Module Code & Module Title

CS6PO5NI Final Year Project

Assessment Weightage & Type

40% Final Project Report

Year and Semester

2022 Spring
Project Title: Sentimento (Social media assisting platform with sentiment analysis)

Student Name: Prashant Ghimire

London Met ID: 19031371
College ID: np01cp4a190008
Internal Supervisor: Bibek Khanal
External Supervisor: Subash Basnet
Assignment Due Date: 2022 April 27
Assignment Submission Date: 2022 April 26
Word Count: 8800 approx.

I confirm that I understand my coursework needs to be submitted online via Google Classroom under the
relevant module page before the deadline in order for my assignment to be accepted and marked. I am
fully aware that late submissions will be treated as non-submission and a mark of zero will be awarded.
 Acknowledgement

I would like to take this occasion to express my gratitude to Mr. Subash Basnet and Mr.
Bibek Khanal, our esteemed Project Supervisors, for their unwavering support and advice
throughout our final year project. Their continuous supervision and monitoring have been
really beneficial in completing this project.

I would also like to express my gratitude to our university and college for allowing us to
learn and work on a self-project. I am grateful to my helping teachers, faculty members,
and the entire Islington community for allowing us to grow with knowledge and wisdom.
 Abstract

The following report is a compilation of final year project work that tracks the project's
overall progress. The introduction, background information, development progress
documents, project outcome, and future possible engagement are all included in this
report. The research culminates in a fully functional mobile application that performs
sentiment analysis on YouTube comments, tweets, and custom texts. Spam and sarcasm
detection are also included for a more reliable analysis. Being a social media assisting
platform, there is also a feature for connecting job seekers with providers. This project
employs a Multinomial Naive Bayes Probabilistic Classifier. Flutter and Flask are utilized
to create a user-friendly application. The project's backend API is hosted on the Heroku
platform.

Keywords: Sentimento, Social Media Sentiment Analysis, Multinomial Naive Bayes,

Flutter, Flask
Table of Contents

Chapter 1: Introduction .............................................................................................. 1

1.1 Project Description ............................................................................................. 1

1.2 Problem Domain ................................................................................................ 2

1.3 Current Scenario ................................................................................................ 3

1.4 Project as a solution ........................................................................................... 4

1.5 Aims and Objectives .......................................................................................... 5

1.6 Structure of the report ........................................................................................ 6

1.6.1 Background ................................................................................................. 6

1.6.2 Development ............................................................................................... 6

1.6.3 Testing and Analysis ................................................................................... 7

1.6.4 Conclusion ................................................................................................... 7

Chapter 2: Background .............................................................................................. 8

2.1 About end/target users ....................................................................................... 8

2.2 Understanding solution .................................................................................... 10

2.2.1 Sentiment Analysis Part............................................................................. 11

2.2.2 Backend API Development ........................................................................ 15

2.2.3 Frontend Mobile Application Development ................................................ 15

2.3 Similar projects................................................................................................. 16

2.4 Comparison and uniqueness in my project ...................................................... 19

Chapter 3: Development .......................................................................................... 20

3.1 Considered Methodologies............................................................................... 21

3.2 Selected Methodology...................................................................................... 23

3.3 Phases of methodology .................................................................................... 25

 3.4 Survey Results ................................................................................................. 28

3.4.1 Pre-survey results ...................................................................................... 29

3.4.2 Post-survey results .................................................................................... 35

3.5 Requirement Analysis ...................................................................................... 41

3.6 Design and Logical Diagrams .......................................................................... 44

3.6.1 Entity Relationship Diagram ...................................................................... 44

3.6.2 System Architecture .................................................................................. 45

3.6.3 Overall Use Case Diagram ........................................................................ 46

3.6.4 Overall Context Level Diagram .................................................................. 47

3.6.5 Flowchart of sentiment analysis module .................................................... 48

3.6.6 Communication diagrams .......................................................................... 49

3.6.7 Sequence Diagrams .................................................................................. 51

3.6.8 Structure chart for deleting vacancy .......................................................... 54

3.6.9 Activity Diagram to view saved sentiment reports ..................................... 55

3.6.10 Data flow diagrams................................................................................. 56

3.7 Implementation following Scrum ...................................................................... 58

3.7.1 Sprint 1: 1st and 2nd Increment ................................................................... 58

3.7.2 Sprint 2: 3rd, 4th, 5th Increment ................................................................... 60

3.7.3 Sprint 3: 6th and 7th Increment ................................................................... 64

3.7.4 Sprint 4: 8th, 9th and 10th Increment ........................................................... 68

3.7.5 Sprint 5: 11th, 12th and 13th increment ........................................................ 70

3.7.6 Sprint 6: 14th and 15th increment................................................................ 72

Chapter 4: Testing and Analysis .............................................................................. 74

4.1 Test Plan .......................................................................................................... 74

4.1.1 Unit testing plan ......................................................................................... 74

 4.1.2 System testing plan ................................................................................... 77

4.2 Unit Testing ...................................................................................................... 78

4.3 System Testing .............................................................................................. 105

4.4 Critical Analysis .............................................................................................. 110

Chapter 5: Conclusion ........................................................................................... 111

5.1 Legal, Social and Ethical Issues..................................................................... 111

5.2 Advantages .................................................................................................... 112

5.3 Limitations ...................................................................................................... 114

5.4 Personal experience and Future work ............................................................ 115

5.4.1 Self-reflection or personal experience: .................................................... 115

5.4.2 Here’s how I will be further engaged on this project in future: ................. 115

Chapter 6: References ........................................................................................... 116

Chapter 7: Bibliography.......................................................................................... 119

Chapter 8: Appendix .............................................................................................. 123

Appendix 1: Additional Diagrams/Figures ............................................................. 123

i. Gantt chart ..................................................................................................... 123

ii. Work breakdown structure ............................................................................. 124

iii. Milestone chart............................................................................................ 125

iv. Layers in the designed sentiment analysis algorithm .................................. 126

v. Product Backlog .......................................................................................... 127

vi. Sprint Backlogs ........................................................................................... 128

Appendix 2: Reasons to use consumed libraries/framework ................................ 134

i. Using BLOC for state management ............................................................... 134

ii. Using Flask over fully featured framework like Django ................................... 134

iii. Multinomial Naive Bayes algorithm over available package like TextBlob .. 134
Appendix 3: Software Requirement Specification ................................................. 135

Appendix 4: User Feedbacks ................................................................................ 144

i. Google Play Store Reviews............................................................................ 144

ii. User Feedback Survey Report ....................................................................... 145

Appendix 5: Weekly/Increment task information for development......................... 148

Appendix 6: Essential code snippets .................................................................... 150

i. Backend ......................................................................................................... 150

ii. Frontend ......................................................................................................... 157

List of Figures

Figure 1: Snapshots of built mobile application ............................................................... 1

