voice tag to each twitter tweet.

ii. Gather entire descriptor for whole twitter tweet.

iii. Fabricate a famous word combination made out of most noteworthy words.

iv. Explore the entirety of tweet inside training example to make the accompanying:

• Assortment of +ve texts

• Assortment of –ve texts

1
• Existence, nonattendance or recurrence of every word [19] gave some case of switch
nullification, refutation only to invert the extremity of the dictionary: regularly evolving
lovely (+3) into not staggering (- 3). extra models:

She isn't tremendous (6-5=1) anyway not horrendous (- 6+5=-1) either. For this situation,
the nullification of an intensely negative or positive value mirrors a blended point of view
that is appropriately caught inside the moved worth. Be that as it may, [21] has
referenced the restriction of AI based way to deal with be extra fitting for Twitter than the
lexical principally based approach.

Moreover, [20] communicated that AI ways will produce a set assortment of the chief
frequently occurring in style words that named a number cost for the benefit of the
recurrence of the word inside the Twitter.

2.6 Techniques of Sentiment Analysis

The etymology thoughts of substances separated from tweets is wont to live the general
connection of a lot of elements with a given assessment extremity [12]. Extremity alludes
to the chief fundamental sort, that will be that if a book or sentence is certain or negative
[25]. Be that as it may, supposition examination has strategies in conveyance extremity,
for example:

1. Natural Language Processing

NLP strategies are upheld AI and especially applied math discovering that uses a general
learning algorithmic guideline joined with an outsized example, a corpus, of information
to get familiar with the establishments [24]. Opinion examination has been taken care of
as an etymological correspondence process meant normal language handling, at a few
degrees of unpleasantness. Extending from being an archive level characterization task

1
[17], it's been taken care of at the sentence level [18] and extra as of late at the expression
level [13]. Normal language preparing might be a field in registering that includes
making PCs get that implies from human language and contribution as the most
straightforward method for interfacing with the $64000 world.

2. Case-based Reasoning

CBR is one in everything about procedures available to actualize conclusion

investigation. CBR is comprehended by reviewing the past with progress settled issues
and utilize similar answers for resolve this firmly associated issues. [15] known some of
the advantages of abuse CBR that CBR needn't bother with a specific area model then
incitement turns into an assignment of social occasion care chronicles and CBR
framework will learn by accomplishment new data as test cases. Furthermore, the
utilization of data procedures assembles the upkeep of enormous sections of information
simpler [15].

3.Artificial Neural Network

[13] Mentioned that ANN or alluded to as neural system might be a maths method that
joins and interleaves group of counterfeit neurons. It will strategy data abuse the
associations way to deal with calculation.

4. Support Vector

SVM is used to discover the feelings of twitter post. [10] communicated SVM is in a
situation to concentrate and investigate to get up to (70%-81.3%) of exactness upon
check case. [29] gathered instructing information from 3 totally unique Twitter notion
recognition sites that fundamentally utilize some pre-manufactured assessment
dictionaries to name each tweet as positive or negative. abuse SVM prepared from these

1
vociferous labelled information, they acquired eighty-one.3% in opinion grouping
precision.

2.7 Application Programming Interface

Speculative chemistry API achieves higher than the others as far as the standard and
furthermore the measure of the extricated elements. As time passed the Python Twitter
API is made by gathered tweets [30]. Python will precisely compute recurrence of texts
being retweeted every one hundred second, arranged the most noteworthy 200 texts
upheld there-tweeting recurrence, and keep them inside the chose data. Since the Python
TwitterAPI exclusively encased Twitter posts for the preeminent ongoing 6 days,
gathered the data required to be keep during a very surprising data [14].