Figure 2 Use case scenario of sentiment analysis (Liu, 2016) ........................................ 3
Figure 3 Type of potential users of the developed platform............................................. 8
Figure 4 Built solution work breakdown ......................................................................... 10
Figure 5 General approach of sentiment analysis (Binita Verma, 2018) ....................... 11
Figure 6 Various methods of sentiment analysis (Walaa Medhat, 2014) ...................... 12
Figure 7 Bayes rule with labelling (Jha, 2017)............................................................... 14
Figure 8 Sentiment analysis outcome between Taylor Swift and Jake Paul.................. 16
Figure 9 Analysis of tweets on NEPSE ......................................................................... 17
Figure 10 Searching social media related jobs in LinkedIn ........................................... 18
Figure 11 Scrum methodology (scrum.org, 2020) ......................................................... 23
Figure 12 Sample of daily scrum board ......................................................................... 26
Figure 13 Social media user’s familiarity with sentiment analysis ................................. 29
Figure 14 User's awareness on precise social media activity........................................ 30
Figure 15 Chart showing user's interest to comment .................................................... 31
Figure 16 Chart showing people's interest to tweet ....................................................... 32
Figure 17 Chart showing user's interest to use sentiment tool ...................................... 33
Figure 18 Chart showing people's interest looking for LinkedIn alternatives ................. 34
Figure 19 Chart showing social platform demand for analysis ...................................... 35
Figure 20 Chart showing user's interest to use their own third-party credentials........... 36
Figure 21 Chart showing user's login preferences ........................................................ 37
Figure 22 Chart showing what sentiment labels user want ........................................... 38
Figure 23 Chart showing user's interest on report saving feature ................................. 39
Figure 24 Chart showing user's opinion on limiting vacancy post at a time ................... 40
Figure 25 ERD of Sentimento's database ..................................................................... 44
Figure 26 System architecture of Sentimento platform.................................................. 45
Figure 27 Use Case Diagram of overall system ............................................................ 46
Figure 28 Context level diagram ................................................................................... 47
Figure 29 Flowchart of sentiment analysis module ....................................................... 48
Figure 30 Communication diagram for performing sentiment analysis .......................... 49
Figure 31 Communication diagram for posting vacancy................................................ 50
Figure 32 Sequence diagram for logging in user ........................................................... 51
Figure 33 Sequence diagram for tweet sentiment analysis ........................................... 52
Figure 34 Sequence diagram for viewing freelancers ................................................... 53
Figure 35 Structure chart for deleting vacancy .............................................................. 54
Figure 36 Activity diagram to view saved reports .......................................................... 55
Figure 37 Level 0 DFD for login/registration .................................................................. 56
Figure 38 Level 1 DFD for login/registration .................................................................. 56
Figure 39 Level 2 DFD for login/registration .................................................................. 57
Figure 40 Initial user authentication interface ................................................................ 58
Figure 41 Change in meaning of a sentence while removing stop words...................... 59
Figure 42 Function to clean sentence ........................................................................... 60
Figure 43 Function to make updates in serialized training model .................................. 61
Figure 44 Function to predict probability of each sentence to fall in 0 or 1 category ..... 62
Figure 45 Single class to access all algorithmic feature of system ................................ 63
Figure 46 Function to extract tweets of chosen topic .................................................... 64
Figure 47 Function to extract YouTube comments of chosen video .............................. 65
Figure 48 Function to handle login request ................................................................... 66
Figure 49 User login request task ................................................................................. 67
Figure 50 API handling sentiment analysis request on chosen YouTube video ............ 68
Figure 51 Viewing user's saved sentiment reports ........................................................ 69
Figure 52 Git commit showing front end user authentication pages task done ............. 69
Figure 53 Bloc state management architecture (Aqeel, 2020) ...................................... 70
Figure 54 Git commit showing tasks of fifth sprint ......................................................... 71
Figure 55 API hosted in Heroku platform ...................................................................... 72
Figure 56 Application published to Google Play Store .................................................. 73
Figure 57 Trying to register with already used email ..................................................... 78
Figure 58 Trying to login without being registered ......................................................... 79
Figure 59 Registering user in the platform .................................................................... 80
Figure 60 Logging inside the platform ........................................................................... 81
Figure 61 Trying to log in with incorrect password ........................................................ 82
Figure 62 Showing profile icon to enter profile page ..................................................... 83
Figure 63 User navigated to profile screen after loading completes .............................. 84
Figure 64 Trying to perform sentiment analysis without providing tokens ..................... 85
Figure 65 Entering coupons and video information ....................................................... 86
Figure 66 Showing sentiment report after loading completes ........................................ 87
Figure 67 Polarity check of manual sentence ................................................................ 88
Figure 68 Providing invalid coupon ............................................................................... 89
Figure 69 Trying to perform analysis using invalid coupon............................................ 89
Figure 70 Toast message about invalid request ............................................................ 89
Figure 71 Saved sentiment reports of user ................................................................... 90
Figure 72 Trying to view reports without performing any analysis ................................. 91
Figure 73 Dashboard in portrait view ............................................................................ 92
Figure 74 Dashboard in landscape view ....................................................................... 93
Figure 75 Posting vacancy in the platform .................................................................... 94
Figure 76 User having five open vacancies trying to add another new vacancy ........... 95
Figure 77 User navigated to phone call window ............................................................ 96
Figure 78 Deleting previously added post ..................................................................... 97
Figure 79 Logging out from the platform ....................................................................... 98
Figure 80 Turning off Wi-Fi and Mobile data ................................................................. 99
Figure 81 Trying to view reports without internet connection ........................................ 99
Figure 82 Dashboard after logging in .......................................................................... 100
Figure 83 Dashboard after opening for second time ................................................... 100
Figure 84 Trying to access profile without providing authentication header ................ 101
Figure 85 Viewing vacancies information from web browser....................................... 102
Figure 86 Trying to delete another user's vacancy ...................................................... 103
Figure 87 Trying to perform sentiment analysis on more than 1000 data.................... 104
Figure 88 Checking hosted backend API on Heroku ................................................... 105
Figure 89 Count before performing analysis ............................................................... 106
Figure 90 Performing polarity check ............................................................................ 106
Figure 91 Count after performing analysis .................................................................. 106
Figure 92 Performing sentiment analysis on tweets .................................................... 107
Figure 93 Viewing recently saved report ..................................................................... 108
Figure 94 Flutter Application Compilation.................................................................... 109
Figure 95 Distinct use cases of sentiment analysis (Samadhan Engineering, 2021) .. 112
Figure 96 Snapshot of a LinkedIn learning course for Flask ....................................... 116
Figure 97 Snapshot of Flask documentation web page .............................................. 117
Figure 98 Snapshot of Flutter package manager ........................................................ 117
Figure 99 Snapshot of scikit-learn library .................................................................... 118
Figure 100 Gantt Chart of the project work division .................................................... 123
Figure 101 Work breakdown structure of overall project ............................................. 124
Figure 102 Milestone chart of overall project............................................................... 125
Figure 103 Layers showing how sentiment analysis works ......................................... 126
Figure 104 Product backlog design ............................................................................. 127
Figure 105 Sprint 1 Backlog ........................................................................................ 128
Figure 106 Sprint 2 Backlog ........................................................................................ 129
Figure 107 Sprint 3 Backlog ........................................................................................ 130
Figure 108 Sprint 4 Backlog ........................................................................................ 131
Figure 109 Sprint 5 Backlog ........................................................................................ 132
Figure 110 Sprint 6 Backlog ........................................................................................ 133
Figure 111 User feedback from Google Play Store reviews ........................................ 144
List of Tables

Table 1 Comparison of other projects and sentimento .................................................. 19

Table 2 General principles of SDLC (Prokopisko, 2019) ............................................... 20
Table 3 Survey volunter's information ........................................................................... 28
Table 4 Functional requirements ................................................................................... 42
Table 5 Unit testing for Flutter application ..................................................................... 75
Table 6 Unit testing for Flask API .................................................................................. 76
Table 7 System testing for built platform ....................................................................... 77
Table 8 Unit test 1 ......................................................................................................... 78
Table 9 Unit test 2 ......................................................................................................... 79
Table 10 Unit test 3 ....................................................................................................... 80
Table 11 Unit test 4 ....................................................................................................... 81
Table 12 Unit test 5 ....................................................................................................... 82
Table 13 Unit test 6 ....................................................................................................... 83
Table 14 Unit test 7 ....................................................................................................... 85
Table 15 Unit test 8 ....................................................................................................... 86
Table 16 Unit test 9 ....................................................................................................... 88
Table 17 Unit test 10 ..................................................................................................... 89
Table 18 Unit test 11 ..................................................................................................... 90
Table 19 Unit test 13 ..................................................................................................... 92
Table 20 Unit test 14 ..................................................................................................... 94
Table 21 Unit test 15 ..................................................................................................... 95
Table 22 Unit test 16 ..................................................................................................... 96
Table 23 Unit test 17 ..................................................................................................... 97
Table 24 Unit test 18 ..................................................................................................... 98
Table 25 Unit test 19 ..................................................................................................... 99
Table 26 Unit test 20 ................................................................................................... 100
Table 27 Unit test 21 ................................................................................................... 101
Table 28 Unit test 22 ................................................................................................... 102
Table 29 Unit test 23 ................................................................................................... 103
Table 30 Unit test 24 ................................................................................................... 104
Table 31 System test 1................................................................................................ 105
Table 32 System test 2................................................................................................ 106
Table 33 System test 3................................................................................................ 107
Table 34 System test 4................................................................................................ 108
Table 35 System test 5................................................................................................ 109
Table of Abbreviations

NLTK: Natural Language Toolkit

API: Application Programming Interface

BLOC: Business Logic Component

CRUD: Create Read Update Delete

ML: Machine Learning

AI: Artificial Intelligence

SDLC: Software Development Life Cycle

FR: Functional Requirement

SR: System Requirement

ERD: Entity Relationship Diagram

DFD: Data Flow Diagram

HTTP: Hypertext Transfer Protocol

Git: Global Information Tracker

UI: User Interface

UT: Unit Testing

ST: System Testing

SRS: Software Requirement Specification

Sentimento: Name of the application/platform of this project

CS6PO5NI FINAL YEAR PROJECT

Chapter 1: Introduction

1.1 Project Description

Sentimento, a social media assisting platform with sentiment analysis feature is a

mobile application built using Flutter for front end, Flask for back end and Multinomial
Naive Bayes Theorem for the algorithmic part.