2.8 Python

It was founded by Guido Van Rossum in Netherlands, 1989 that has been open in
1991.It’s a programing language that is open and settles a workstation drawback that is
giving a direct gratitude to work out an answer. [22] referenced that Python might be
referred to as a scripting language. Besides, [22] and [23] moreover upheld that really
Python might be a just portrayal of language because of it might be one composed and
run on a few stages. also, [24] referenced that Python might be a language that is pleasant
for composing a model because of Python is a littler sum time exceptional and managing
model gave, differentiation with various programming dialects.
Numerous specialists are voice correspondence that Python is practical, notably for an
extravagant venture, Python is fitting for interpersonal companies or news steaming
comes that most perpetually are an electronic which is driving a huge data. [34] gave the
clarification that because of Python will deal with and deal with the memory utilized.
Other than Python makes a generator that licenses partner tedious technique for things,

1
each thing in turn and grant program to snatch supply data each thing in turn to go each
through the complete procedure chain.

1
 Chapter 3: System Development

In this chapter we have given a basic idea about NLP which is the domain under which
our project lies. We have also mentioned about the language used and platform. We
discussed about the different modules in this project such as tweepy, numpy, pandas.

3.1 NLP

Natural Language process (NLP) is that the intersection of applied science, Linguistics
and Machine Learning that's attached the interaction between computers and humans in
tongue.

Fig 2: Venn diagram for NLP

NLP is far toward empowering PCs to grasp and deliver human idiom. Uses of
informatics systems are used in separating of text, machine interpretation and Voice
Agents like Alexa and Siri. Informatics is one amongst the fields that are profited from

1
the advanced methodologies in Machine Adapting, significantly from Profound Learning
methods.

Regular idiom making ready methodology utilize the characteristic dialect tool cabinet
for creating the principle organize in python tasks to figure with human dialect
information. this is often easier to-use by giving the interfaces to a minimum of one than
forty corpora and lexicon resources, for portrayal, for half passages sentences and to urge
the words in its distinctive frame Marking, parsing, and gloss thinking for current
reasoning quality basic idiom handling libraries, and for dynamic discourse. The NLTK
will utilize a huge instrument space and can create some facilitate for people with the
complete basic idiom taking care of system. this can assist people with “part sentences
from sections, to half up words, seeing the syntactical segments of these words, denoting
the elemental subjects, doing this it serves to your machine by acknowledging the most
factor to the substance.

3.2 Platform Used

Windows 10

Windows ten is outlined because the Microsoft that works with the actual framework for
PCs, tablets, and inserted gadgets etc. Microsoft discharged Windows ten is follow-up to
Windows eight. it had been aforesaid on Gregorian calendar month that the window ten
are invigorated rather than discharging it and framework as a successor.

When the window ten is chosen or received will be updated by inheritance squarely from
window seven, eight or window ten. While not activity meddling and also the framework
design methodology. For maintenance shoppers run the windows ten that helps in
exchanging the applying on the past software package and setting to window 10.
Shoppers pickup and fill or refresh window ten. With the assistance of window refresh
partner window ten will be redesigned to physically begin associate degree overhaul for
Windows.

1
Windows ten is employed to focus on add capacities that through which IT offices allows
to utilize mobile phones the board (MDM) programming to anchor and management
gadgets helps in running operating framework. For given boarding programming as an
example, Microsoft Framework Centre Arrangement Chief. Microsoft Windows ten is
employed for varied validation advances, as an example, good cards and tokens. Further,
Windows Hi has the biometric verification to Windows ten, wherever shoppers will sign
on with a novel finger impression, or facial acknowledgment.

The framework is employed to include virtualization-based security tools, as an example,

Secluded shopper Mode, Windows Safeguard Gizmo Watch and Windows shielded
Qualification Monitor. The Windows ten is employed to stay the highlights of explicit
data, procedures and shopper certifications separated attempting to resolve the matter
from any strike. Windows ten is newer version for Bit Locker secret writing to
substantiate data between clients' gadgets, reposting instrumentation, messages and cloud
administrations. Windows eight came up with the new plan and gave touch-empowered
motion driven UI like those on cell phones and tablets, however there wasn't abundant
interpretation of well to customary work space and digital computer PCs, significantly in
business settings. In Windows ten, Microsoft venture to deal with this issue and totally
different behavior of Windows eight, as an example, associate degree absence of massive
business neighborly highlights. The declaration of Windows ten in Gregorian calendar
month 2014 from Microsoft was created and window business executive was made that
point. There was the discharge from Microsoft to Windows ten by seeing the full
population in Gregorian calendar month 2015. then shoppers discovered that Windows
ten is cordial than Windows eight as a result of it had been additional typical interface,
that echoes the work space partaking format of Windows seven. The Windows ten
consecrate Refresh, that clad in August 2016, created some modifications to the
assignment bar and start Menu. It to boot bestowed program augmentations in Edge and
gave client’s access to Cortana on the bolt screen. In Apr 2017, Microsoft discharged the

1
Windows ten manufacturers Refresh, that created Windows hello facial acknowledgment
innovation faster and enabled shoppers to spare tabs in Microsoft Edge to examine later.

The Windows ten fall manufacturers refresh appeared in Gregorian calendar month 2017,
adding Windows Safeguard journey monitor to secure against zero-day assaults. The
refresh likewise enabled shoppers and IT to place applications running out of sight into
vitality productive mode to safeguard battery life and enhance execution.

3.3 Python

Python is termed as taken, object-oriented, high-level artificial language with dynamic

linguistics. The high-level is formed in information structures, together with dynamic
writing and binding, that helps in creating it terribly engaging for fast Application
Development. It helps in scripting languages wherever the parts are along. Python is
termed as terribly straightforward, straightforward to be told and has straightforward
syntax. It helps in reducing the value. Python supports packages which inspires program
code apply. The in depth commonplace library is out there in supply and might be freely
distributed.
Often, programmers like committal to writing in Python as a result of it provides
productivity. Written material testing and debugging cycle is extremely quick. Debugging
is extremely straightforward in python. Whenever the interpreter finds a slip it generates
the exception. If this doesn't happen then interpreter prints a stack trace. A supply level
computer program helps in examination of native and international variables, analysis of
capricious expressions, setting breakpoints, stepping through the code a line at a time,
and so on. The quickest thanks to rectify a program is by adding few print statements to
the supply.

1
3.4 Modules Used

3.4.1 Tweepy
Python is pleasant language for a wide range of things. Awfully dynamic engineer
network makes a few libraries that broaden the language and construct it simpler to
utilize fluctuated administrations. one in everything about libraries is tweepy. Tweepy is
publicly released, facilitated on GitHub and enables Python to talk with Twitter stage and
utilize its API.
Installing tweepy is straightforward and simple and can be installed from GitHub:

Fig 3: Installing tweepy

Either means provides you with the newest version.

3.4.1.1 Authentication Tweepy

Tweepy supports accessing Twitter via Basic Authentication and therefore the newer
methodology, OAuth. Twitter has stopped accepted Basic Authentication thus OAuth is
currently the sole thanks to use the Twitter API.

2
Here may be a sample of a way to access the Twitter API exploitation tweepy with
OAuth:

Fig 4: Accessing Twitter API using OAuth

The results of this code is:

Fig 5: Updated status using Twitter API.

The major distinction among Basic and OAuth authentication are the consumer
and access keys. You are only required to provide a username and a passkey for
accessing using basic authentication, but from 2010 since OAuth became

2
mandatory by twitter, the method has become a lot more cumbersome. A twitter
app must be created at dev.twitter.com.

OAuth could be a bit a lot of difficult at the start than Basic Auth entication,
since it needs a lot of effort, however the advantages it offers are terribly
lucrative:

• Tweets is tailor-made to acquire a string that recognizes the application

which was utilized.

• The user password is not provided making it safer.

• It's simpler to deal with the authorizations, for instance a gathering of

tokens and keys is created that exclusively allows perusing from the courses
of events, in this way just in the event that someone acquires those
certifications, he/she won't have the option to compose or send direct
messages, limiting the opportunity.

• The app does not answer on a secret word, therefore regardless of whether

the client transforms it or not the app will function.

Subsequent to signing in to the gateway, and navigating to Apps another app can
be made which is able to offer the required information for act with Twitter
API.