Figure 1: Snapshots of built mobile application

This application is built for folks who need social media in their professional lives in some
way. Users can perform sentiment analysis on chosen YouTube video comments and
tweets using the developed platform. This platform allows social media-related job
recruiters and job seekers to connect with one another.

1
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

1.2 Problem Domain

The comments, feedback and public opinion on social medias can be really
applicable for various purposes to companies, content creators and for similar
professions. User’s opinion data is available on the internet but there is lack of optimal
utilization of it. The prime difficulties I found in the social media engaged professionals
are:

i. The feedbacks of any range of customers can be truly important to sharpen the
qualities to a company, content creators and similar professionals but the lack of
proper tools, the power of feedbacks cannot be optimally utilized (John Hattie,
2007).

ii. The Online Labour Index (OLI) provides data related to gig economy (freelancers).
In 2017, the index reported that skills from diverse ranges were growing from all
around the world where USA was ahead for writing/translation and India was
ahead for software development and technology. Even with the availability of
multiple job searching platforms, the freelancers engaged in social media works
were found to have difficulties in connecting with their specific type of jobs (Rajat
Kathuria, 2017).

2
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

1.3 Current Scenario

In the early days of telecommunication, digital communication was a means of

luxury to the very specific numbers of people. Coming to the present days, Social Medias
have established themselves as a necessity in every day human’s life. The social media
penetration took place instantaneously within a generation period. With the advancement
of social communication technology, concepts such as mass sentiment analysis, targeted
advertising, digital economy concepts like Metaverse are also growing in a consistent and
with well-grounded pace.

Figure 2 Use case scenario of sentiment analysis (Liu, 2016)

Contemplating the potential use cases of strong human presence in social medias and
my genuine interest in this, I was self-motivated to build my final year project and future
career in this sector. Sentimento is a portable platform or mobile application for
professionals who are engaged in any form of social media work. It is built primarily for
social media influencers and digital content creators who has presence in the social
media in a professional life. The professionals engaged in digital marketing/advertising,
mass psychology, political campaigns and similar field professionals can be benefited
from my built platform.

3
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

1.4 Project as a solution

Sentimento, the platform I have developed is a tool for professionals who are
engaged in multiple forms of social media works like content creators, freelancers,
marketing personals, etc. The solutions that I have come up with to the previously stated
challenges are as follows:

i. To keep track of audience feedbacks, content providers can't look through every
single comment on their YouTube videos. This takes a long time and is frequently
unfeasible. The Sentimento platform may be used to scan comments in bulk and
content creators can be provided with a summary feedback report.

ii. Not every company or workforce affords expensive cost of data science team, the
platform can be utilized by marketing and advertisement agencies to find what
people are saying to particular things. At this stage of prototype level, YouTube
comments and Twitter tweets are available as means of data.

iii. Freelancers related to social media work can be connected with relevant work
vacancies in the platform. I considered this feature to include to solve the second
problem I mentioned in previous heading.

iv. Political Campaigns or similar events has their own objectives. To monitor what
people are saying about the particular event can be monitored from the platform
using tweets sentiment analysis to specific hashtag/topic.

4
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

1.5 Aims and Objectives

Developing a prototype level platform for social media professionals where

sentiment analysis can be performed and connecting freelancers with relevant work
vacancies is the main aim of this project.

The activities of my involvement or the objectives to fulfil the aim of the project are as
follows:

i. To learn about Sentiment Analysis in Natural Language Processing which will be

helpful for me to study Data Science after Bachelors level.

ii. To learn Flutter, cross-platform application development framework which is also

the interest of mine to engage in application development field.

iii. To learn about backend and REST API development using Python’s Flask.

iv. To research and self-study about how can I implement the sentiment analysis part
in my project.

v. To perform additional research on documentation part for authentic facts/data and

tackle difficulties I may face on development part.

vi. To implement user interfaces and front-end development works on Flutter that I
learned from my internship period.

vii. To carry out database related works in real project that we have been learning from
first year of the college.

5
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

1.6 Structure of the report

In this section, the chapters of the report are shortly introduced.

1.6.1 Background

This section of the documentation is devoted to a literature review. It includes information

about other similar projects, as well as details on the differences and improvements made
to the project. The mobile application is created to meet the needs of the possible
consumers of the generated application/solution. This section of the report goes over the
algorithmic part of the sentiment analysis feature utilizing Multinomial Nave Bayes in
detail. The backend and frontend development work structures are also documented to
ensure that the whole system is properly understood.

1.6.2 Development

This section of the report documents the technical work performed during the
development of this project, including system design, database and system architecture,
frontend and backend development, software development methodology. This section of
the report also includes a survey done at various stages of project development, as well
as diagrammatic representations of various components of the solution. The codes for
various sections of the development, such as the algorithmic component, the
backend/frontend side, and the cloud hosting work, are also attached here with sufficient
elaboration.

6
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

1.6.3 Testing and Analysis

This section of the project documentation offers a summary of the testing procedures for
the created system's various functionalities. The testing methods or plans for completing
system and unit testing are first thoroughly outlined. Under system testing, each of the
developed system's essential features is tested. Under unit testing, the smaller logical
functions or the interface with which the user interacts are also tested. The testing
procedure includes tabular data and screenshots of the interface depending on the
expected and actual/output results.

Following the testing, there is a critical analysis portion in which the overall testing,
whether the development went as intended, and other components of the development
work are critically examined, as well as personal experience.

1.6.4 Conclusion

The report's final but crucial section offers information on the project's legal, social, and
ethical difficulties. This section also includes the project's various use cases/benefits as
well as its limitations. This section of the documentation work also discusses additional
work that can be done to improve the project, how further studies in this topic can be done
in further studies of my academic and career pathway.

7
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Chapter 2: Background

2.1 About end/target users

Sentimento, a social media assisting platform, can be used by various social media
related professionals. Primarily, we can categorize the end users based in following tree
structure:

Figure 3 Type of potential users of the developed platform

8
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Specific client of sentimento platform

The platform is designed and developed to fit all the previously mentioned potential
users. It clarifies that this project is not made for any particular client. The platform is
developed keeping in mind to fulfil the professional needs of potential users. The user
interface and database architecture are structured for all the general users of the platform.

Because this is a prototype mobile application, no specific clients or corporations have

been approached to create a toolkit for a specific company. However, with a few tweaks,
the sentiment analysis tool may be tailored to meet the specific and detailed demands of
any given client.

9
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

2.2 Understanding solution

Figure 4 Built solution work breakdown

The project was finished with the help of a variety of tools and technologies that were
already available in the community. The total solution is broken into numerous sections
in this part, and each module is well explained.

10
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

2.2.1 Sentiment Analysis Part

Generic Introduction to Sentiment Analysis

Sentiment Analysis in the context of computer science is the phenomenon of

extracting human sentiment from the provided texts. It is the concept of mining human
opinion regarding particular topic, event, individual/party, issues or similar aspects.
Sentiment/Opinion Analysis is a part of Natural Language Processing. The outcome of a
sentiment analysis project is a system which helps to understand the human sentiments
like positive, negative and neutral expressions regarding any topic (Ludvig Persson,
2018).

Figure 5 General approach of sentiment analysis (Binita Verma, 2018)

The above diagram depicts a graphical summary of the sentiment analysis workflow. The
flow of the process is depicted in the diagram above, from the beginning phase of
gathering data to the processing phases and finally to the finalization of sentiment scores.
11
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Distinct approaches of performing sentiment analysis:

There are multiple approaches of performing sentiment analysis. Depending on
the use cases, either machine learning approaches or lexicon-based approaches can be
implemented. More details on these distinct approaches are described below.

Figure 6 Various methods of sentiment analysis (Walaa Medhat, 2014)

Lexicon-based Approach
In this approach, a collection or dictionary of positive and negative labelled words are
required. For calculating the overall sentiment/polarity of the data, a function to sum the
average polarity of words is defined. The final average polarity is determined as the
overall sentiment (positive or negative) score for a sentence or a piece of testing data
(Mulvenna, 2015).

12
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Machine Learning Approach

Supervised and Unsupervised learning are two categories in machine learning

approach. In unsupervised learning approach, a model or machine is trained consuming
the unclassified/unlabelled information. The developed algorithm is solely responsible to
act on those unclassified data without guidance. The model classifies the information
based on the observed similarities, differences and patterns from the provided
information. Clustering and Association are two categories of unsupervised learning.
Clustering is about finding patterns in collection of unlabelled data. For finding the relation
or link among data in large unclassified data, association techniques are used (Shimon
Ullman, 2014).

In supervised learning, there are multiple classification techniques like Decision Tree
Classifier, Linear Classifier, Rule Based Approach, Probabilistic Classifier. Each of these
classification approaches have their own certain variations and multiple algorithms.
Decision Tree Classifier can be used for both classification and regression problems. It is
a tree-structured classifier where there are decision node and leaf node. Decision nodes
are used to perform decision, have multiple branches whereas leaf nodes are output of
those decisions (Lior Rokach, 2015).

Naive Bayes is popular probabilistic classification technique derived from Bayes

Theorem. This theorem is a mathematical formula used for calculating conditional
probabilities. Naive Bayes Classifier is easy and simple to implement. It can work with
less amount of training data too. It is fast and can be used for real-time predictions
(Stanford Edu, 2009).

13
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Figure 7 Bayes rule with labelling (Jha, 2017)

Elaborating the above rule with an example

Let’s assume ‘c’ as shopping and ‘x’ as raining. The posterior probability denotes
the conditional probability of shopping when its already raining. P (x | c) denotes
probability of occurrence of rain while shopping. P (c) denotes probability of raining.
When there is occurrence of discrete features (e.g., word counts for text classification),
the multinomial naive bayes classifier is suitable option. It requires integer feature counts
and based on frequency of the features and its label or classification. It can be used in
multinomially distributed data (Scikit-learn, 2021).

14
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

2.2.2 Backend API Development

Flask, a micro web framework has been utilized for the API development of the
sentimento project.

Reasons to use Flask over other backend tools for this project:
i. Flask is lightweight, only useful modules to our system can be imported, less space
consuming comparing to Django or other full-featured framework.
ii. Previous experience of working with Flask framework.
iii. Only API development is necessary with no need of user-interacting web pages.
So, there is no need to use web page templates. In such case, Flask is one of the
best tools.

2.2.3 Frontend Mobile Application Development

Flutter, cross platform framework is used for the user interface development work
for the project.

Reasons to use Flutter instead of React Native for this project:

i. Previous experience of working with Flutter during internship.
ii. Easy to learn if there is clear understanding of object-oriented programming as
Flutter is focused on widget-based interface development.
iii. React Native uses middleware to translate developer’s written JavaScript code to
native components. Such code translation can create difficulty for smooth
interaction (run time animations) on user interfaces. On the other hand, Flutter
compiles code with support of C++ engine and with native iOS and Android
platform. This makes Flutter to contain lesser performance issues. (Doshi, 2021)
In Sentimento project, there is possibility to render user’s saved reports in larger
quantity. For such purpose, Flutter is suitable on this project.

15
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

2.3 Similar projects

Similar types of projects are sought from the project's research stage to
requirement analysis and development, and inspiration is collected to produce a better
solution for Sentimento's project.

i. TSentiment (Mobile Application for tweet sentiment analysis)

It is an iOS application for tweets analysis based on hashtag. User has to choose
a topic to which analysis is performed. The tweets analysed and percentage distribution
of positivity/negativity found is shown in this platform (C9 Apps Desenvolvimento de
Software Ltda ME, 2017).

Figure 8 Sentiment analysis outcome between Taylor Swift and Jake Paul

Publisher of the application: C9 Apps Development de Software Ltd. ME

Platform: iOS and MacOS

16
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

ii. Sentiment Analysis of Twitter (Mobile Application)

It is an android application for tweets analysis. Users can choose the topic and
number of total tweets they want to analyse. The type of visualization output can also be
chosen whether they want to view in graph or pie-chart. There is option to choose
between polarity visualization and emotion. There is also feature to show recent searches
by a user. Maximum 100 recent tweets would be analysed and tweets would be 7 days
old. The application works with Naive Bayes Classifier and generates the sentiment in
positive and negative polarity (Jay Khatri, 2020).

Figure 9 Analysis of tweets on NEPSE

Developers: Jay Khatri, Isha Khimsurya

Platform: Android

17
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

iii. LinkedIn (Professional Networking Platform)

It is a successful and well-established platform which serves in professional

networking, job portals and similar services. The vision of this platform is to create
economic opportunity for every member of the global workforce (LinkedIn, 2021).

Figure 10 Searching social media related jobs in LinkedIn

LinkedIn is considered as similar projects to mine because there will be social media
related job networking feature in my ‘Sentimento’ platform.

18
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

2.4 Comparison and uniqueness in my project

S. N Comparison topic Project 1 Project 2 Sentimento

1 User Authentication System ✖ ✔ ✔
2 YouTube Comments Analysis ✖ ✖ ✔
3 Tweets Analysis ✔ ✔ ✔
4 Custom Sentence/Paragraph Analysis ✖ ✖ ✔
5 User’s defined maximum data to pick ✖ ✔ ✔
6 Report Saving Feature ✖ ✖ ✔
7 Custom API Keys ✖ ✖ ✔
8 Social Media Job Networking ✖ ✖ ✔
Table 1 Comparison of other projects and sentimento

According to the information supplied in the Google Play application description, the
second project uses the Naive Bayes Classifier. Multinomial Naive Bayes is used in my
research to classify data as sarcastic or not. The frequency of an event occurring is
represented by Multinomial Nave Bayes. Because the goal is to increase the accuracy of
the analysis, improvements to the algorithm may be made in future sprints. However, the
Naive Bayes algorithm is set to be used in the classification section.

Personal view on how the developed solution is better than similar projects

Sentimento platform isn't just for analysing tweets. YouTube comments and custom
sentence/paragraphs are also supported. The sentiment report saving function allows
users to access reports in the future and compare sentiment discrepancies between past
and present sentiment changes on the same topic.

19
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Chapter 3: Development

Software Development Life Cycle (SDLC):

It is a defined collection of rules/protocols for the creation of production-ready or

large-scale projects. Following any software development life cycle, the project's integrity,
general documentation, and security are the critical aspects for project success. Following
software development principles, the project involves technical, management, finance,
and every department of personnel (Young, 2013).

Table 2 General principles of SDLC (Prokopisko, 2019)

Software Development can be carried out following several methodologies based on each
project’s requirement and conditions. Waterfall, Agile, Rapid Application Development are
popular methodologies. These methodologies have smaller branches with certain
differences.

20
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.1 Considered Methodologies

Waterfall
It is a linear and traditional approach of software development. Each phase of process is
carried out in sequential order. Next task is not executed without finishing the previous
one (Young, 2013).

Good points to considerate:

i. The straight goal focused approach makes the project well planned.
ii. Minimizes cost for changing requirements and project is estimated to be well
documented.

Bad points to considerate:

i. The initial task plan limits further feature to implement in the middle of
development.
ii. The poor flexibility in making changes make essential new updates to come very
late in the production. This is not feasible for current market scenario where rapid
changes are essential to make if needed.

Spiral Model
This model is appropriate for unique projects where there is high risk of project failure.

Good points to considerate:

i. Better for complicated projects where task to accomplish is very less clear.
ii. Strong documentation control over approval of development stages.

Bad points to considerate:

i. Not suitable for project with smaller scope and low risk projects.
ii. End of the project is difficult to estimate.

21
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Unified Process
It is an interactive approach of development for larger projects with bureaucratic
development teams. The principles of this methodology stand between waterfall and agile
principles.

Good points to considerate:

i. Highest risk containing tasks are prioritized first to maintain early break point. This
allows easy cancellation of project in case of high failure possibility.
ii. Performing several tasks like documentation, development and testing in parallel,
labour pool can be utilized efficiently (Young, 2013).

Bad points to considerate:

i. Documentation, artifact making is heavier than Agile.
ii. The system to be developed for this project is moderate with less complexity, in
such case, agile is more appropriate than unified process.