2
This is a screen that has all of the information required to speak to Twitter
network. it's vital to notice that by default, the app has no access to direct
messages, thus by attending to the settings and dynamical the suitable choice to
“Read, write and direct messages”, you'll be able to modify your app to possess
access to each Twitter feature.

Fig 6: Twitter application settings

2
3.4.1.2 Twitter API

Tweepy gives entry to the well mentioned Twitter API. With tweepy, it's
attainable to induce any object and use any methodology that the official
Twitter API offers. For instance, a User object has its documentation at
https://dev.twitter.com/docs/platform-objects/users and following those tips,
tweepy will get the suitable data.

Fig 7: Accessing user entities

Main Model categories within the Twitter API are Tweets, Users, Entities and Places.
Access to every returns a JSON-formatted response and traversing through data is
extremely straightforward in Python.

The above set of statements in figure 7 gives the following result:

Name: Robin Singh

Location: India

Friends: 55

2
3.4.1.3 Tweepy Streaming API

One of the most usage cases of tweepy is observance for tweets and doing
actions once some event happens. Key part of that's the Stream Listener object,
that monitors tweets in real time and catches them. Stream Listener has many
ways, with on_data() and on_status() being the foremost helpful ones. Here may
be a sample program that implements this behavior:

Fig 8: Stream Listener methods

So, this program contains a Stream Listener enforced and therefore the code is
about up to use OAuth. The Stream object is made, that uses that hearer as
output. Stream, being another necessary object in tweepy conjointly has several

2
ways, during this case filter() is employed with parameters passed. "Follow"
may be a list of followers whose tweets are monitored, and "track" may be a list
of hashtags which is able to trigger the Stream Listener.

In this example, we've got used my user ID to follow and therefore the
#pythoncentral hashtag as a condition. when running the program and tweeting
this status:

Fig 9: Tweeting a status

The program virtually instantly catches the tweet, and calls the on_status()
methodology, that produces the subsequent output within the console:

Fig 10: Result of on_status()

2
Besides printing the tweet, within the on_status() methodology there are some
further things that illustrate the amount of potentialities which will be finished
the tweet data:

Fig 11: Prints all the hashtag values

This code traverses through entities, picks the "hashtags" one and for every
hashtag the tweet contains, it prints its value. This is often simply a sample; a
whole list of tweet entities can be found on twitter’s website.

To sum up, tweepy may be a nice ASCII text file library that gives entry to the
TwitterAPI for Python. Despite the fact that the documentation for tweepy might
be somewhat rare and doesn't have a few models, the undeniable reality is that it
intensely relies upon the Twitter API, that has wonderful documentation, makes
it in all probability the most effective Twitter library for Python, particularly
once seeing the Streaming API help. Alternative modules similar to python-
twitter give several options also, however the tweepy has most alive association
and large carry out to the implementation in the previous years.

2
3.4.2 TextBlob

TextBlob is a python module and provides a simplistic API to use its methods and carry
out NLP tasks. TextBlob aims to provide access to common text-processing operations
through a familiar interface. You can treat TextBlob objects as if they were Python
strings that learned how to do Natural Language Processing.

A nice thing regarding TextBlob is that it is similar to strings. So, you can use it the same
way just as strings. Below, I have performed some of the simple tasks. The below code is
used to show that TextBlob is very same to string and the syntax is just to explain things
right.

Fig12: Similarity between TextBlob and Strings

2
3.4.2.1 Setting up TextBlob

Installation of TextBlob in your system in a simple task, all you need to do is

open anaconda prompt (or terminal if using Mac OS or Ubuntu) and enter the
following command:

pip install -U textblob

For the uninitiated – practical work in Natural Language Processing typically

uses large bodies of linguistic data, or corpora. To download the necessary
corpora, you can run the following command:

python -m textblob.download_corpora

3.4.2.2 Pros and Cons

Pros:
1. Since, it is built on the shoulders of NLTK and Pattern, therefore making it
simple for beginners by providing an intuitive interface to NLTK.
2. It provides language translation and detection which is powered by Google
Translate (not provided with Spacy).