Agile
Agile principles are very adaptive towards change in requirements and plans. The minimal
cost and quality software are the key principles of agile methodology. It is similar to unified
process but iterations are much short period of time in agile.

Good points to considerate:

i. The agile frameworks bring strong bond in team work as it promotes fail-early-fail-
fast approach in the work. This empowers team to be self-managed.
ii. The quality of the product is significantly better developed with agile frameworks
as it focuses result over excessive documentation.

Bad points to considerate:

i. The lack of work-in-progress control can result too many running tasks. This can
generate long queues (Perisic, 2014).
22
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.2 Selected Methodology

The Scrum framework of agile methodology will be utilized for this project because
it is appropriate for this project type. Agile characteristics such as flexible work divisions
for individuals, working software over exhaustive documentation, and adapting to
changes are all part of the Scrum methodology. Each sprint of the scrum is dedicated to
completing specific tasks.

Figure 11 Scrum methodology (scrum.org, 2020)

The Scrum methodology is about carrying out the project based on multiple sprints. In
each sprint, we carry out certain feature. Each sprint can be of one to two week long. The
tasks to be done, the tasks to be reviewed are maintained in scrum. There is no hard rule
for task assignment. Instead, the tasks can be chosen by each individual and primary
focus is for completion of tasks in a sprint. Products are built within certain time frame
and improves on each iteration from the previous one.

Three points to consider on sprint planning are as follows (Star Agile, 2020):

i. What can be delivered in increment resulting from upcoming sprint?

ii. How can the work needed to deliver an increment be achieved?
iii. When is word considered ‘done’?
23
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Reasons to choose Scrum for this project ‘Sentimento’ are as follows:

i. This is an individual project; all the tasks are self-done. Roles are not specified in
scrum but main focus is on task completion. This provides flexibility to carry out
multiple tasks in self-easiness.

ii. Previous experience of scrum during my internship helped me to manage or

breakdown work structure following scrum principles.

iii. Technical challenges in my project are higher than documentation and scrum itself
is effective to develop product more than artifacts focus.

Roles for ‘Sentimento’ project are as follows:

Product Owner, Scrum Master (Project Manager), Development Team are three roles in
Scrum project (Glamazdina, 2021). As this is an individual final year project, all the three
roles are allocated to myself as a sole performer of the project.

24
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.3 Phases of methodology

Scrum phases of Agile methodology are as follows:

i. Product Backlog Creation

The functional and non-functional requirements, as well as product features, are recorded
at this phase for later implementation. The task description provided by the client is broken
down into smaller technical components to complete. The product owner or client party
meets with the scrum master to discuss the product's development. This engagement can
happen at any time during the development process.

ii. Sprint planning and Backlog creation

The set of product backlog items are selected for the next sprint in this phase. Usually, a
sprint last about two weeks, the short duration sprints provide the opportunity to accept
client’s further feedback. This provides enough time to adapt new changes if required.
There is an additional benefit of extra time for bug addressing and debugging. After
breaking down the problem domains of the sprint into smaller tasks, scrum team is also
responsible for prioritizing the works (Star Agile, 2020).

25
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

iii. Working on sprint/ Daily Scrum

Following the planning and appropriate coordination, this is the actual working stage on
the product. A task progress board is kept up to date. Task cards are divided into four
categories: tasks to accomplish, current tasks, tasks to review, and completed tasks.

Figure 12 Sample of daily scrum board

Internal meeting of the developer team exists for less than 15minutes to synchronize the
tasks for next 24hours. The work inspection and work forecast of team members are
carried out for fine tuning of the product.

iv. Sprint review or testing and product demonstration

This phase is about viewing the tasks done in this sprint. The quality of the sprint or
deployment is evaluated and judged if any necessary changes are essential to make. The
product owner describes what product backlog items have been ‘done’ and what not. The
entire team collaborates on what to do next (Khristich, 2020).
26
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

v. Sprint retrospective

It is a final meeting and last phase of the scrum development process. It is attended by
product owner/client, scrum master and development team. This is time boxed for forty-
five minutes a week (Khristich, 2020).

Questions each team member answers in sprint retrospective are as follows (Linders,
2013):
i. What worked well in previous sprint.

ii. What did not work well, difficulties faced, requirements that were not fulfilled.

iii. What should be done differently, changes a team member want to make in working
process if any.

27
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.4 Survey Results

To obtain better and reliable data, the surveys were taken only from people who
regularly uses social media (Instagram, Snapchat, LinkedIn) in their professional life.

Profession of the survey volunteers:

Profession Number of persons

Student 12
Artist/Influencer 1
Data Analyst 1
Job Seeker 2
News Media Officer 2
University Professor 1
Table 3 Survey volunter's information

28
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.4.1 Pre-survey results

Question 1: Are you familiar about sentiment analysis on user’s activity?

Figure 13 Social media user’s familiarity with sentiment analysis

i. The survey shows nearly half of the social media users are aware about how
companies can use our social media activity data.

ii. Forty two percent approximate users not knowing about their user activity indicates
companies do not take sufficient steps to highlight how do they use their
information for multiple purposes.

iii. Sixteen percent of approximate users getting to know about how social media
companies use our data indicates movies and web series are helping to make
people understand about critical information.

29
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Question 2: Do you think any valuable information can be extracted using user's
comment, tweets, and posts?

Figure 14 User's awareness on precise social media activity

i. The answers from volunteers shows majority of the social media users who are
aware about their online data are also precisely confident on what data can be
useful for third factor use cases.

30
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Question 3: Do you comment your opinion on YouTube videos?

Figure 15 Chart showing user's interest to comment

i. Almost half of the volunteers chooses to rarely comment on YouTube videos. This
indicates that there is high probability for most of the comments found on YouTube
are spam, unwanted promotions or sometimes scam initiations.

ii. The answers suggest to make a user’s opinion filtration system from bulk of spam
comments. This will help to perform sentiment analysis on reliable data excluding
senseless spam comments.

31
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Question 4: Do you often tweet about ongoing internet topics, events or news?

Figure 16 Chart showing people's interest to tweet

i. The survey answer tells people are getting less interested to tweet about any
particular topics than in past. Before few years, the tweeting culture was booming
significantly.

ii. People’s awareness on data privacy has made people more conscious on where
to express themselves.

iii. Twenty six percent approximate users ready to tweet on critical social, global
issues indicates people are ready to unite and raise voice on any important issues.

32
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Question 5: If there was a tool that will give you summary report of public sentiment on
any internet topics, will you use such tool?

Figure 17 Chart showing user's interest to use sentiment tool

i. Majority of the survey volunteers are interested in using tools that saves time to
understand mass sentiment on particular topic.

ii. One slice of a pizza or ten percent people are more interested to go through each
of the comments/tweets by themselves.

33
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Question 6: Do you think, a separate platform to connect just social media related job
recruiter and job seeker is a better idea?

Figure 18 Chart showing people's interest looking for LinkedIn alternatives

i. Seventy three percent is a significant number of people interested to use their

profession specific job platform instead of generic job platforms like LinkedIn.

ii. The user’s interest rate makes job seeker and provider connecting feature to add
to the platform requirements.

34
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.4.2 Post-survey results

Question 1: Which social media platforms do you want to use for obtaining mass
sentiment/opinion?

Figure 19 Chart showing social platform demand for analysis

i. Majority of the answers from post-survey showed their interest to use both
YouTube and Twitter platform for sentiment analysis.

ii. Both platforms are now considered to integrate on Sentimento’s system after
requirement analysis.

35
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Question 2: Do you want to use your own third-party access tokens or will it be easier for
you to not care about it?

Figure 20 Chart showing user's interest to use their own third-party credentials

i. The survey shows the platform will be used by both technical and non-technical
users.

ii. There is strong need to implement a system where user can either use their own
third-party access tokens or can be dependent with platform itself to serve both
type of users.

36
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Question 3: About user login preferences

Figure 21 Chart showing user's login preferences

i. Survey report shows all post survey volunteers want to be logged in to the platform
unless they want to logout by themselves.

ii. All the potential users find logging in every time quite tedious. So, considering this,
bearer token mechanism will be used with no date expiry to keep user logged in.

37
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Question 4: Choose one or many sentiment categories you want in the platform

Figure 22 Chart showing what sentiment labels user want

i. Out of total nineteen responses, all the five sentiment categories are strongly
chosen.

ii. All the five sentiment labels will be considered for development work.

38
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Question 5: Do you want to see what analysis you did the last time?