Cons:
1. It is little slower in the comparison to spacy but faster than NLTK. (Spacy >
TextBlob > NLTK)
2. It does not provide features like dependency parsing, word vectors etc.
which is provided by spacy.

2
3.4.2 WordCloud

It is a visual representation tool used for describing text data wherein the size of
individual word tells about the significance of word. Huge literary information focuses
can be featured utilizing a word cloud. Word Clouds are generally utilized for examining
information from social networking websites. To generate Word Cloud in python we are
required to have matplotlib, pandas and WordCloud.

WordCloud can be a little tricky to install. If you only need it for plotting a basic
wordcloud, then “pip install wordcloud’’ would be sufficient. However, the latest version
with the ability to mask the cloud into any shape of your choice requires a different
method of installation as below:

Fig13: Commands to install Word Cloud

Benefits:
1. Assessing client and agent assessment.
2. Knowing about new SEO words to aim.

Disadvantages:
1. WordClouds are not suitable for every situation.
2. Information ought to be advanced for setting.

3
The following is an example of word cloud:

Fig14: Word Cloud of hobbies

This word cloud is formed by different hobby words. As you can see the word “music” is
the biggest in size hence it has been used most frequently as a hobby followed by movies,
reading and so on. The word music is appearing as most frequently but it could change
depending upon the dataset that the particular person is using.

3
3.4.3 Regular Expression(RegEx)

Regular expression is a set of special characters that is used to find a string or a substring
that is provided by the user in a pattern or similar kind of variable. I have used the sub
method of the re to clean up the tweets of unnecessary information and get only the
individuals words. Regular expression is mostly used in Unix environment.

The regular expression modules provide support for Perl like regular expression in

python. Whenever an error is present the re module raises an exception re.error.

3.4.3.1 RegEx Functions

The re module offers a set of functions that allows us to search a string for a match:

Table 1: RegEx functions

Function Description

findall Returns a list containing all matches

search Returns a Match object if there is a match anywhere in the

string

split Returns a list where the string has been split at each match

sub Replaces one or many matches with a string

3
3.4.3.2 Metacharacters

Metacharacters are characters with a special meaning:

Table 2: Metacharacters

Character Description Example

[] A set of characters "[a-m]"

\ Signals a special sequence "\d"

(can also be used to escape
special characters)
. Any character (except "he..o"
newline character)
^ Starts with "^hello"

$ Ends with "world$"

* Zero or more occurrences "aix*"

+ One or more occurrences "aix+"

{} Exactly the specified "al{2}"

number of occurrences
| Either or "falls|stays"

3
 Chapter 4: Performance Analysis

Here we have discussed about the extraction of tweets and storing the attribute fields of
the tweets into a data frame using pandas and numpy module. We have also applied
various mathematical functions to count the number of tweets, likes, retweets etc.

4.1 Tweets extraction

Now that we’ve created a perform to setup the Twitter API, we will use this perform to
make an “extractor” object. After this, we'll use tweepy’s command:

Fig 15: Extracting tweet count from user

This extracts from screen_name’s user the amount of tweets count.

As it is mentioned within the title, I’ve chosen @narendramodi as the user to extract
knowledge for a posterior analysis.

Fig 16: Extracting 5 recent tweets

3
Output:

Fig 17: Result of tweet count and extraction

The 5 recent tweets are:

Had a beautiful interaction with IFS officer trainees. We tend to mentioned a large vary
of problems. Urged the officers to… https://t.co/qtnNEHVSob

Sharing Associate in Nursing animated video that may teach you the ropes of
Shalabhasana. #FitIndia #4thYogaDay https://t.co/zHZn9MsUwG

आप सभी के साथ शलभासन का एक एननमेटेड वीनडयो शेयर

कर रहाहूं।
https://t.co/YG4DC5MrYK

Saddened by the loss of lives thanks to storms in some elements of the country.
Condolences to the bereft families. I p… https://t.co/YeJCdfv6he

Veena Abhyankar foreign terrorist organization from Pune wrote a beautiful letter to ME,
mentioning that she has been learning paper-quilling f… https://t.co/D5DOJ4Ts5K

3
4.2 Making a (pandas) Data Frame

We currently have initial data to construct pandas knowledge Frame, so as to control the
information in a very straightforward approach.