Figure 23 Chart showing user's interest on report saving feature

i. Nearly eighty percent of survey volunteers want to view their past sentiment
reports.

ii. Twenty one percent of people not agreeing to record their past reports is also a
considerable number of users. Considering this, instead of storing all the past
reports, the platform will have feature where users can save the sentiment report
if they want to.

39
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Question 6: Do you think users should be limited on how many job vacancies they can
post at a time?

Figure 24 Chart showing user's opinion on limiting vacancy post at a time

i. Eighty four percent of people are conscious about chances of false vacancies post
if users can post any number of vacancies.

ii. Near to 16 percent of people desires to post any number of vacancies they want.
Considering both type of users, a threshold limit of five posts can be made at a
time. Any previous post can be deleted to other new posts. This feature satisfies
both kind of user preferences.

40
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.5 Requirement Analysis

After doing research about social media platforms, analysing what can be done for
the project with supervisors and evaluating the conducted survey results, following
requirement analysis has been done:

Functional requirements:

i. Primary Features

Req.ID Requirement Description

FR. 1 User can register to system with unique and valid registration detail
System Requirement
SR. 1 User can provide their correct registration information
SR. 2 Backend responds with registration success or fail according to
user provided information

FR. 2 User can login to system with their credentials used on registration
System Requirement
SR. 3 User can provide their correct login credentials
SR. 4 Backend responds with login fail in case of invalid information or
else user will be forwarded to dashboard

FR. 3 User can perform sentiment analysis on YouTube comments and tweets
System Requirement
SR. 5 User can provide YouTube video URL or topic of tweet to perform
sentiment analysis
SR. 6 Backend responds with sentiment report if valid input is given by
user.

41
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

FR. 4 User can view open job vacancies and skill offering from other users of
platform
System Requirement
SR. 7 User can choose either job vacancies or freelancing services to
view
SR. 8 Backend responds with currently available vacancies posted by
other users of platform

FR. 5 User can post social media related job vacancies and their skill offering in the
platform
System Requirement
SR. 9 User can provide information about their vacancy or freelancing
skill.
SR. 10 Backend stores the data in the database that is accessible by
other users later

FR. 6 User can view their detailed profile

System Requirement
SR. 11 Profile viewing REST API request can go from frontend to
backend with user’s authentication detail
SR. 12 Backend responds with user’s profile data if authentic request is
provided.

FR. 7 User can save and view their past sentiment analysis reports
System Requirement
SR. 13 User can have the option to save sentiment report for further
uses.
SR. 14 Backend stores the report in the database.

Table 4 Functional requirements

42
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

ii. Authentication
Bearer authentication tokens will be used by the system to authenticate user’s request to
backend.

iii. Reporting/Documentation requirements

Use case diagram, system architecture, flowchart and similar diagrams are designed
before the start of development of the project.

Non-functional requirements:

i. Performance: To reduce the memory usage and shorten sentiment analysis

algorithm processing time, the machine learning model is to be serialized into
pickle file.

ii. Scalability: The backend of the project is to be hosted in cloud for larger scale
accessibility of the project.

iii. Security: For user’s security reasons, the platform should not store third party
access tokens. Considering it, the tokens will be locally saved in user’s device.

A proper software requirement specification document made during the requirement

analysis phase of the project is attached in the appendix section of the report.

43
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.6 Design and Logical Diagrams

3.6.1 Entity Relationship Diagram

Figure 25 ERD of Sentimento's database

44
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.6.2 System Architecture

Figure 26 System architecture of Sentimento platform

45
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.6.3 Overall Use Case Diagram

Figure 27 Use Case Diagram of overall system

46
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.6.4 Overall Context Level Diagram

Figure 28 Context level diagram

47
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.6.5 Flowchart of sentiment analysis module

Figure 29 Flowchart of sentiment analysis module

48
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.6.6 Communication diagrams

3.6.6.1 Communication diagram for sentiment analysis

Figure 30 Communication diagram for performing sentiment analysis

49
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.6.6.2 Communication diagram for posting vacancy

Figure 31 Communication diagram for posting vacancy

50
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.6.7 Sequence Diagrams

3.6.7.1 Sequence diagram for user login

Figure 32 Sequence diagram for logging in user

51
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.6.7.2 Sequence diagram for performing sentiment analysis on tweets

Figure 33 Sequence diagram for tweet sentiment analysis

52
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.6.7.3 Sequence diagram for viewing freelancers

Figure 34 Sequence diagram for viewing freelancers

53
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.6.8 Structure chart for deleting vacancy

Figure 35 Structure chart for deleting vacancy

54
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.6.9 Activity Diagram to view saved sentiment reports

Figure 36 Activity diagram to view saved reports

55
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.6.10 Data flow diagrams

3.6.10.1 Data flow diagram for user login and registration

Figure 37 Level 0 DFD for login/registration

Figure 38 Level 1 DFD for login/registration

56
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Figure 39 Level 2 DFD for login/registration

57
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.7 Implementation following Scrum

3.7.1 Sprint 1: 1st and 2nd Increment

Highlights of sprint 1

i. The initial user interface or tiny application using NLTK library is developed for
demo showcase showing small glance of how the system will look like.

Figure 40 Initial user authentication interface

58
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

ii. Faced memory allocation error while count vectorization of sentences and
decided to remove ‘not’ as stop words from sentences to fix issue but this
created another big issue of change of meaning of a sentence.

Figure 41 Change in meaning of a sentence while removing stop words

59
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.7.2 Sprint 2: 3rd, 4th, 5th Increment

Highlights of sprint 2:

i. Data cleaning or pre-processing

Figure 42 Function to clean sentence

60
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

ii. Multinomial Naive Bayes module for both debug and production level

Figure 43 Function to make updates in serialized training model

In the above code, ‘fit_transform’ method is used for training data whereas
‘transform’ method is used for testing data. This is done for removing exclusion biasness
from the system towards particular features. Exclusion biasness occurs if features of
training data are not evaluated just because it does not exist in testing data. ‘fit_transform’
is used for training data to learn mean and variance of the features from training data.
The mean and variance of the testing data needs to be new for our model so we use
‘transform’ for it (Khanna, 2020).
61
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Figure 44 Function to predict probability of each sentence to fall in 0 or 1 category

The scikit-learn library has been used for the Multinomial Naïve Bayes calculated.
The detailed theory is explained in the previous solution section of the report. There are
two functions which does the same work but one is for debug level which uses textual
training data for each run while the production ready function uses already existing trained
model.

In the production level, using the textual data for training data too can cost higher memory
usage. And it is not efficient to do for production. To shorten the runtime duration, once
calculated object of the run time of CountVectorizer class is dumped in external file. The
once calculated object of Multinomial class is also dumped in another external file. Saving
a machine learning model is also known as serialization (GeeksforGeeks, 2021).

62
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Figure 45 Single class to access all algorithmic feature of system

This class is used to make objects of result of sentiment analysis by the backend
developed with flask and returns JSON response using the object created from this class.

63
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.7.3 Sprint 3: 6th and 7th Increment

Highlights of the sprint 3:

i. YouTube and Twitter data integration

For extracting both the third party’s data, their official API guidelines or documentation is
studied to make these two functions.

Figure 46 Function to extract tweets of chosen topic

64
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Figure 47 Function to extract YouTube comments of chosen video

65
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

ii. Flask setup with user authentication

Figure 48 Function to handle login request

66
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Figure 49 User login request task

67
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.7.4 Sprint 4: 8th, 9th and 10th Increment

Highlights of the sprint 4:

i. Sentiment analysis module integrated with backend

Figure 50 API handling sentiment analysis request on chosen YouTube video

68
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

ii. Report and Vacancy saving/viewing feature

Figure 51 Viewing user's saved sentiment reports

iii. Sign in, Signup user interfaces for front end application

Figure 52 Git commit showing front end user authentication pages task done

69
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.7.5 Sprint 5: 11th, 12th and 13th increment

Highlights of sprint 5:

BLOC state management is utilized and all the essential remaining front-end tasks like
profile viewing, vacancy posting, managing coupons are completed during this long sprint.

State Management:

Business Logic Component (BLOC) is used for state management of the built Flutter
application. Building production level or qualitative applications, managing current state
becomes critical. State management helps many developers to collaborate and work in a
single code base seamlessly following pattern (Bloc Library, 2021).

Figure 53 Bloc state management architecture (Aqeel, 2020)

For this application, multiple blocs are created for several use cases like authentication,
sentiment analysis, viewing vacancies. Wherever there is presence of multiple state like
data loading and loaded state, Bloc has been consumed in the application.

70
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Figure 54 Git commit showing tasks of fifth sprint

71
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

3.7.6 Sprint 6: 14th and 15th increment

Highlights of sprint 6:

Normally while following Scrum in corporate or office environment, one sprint lasts
around one to four weeks. For this sprint (six) of the project, this sprint lasted quite longer
than a month as there were no past experiences of publishing a fully featured application
to Google Play Store. The policy issues and formalities to publish this application took
longer time making this sprint longer than usual Scrum’s sprint.

Figure 55 API hosted in Heroku platform

72
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Figure 56 Application published to Google Play Store

73
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Chapter 4: Testing and Analysis

4.1 Test Plan

During the multiple sprints of development of this project ‘Sentimento’, the features
built or integrated in the platform are tested properly with brief documentation of the
individual test. The testing plans are listed below as following:

4.1.1 Unit testing plan

It is an individual component testing method where fractional logical parts of the

system are tested. Unit testing validates that each unit of the software codes performs as
expected. It can be performed for a method, user interface functionalities and even for
any loops or conditions that plays logical part in the system (Khorikov, 2020).

For the unit testing of this project, the individual interface components or functionalities of
the built Flutter application is to be tested with precision. Apart from the application where
user interacts, the built Flask API end points is also to be tested. The discrete end point
APIs for sentiment analysis, vacancy without user authorization token and all built
fragmentary backend logical parts are tested using Postman. Also, the access
authentication for third parties is also tested in unit testing of this project.

74
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

The unit testing plan for this project are listed in the table below as follows:

Flutter application test

Test Id Objective
1 To check if user can register with already used credentials.
2 To check if user can login without being registered.
3 To check if user can be registered with valid credentials.
4 To check if user can login with valid credentials.
5 To check if user can login with invalid password.
6 To check if user can view their profile.
7 To verify user cannot access sentiment analysis feature without providing
tokens or coupons.
8 To test if user can access sentiment analysis feature after providing tokens
or coupons.
9 To test user can perform polarity check on manual sentences.
10 To test user cannot perform sentiment analysis with invalid coupon or token.
11 To test user can view their saved sentiment reports.
12 To verify user will be shown error message if no any sentiment reports are
saved.
13 To test if dashboard is responsive to screen size on multiple devices.
14 To show user can post a vacancy in the platform.
15 To show user cannot post more than five vacancies.
16 To show user redirects to phone call when presses contact button.
17 To verify user can delete their previously posted vacancy.
18 To test user can logout from the platform.
19 To show error message will be displayed if user tries to access saved reports
without internet connection.
20 To show user’s name and email in the interface when user is navigated from
login screen to dashboard.
Table 5 Unit testing for Flutter application
75
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Flask API test

Test Id Objective
1 To read a user's profile without supplying a bearer token.
2 To show open vacancies can be viewed without user authentication.
3 To show user cannot delete another user’s vacancy post.
4 To show user cannot perform analysis on more than 2000 comments.
Table 6 Unit testing for Flask API

76
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

4.1.2 System testing plan

It is a comprehensive testing of the system itself. The fundamental or core features

implemented in the platform are tested during system testing of the project. A complete
project can be divided into multiple segments like frontend, backend. These each
development works are integrated together to achieve a fully functional application
(Hamilton, 2022).

System testing is about validating that all the integrated systems are merged to make a
sole application perform its functionalities. Integration testing justifies the unity of team
work of the built platform. Those features included in the software requirement
specification document are tested in the system testing of this report.

The system testing plan for this project are listed in table below:

Test Id Objective
1 To validate backend server is live and responsive in cloud.
2 To show backend is properly integrated with frontend.
3 To show user can perform sentiment analysis on tweets.
4 To show user can save sentiment report.
Table 7 System testing for built platform

77
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

4.2 Unit Testing

UT1: To check if user can register with already used credentials.

Action Registration information was provided with already used email.

Expected Result A toast message about invalid registration would be shown.
Actual Result A toast message was shown stating user is already registered with
this email.
Conclusion The platform is able to prevent multiple use of same email which
states the test is successful.
Table 8 Unit test 1

Figure 57 Trying to register with already used email

78
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT2: To check if user can login without being registered.

Action Not registered email address was provided while logging in

Expected Result A toast message about invalid login would be shown.
Actual Result A toast message was shown stating that user with the provided email
does not exist in the system.
Conclusion Users are not allowed to log in without being registered which states
that the test is successful.
Table 9 Unit test 2

Figure 58 Trying to login without being registered

79
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT3: To check if user can be registered with valid credentials.

Action A new user information was provided to register.

Expected Result User will be navigated to dashboard.
Actual Result After clicking register button, loading interface was shown and was
navigated to dashboard
Conclusion Users can register themselves to platform stating this test is
successful.
Table 10 Unit test 3

Figure 59 Registering user in the platform

80
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT4: To check if user can login with valid credentials.

Action Valid email address and password was provided to login.

Expected Result User will be navigated to dashboard.
Actual Result After clicking login button, loading interface was shown and was
navigated to dashboard.
Conclusion User can log in to the platform which makes this test successful.
Table 11 Unit test 4

Figure 60 Logging inside the platform

81
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT5: To show user cannot login with invalid password.

Action Valid email address and invalid password was provided to login.
Expected Result User will be shown error message.
Actual Result After clicking login button, loading interface was shown and toast
message about incorrect password was shown.
Conclusion User cannot log in to the platform with incorrect password which
makes this test successful.
Table 12 Unit test 5

Figure 61 Trying to log in with incorrect password

82
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT6: To show that user can view their profile.

Action Profile viewing icon was clicked.

Expected Result User will be shown their profile.
Actual Result After clicking profile button, loading interface was shown and profile
information was shown
Conclusion User can view their profile which states the test is successful.
Table 13 Unit test 6

Figure 62 Showing profile icon to enter profile page

83
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Figure 63 User navigated to profile screen after loading completes

84
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT7: To verify user cannot access sentiment analysis feature without providing tokens
or coupons.

Action YouTube analysis was tried to access without providing token or

coupon.
Expected Result A toast message is expected to be shown.
Actual Result A toast message was shown stating tokens to enter first.
Conclusion User cannot perform sentiment analysis without entering tokens
which states the test is successful.
Table 14 Unit test 7

Figure 64 Trying to perform sentiment analysis without providing tokens

85
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT8: To show user can access sentiment analysis feature after providing tokens or
coupons.

Action Coupons were entered and YouTube comments analysis was

selected, video URL and 500 numbers of comments were asked to
analyse.
Expected Result A successful sentiment analysis report would be shown.
Actual Result Report interface with thumbnail and analysis data was shown.
Conclusion User can perform sentiment analysis after entering tokens which
states the test is successful.
Table 15 Unit test 8

Figure 65 Entering coupons and video information

86
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Figure 66 Showing sentiment report after loading completes

87
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT9: To test user can perform polarity check on manual sentences.

Action A sentence was provided to check polarity.

Expected Result The polarity of the sentence will be shown.
Actual Result The indicator changed its position and polarity was shown.
Conclusion User can perform sentiment analysis on manual sentences stating
this test is successful.
Table 16 Unit test 9

Figure 67 Polarity check of manual sentence

88
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT10: To show user cannot perform sentiment analysis with invalid coupon.

Action Invalid coupon was provided to perform YouTube comments

analysis.
Expected Result A toast message would be shown.
Actual Result A toast message about invalid request was shown.
Conclusion User cannot perform sentiment analysis using invalid coupons
stating this test is successful.
Table 17 Unit test 10

Figure 68 Providing invalid coupon

Figure 69 Trying to perform analysis using invalid coupon

Figure 70 Toast message about invalid request

89
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT11: To verify that user can view their saved sentiment reports.

Action ‘My Reports’ button was clicked then ‘Show Details’ was clicked.
Expected Result All the saved sentiment reports were expected to appear.
Actual Result All the saved reports were shown in order.
Conclusion User can view their saved reports stating this test is successful.
Table 18 Unit test 11

Figure 71 Saved sentiment reports of user

90
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT12: To verify user will be shown error message if no any sentiment reports are saved.

Figure 72 Trying to view reports without performing any analysis

91
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT13: To test if dashboard is responsive with screen size on multiple devices.