Python’s perform show plots Associate in nursing output in an exceedingly friendly

approach, and therefore the head methodology of an information frame permits us to
check the primary five components of the information frame (or the first variety of
elements that are passed as Associate in argument).

So, exploitation Python’s list comprehension:

Fig 18: Creating and displaying a data frame.

This will produce following result:

Tweets

0 Had a beautiful interaction with IFS officer t...

1 Sharing Associate in Nursing animated video that may teach you ...

2 आप सभी के साथ शलभासन का एक एननमेटेड वीनडयो शेय...

3 Saddened by the loss of lives thanks to storms in...

4 Veena Abhyankar foreign terrorist organization from Pune wrote a beautiful...

3
5 My Asian nation visit was historic. It gave ME a grea...

6 यस भ्रमण मार्फ़ त भारत-नेपाल सम्बन्धमा नयााँ उर् ...

7 मेरो नेपाल भ्रमण ऐनतहानसक रह्यो । यस भ्रमणले म...

8 At the programme in Kathmandu, I reiterated In...

9 I convey the folks of Kathmandu for the memora...

The fascinating half from here is that the amount of information contained in an
exceedingly single tweet. If we would like to get knowledge like the creation date, or the
supply of creation, we will access the information with this attributes. Let’s look at an
e.g.

Fig 19: Accessing the information inside a single tweet

3
Output:

Fig 20: Attributes of a tweets

Now, we try to find the most liked tweet and number of liked on that tweet. Also, we find

the number of characters contained in that particular tweet. Similarly, we also do this for

the most retweeted tweet.

3
Code:

Fig21: Most liked and most retweeted tweet.

Output:

Fig 22: Result of tweet analysis

3
4.3 Cleaning tweets

The method of re library that has been used for cleaning text is sub.

Fig 23: Syntax of substitute method

The sub method finds for the given pattern in the string that the user has given and if
there are any occurrences it is replaced by the substitute pattern. It then returns the
newly altered string.

To clean up the tweets of any hash tags, RT, @ mentions and hyperlinks we have
proceeded in the way shown below:

Fig 24: cleaning up the tweets

4
4.4 Calculation of Polarity and Subjectivity

I have used the TextBlob module to find the sentiment associated with the given tweet of
the respective user.

The TextBlob module has a sentiment function which return either the polarity or the
subjectivity. Polarity determines whether the statement is positive statement, negative
statement or neutral statement whereas subjectivity is personal opinion and objectivity is
factual information.

Polarity lies in the range of [-1,0,1] whereas subjectivity lies in the range of [0,1]. I have
implemented the below given code to successfully determine the polarity and
subjectivity.

Fig 25: Polarity and subjectivity methods

4
4.5 Visualizing the tweets

I have used WordCloud library to determine the most prominently used word in the given
set of tweets. Using WordCloud helps one to determine the most frequently used word in
the given set of text and enables the person to get a better look and understanding of the
vocabulary of the person who has written the sentence.

Fig 26: Generating WordCloud.

Herein, I have joined all the tweets under the Tweets column name in the data frame.
This forms the set of text which I have used for my WordCloud. Then, I have used the
generate method of the WordCloud library and passed the allWords variable to generate a
WordCloud.

4
 Chapter 5: Conclusion

Nowadays, sentiment analysis or opinion mining could be a hot topic in machine

learning. We have still so much to find regarding the feelings of corpus of texts terribly

accurately thanks to the complexity within the English language. In this project we are

tending to specialize in sentiments analysis. There is capability of labor within the range

of sentiment analysis with slightly accepted background. For eg. we tend to seen that

clients as a rule utilize our site for explicit sorts of watchwords which can be partitioned

into a couple of particular classes, to be specific: governmental issues/lawmakers, big

names, items/brands, sports person, media and music. Subsequently we will attempt to

perform separate feeling investigation on tweets that exclusively have a place with 1 of

those categories (for example the training data wouldn't be general anyway explicit to 1

of those classifications) and analyze the outcomes we tend to get if we apply general

sentiment analysis on that instead.

Twitter’s API is vastly helpful in data processing applications, and may offer large

insights into the general public opinion if the Twitter API and large information analytics

are a few things you've got more interest in. Twitter API can be used in most of the

difficult sentiment gathering, involving people, trends, and social graphs that is very

different for the human mind to get.

4
References

[1] M.Rambocas , and J. Gama, “Marketing Research: The Role of Sentiment Analysis”.
The 5th SNA-KDD Workshop’11. University of Porto, 2013.

[2] A. K. Jose, N. Bhatia, and S. Krishna, “Twitter Sentiment Analysis”. National

Institute of Technology Calicut, 2010.

[3] P. Lai, “Extracting Strong Sentiment Trend from Twitter”.

Stanford University, 2012.

[4] Y. Zhou, and Y. Fan, “ A Sociolinguistic Study of American Slang,” Theory and
Practice in Language Studies, 3(12), 2209–2213, 2013.
doi:10.4304/tpls.3.12.2209-2213

[5] M. Comesaña, A. P.Soares, M.Perea, A.P. Piñeiro, I. Fraga, and A.Pinheiro, “

Author’s personal copy Computers in Human Behavior ERP correlates of masked
affective priming with emoticons,” Computers in Human Behavior, 29, 588–595, 2013.

[6] A.H.Huang, D.C. Yen, & X. Zhang, “Exploring the effects of emoticons,”
Information & Management, 45(7), 466–473, 2008.

[7] D. Boyd, S. Golder, & G. Lotan, “Tweet, tweet, retweet: Conversational aspects of
retweeting on twitter,” System Sciences (HICSS), 2010 ….
Retrieved from http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=5428313

[8] T. Carpenter, and T. Way, “Tracking Sentiment Analysis through Twitter,”. ACM
computer survey. Villanova:VillanovaUniversity, 2010.

4
[9] D. Osimo, and F. Mureddu, “Research Challenge on Opinion Mining and Sentiment
Analysis,” Proceeding of the 12th conference of Fruct association, 2010, United
Kingdom.

[10] A. Pak,and P. Paroubek, “Twitter as a Corpus for Sentiment Analysis and Opinion
Mining,” Special Issue of International Journal of Computer Application,
France:Universitede Paris-Sud, 2010.

[11] S.Lohmann, M. Burch, H. Schmauder and D. Weiskopf, “Visual Analysis of

Microblog Content Using Time-Varying Co-occurrence Highlighting in Tag Clouds,”
Annual conference of VISVISUS.
Germany: University of Stuttgart, 2012.

[12] H. Saif, Y.He, and H. Alani, “Semantic Sentiment Analysis of Twitter,”

Proceeding of the Workshop on Information Extraction and Entity Analytics on Social
Media Data. United Kingdom: Knowledge Media Institute, 2011.

[13] A. Agarwal, B. Xie, I. Vovsha, O. Rambow, and R.Passonneau,

“Sentiment Analysis of Twitter Data,” Annual International Conferences. New
York:Columbia University, 2012.

[14] J. Zhang, Y. Qu, J. Cody and Y. Wu, “ A case study of Microblogging in the
Enterprise: Use, Value, and Related Issues,” Proceeding of the workshop on Web 2.0.,
2010.

[15] G. Kalia, “A Research Paper on Social Madia: An Innovative Educational Too”,

Vol.1, pp. 43-50, Chitkara University, 2013.

4
[16] Internet World Start, “Usage and Population Statistic”, Retrieved 10 15, 2013 from:
http://www.internetworldstats.com/stats.htm

[17] A.M. Kaplan, and M, Haenlein, “Users of the world, unite! The challenges and
opportunities of Social Media,” France: Paris, 2010.

[18] Q. Tang, B. Gu, and A.B. Whinston, “Content Contribution in Social Media: The
case of YouTube”, 2nd conference of social media. Hawaii: Maui, 2012.

[19] M.Taboada, J. Brooke, M. Tofiloski, K. Voll, and M. Stede, “ Lexicon- Based

Methods for Sentiment Analysis,” Association for Computational Linguistics, 2011.

[20] M. Annett, and G. Kondrak, “A Comparison of Sentiment Analysis Techniques:

Polarizing Movie Blogs,” Conference on web search and web data mining (WSDM).
University of Alberia: Department of Computing Science, 2009.

[21] P. Goncalves, F. Benevenuto, M. Araujo and M. Cha, “Comparing

and Combining Sentiment Analysis Methods”, 2013.

[22] E. Kouloumpis, T. Wilson, and J. Moore, “Twitter Sentiment Analysis: The Good
the Bad and the OMG!”, (Vol.5). International
AAAI, 2011.

[23] S. Sharma, “Application of Support Vector Machines for

Damage detection in Structure,” Journal of Machine Learning
Research, 2008.

[24] A.Sharma, and S. Dey, “Performance Investigation of Feature SelectionMethods and

Sentiment Lexicons for Sentiment Analysis,” Association for the advancement of
Artificial Intelligence, 2012.
4
 JAYPEE UNIVERSITY OF INFORMATION TECHNOLOGY,
WAKNAGHAT PLAGIARISM VERIFICATION REPORT
Date: ...............................
Type of Document (Tick): B.Tech Project Report

Name: Robin Singh Department: _C=S-=E'------- Enrolment No

161240 Contact No. _---=..7. : ::0. .:. 1.:::.8.:::.94. .:. 0=.:2:;0�8::.- E-mail. _. :. r-=-o-=bi:.:..;nc.:.1:....:2= 2=-
4 :.... : 9 :....:. 7 ...,, @=g .:...; . m .:..: : a ::. . : . ; i l . co:;..:: . ; ;.

; . m;
.______
Name of the Supervisor: __.,D""'r._,_.,_H.!.!,ailr.Li �Sin. !.: g:1-!-h.,_ ____
Title of the Thesis/Dissertation/Project Report/Paper (In Capital letters): TWITTER
SENTIMENT ANALYSIS

UNDERTAKING
I undertake that I am aware of the plagiarism related norms/ regulations, if I found guilty of any
plagiarism and copyright violations in the above thesis/report even after award of degree, the
University reserves the rights to withdraw/revoke my degree/report. Kindly allow me to avail
Plagiarism verification report for the document mentioned above.
Complete Thesis/Report Pages Detail:
- Total No. of Pages = 55
Digitally signed by
- Total No. of Preliminary pages = 9 Robin Singh
Robin
Date: 2020.07.14
- Total No. of pages accommodate Sing 21:12:31 +05'30'

bibliography/references = 3 h
(Signature of Student)
FOR DEPARTMENT
USE
We have checked the thesis/report as per norms and found Similarity Index at. (%). Therefore, we
are forwarding the complete thesis/report for final plagiarism check. The plagiarism verification report
may be handed over to the candidate.

(Signature of Guide/Supervisor) Signature of HOD

FOR LRC USE
The above document was scanned for plagiarism check. The outcome of the same is reported below:
Copy Received on
Excluded Similarity Generated Plagiarism Report
Index (%) Details
(Title, Abstract & Chapters)
• All
Word Counts
Report Generated Preliminary
on Pages
• Bibliography/ Character Counts
Submission
,. ' Ima ID Total Pages Scanned
· ges/Quotes
• 14 Words File Size
String

Signature
Checked by
Name &
Hari Singh
 Digitally signed by Hari Date: 2020.07.15 14:13:26 +05'30' Librarian
Singh

Please send your complete thesis/report in (PDF) with Title Page, Abstract and Chapters in (Word File)
through the supervisor at plagcheck,iult@gmail.com