Action Phone was rotated to landscape from dashboard screen

Expected Result The four menus were expected to align in same line on landscape.
Actual Result All the four menus were aligned on same line as expected.
Conclusion The dashboard menus align according to screen size which states
test is successful.
Table 19 Unit test 13

Figure 73 Dashboard in portrait view

92
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Figure 74 Dashboard in landscape view

93
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT14: To show user can post a vacancy in the platform.

Action All the fields were filled and ‘Post vacancy’ button was clicked.
Expected Result A toast message would be shown.
Actual Result A toast message stating vacancy posted on the platform was shown.
Conclusion The vacancy was added which states test is successful.
Table 20 Unit test 14

Figure 75 Posting vacancy in the platform

94
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT15: To show user cannot post more than five vacancies.

Action A new vacancy is tried to add from the user account which already
has five active posts.
Expected Result A toast message would be shown.
Actual Result A toast message stating vacancy posted on the platform was shown.
Conclusion The vacancy was added which states test is successful.
Table 21 Unit test 15

Figure 76 User having five open vacancies trying to add another new vacancy

95
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT16: To show user navigates to phone call when presses contact button.

Action ‘Contact’ button was pressed from detailed view of a freelancer.

Expected Result User will be taken to phone call user interface.
Actual Result Cellular calling window was opened after pressing ‘Contact’ button.
Conclusion The feature worked as expected which states test is successful.
Table 22 Unit test 16

Figure 77 User navigated to phone call window

96
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT17: To verify user can delete their previously posted vacancy.

Action ‘Delete Post’ button was pressed from detailed view of a freelancer.
Expected Result User will be taken to dashboard after deleting with a toast message.
Actual Result Dashboard was opened with a message stating post deleted.
Conclusion The feature worked as expected which states test is successful.
Table 23 Unit test 17

Figure 78 Deleting previously added post

97
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT18: To test user can logout from the platform.

Action ‘Logout’ icon was pressed from dashboard.

Expected Result User will be taken to Login screen.
Actual Result Login screen was opened after pressing logout icon.
Conclusion The feature worked as expected which states test is successful.
Table 24 Unit test 18

Figure 79 Logging out from the platform

98
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT19: To verify that error message will be displayed if user tries to access saved reports
without internet connection.

Action Both the Wi-Fi and cellular data was turned off and ‘My Reports’ was
clicked.
Expected Result A toast message would be shown.
Actual Result A toast message about no internet was shown.
Conclusion The feature worked as expected which states test is successful.
Table 25 Unit test 19

Figure 80 Turning off Wi-Fi and Mobile data

Figure 81 Trying to view reports without internet connection

99
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT20: To show user’s name and email in the interface when user is navigated from login
screen to dashboard.

Action Valid login credential was provided and navigated to dashboard.

Expected Result User’s name and email was expected to be shown on top of
dashboard.
Actual Result Name and email were not shown while opening application for first
time.
Conclusion User information was shown from second and further opening of
application but not in first time stating the test failed and there is
space to improve.
Table 26 Unit test 20

Figure 82 Dashboard after logging in

Figure 83 Dashboard after opening for second time

100
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

API unit testing:

UT21: Trying to view user's profile without supplying a bearer token.

Action HTTP GET request was done on profile viewing backend URL
Expected Result An error message would be provided.
Actual Result Error message about missing authorization header was shown.
Conclusion The test is successful.
Table 27 Unit test 21

Figure 84 Trying to access profile without providing authentication header

101
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT22: To show open vacancies can be viewed without user authentication.

Action Vacancies viewing URL was opened on browser.

Expected Result JSON data of vacancies would be shown.
Actual Result Vacancies information was shown.
Conclusion Vacancies data was shown without providing authentication header
which makes this test successful.
Table 28 Unit test 22

Figure 85 Viewing vacancies information from web browser

102
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT23: To show user cannot delete another user’s vacancy post.

Action Vacancy id of other user’s post was provided to delete.

Expected Result Response about invalid API request would be provided.
Actual Result Message stating unauthorized request was shown.
Conclusion The vacancies are protected from unauthorized removal stating the
test successful.
Table 29 Unit test 23

Figure 86 Trying to delete another user's vacancy

103
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

UT24: To show user cannot perform analysis on more than 2000 comments.

Action Video id, API key and 5000 number of comments were given as input
for post request.
Expected Result Sentiment analysis would be prevented and error message would
be provided in response.
Actual Result Message guiding how many comments can be taken was shown in
API response.
Conclusion The backend is throttled from bulk comments analysis stating the
test successful.
Table 30 Unit test 24

Figure 87 Trying to perform sentiment analysis on more than 1000 data

104
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

4.3 System Testing

ST1: To validate backend server is live and responsive in cloud.

Action Curl command was applied to domain of the hosted API from
command prompt.
Expected Result The built API documentation would be shown.
Actual Result JSON format information was shown.
Conclusion The Flask built backend API hosted on Heroku transferred data as
response which states server is live and test is successful.
Table 31 System test 1

Figure 88 Checking hosted backend API on Heroku

105
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

ST2: To show backend is properly integrated with frontend.

Action Sentiment analysis count is noted, then a manually provided

sentence analysis is performed then again, the analysis count is
viewed.
Expected Result Sentiment analysis count would increase by 1.
Actual Result The analysis count increased from 83 to 84 after performing
analysis.
Conclusion Increase on analysis count indicates that database is updated
indicating successful integration of frontend, backend and database.
Table 32 System test 2

Figure 89 Count before performing analysis

Figure 90 Performing polarity check

Figure 91 Count after performing analysis

106
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

ST3: To show user can perform sentiment analysis on tweets.

Action Sentiment analysis on ‘Nepse’ topic was done.

Expected Result Analysis report would be shown.
Actual Result The detailed report was shown.
Conclusion Test is successful.
Table 33 System test 3

Figure 92 Performing sentiment analysis on tweets

107
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

ST4: To show user can save sentiment report.

Action ‘Save’ icon was clicked after viewing sentiment report.

Expected Result The recently saved report would be shown on ‘My Reports’ interface.
Actual Result Report on lastly saved analysis was shown.
Conclusion The saved report is saved on database indicating integrity of Report
table and user interface. So, this test is successful.
Table 34 System test 4

Figure 93 Viewing recently saved report

108
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

ST5: To show build Flutter application runs with no compilation errors

Action Android Emulator was selected from VS Code and ‘flutter run’
command was executed from the terminal.
Expected Result The application was expected to be opened in Android emulator.
Actual Result Application was opened after 8.2 second of compilation without any
errors.
Conclusion This test is successful as application opened as expected.
Table 35 System test 5

Figure 94 Flutter Application Compilation

109
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

4.4 Critical Analysis

The above documented unit and system testing demonstrates that all the
requirements are considered and implemented in this platform. By following the SCRUM
principle while carrying out the project, the development was completed within allocated
timeframe and enough time was available for additional improvements.

One of the testings failed for this project. The error was faced in frontend part of the
project. User’s name and email was not shown in the top of profile on first login. This error
existed as there was no state management applied for this part. Apart from this, all the
testings were successful. To summarize the carried-out test, following points are written
as follows:

i. Functionalities mentioned in the requirement analysis worked smoothly as

expected.

ii. Frontend, backend was fully integrated and worked all together.

iii. All the validation, data throttling and logical parts of the interface worked as
expected.

iv. The state management worked smooth and loading screens were shown in every
buffering stage for better user experience.

110
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Chapter 5: Conclusion

5.1 Legal, Social and Ethical Issues

Here’s how the platform is built to prevent any professional issues from happening:

Creating a product with the potential of numerous users to be involved adds a

sensitive responsibility to protect the rights and information of consumers of the product.
Considering this part of the professional ethics and computer law: certain ethics, moral
and legal duties are strictly followed in the development of this project which are as
follows:

i. Users are trusting with their personal information while using this app,
understanding this, passwords are hashed before storing in database. The email that user
provides while registering in the platform is not visible to third person/parties.

ii. The third party’s (YouTube and Twitter) API key user has to enter to perform
sentiment analysis are user’s sensitive information. Keeping this in mind, those API keys
are not stored in system’s database. Instead, the keys are encrypted and locally saved in
the user’s device.

iii. To prevent bad actors from posting spam vacancies, the system throttles users to
post only five open vacancies at a time.

iv. There might be any incorrect information while posting vacancy by user. In such
case user wants to remove such misinformation. Considering this, there is a feature to
delete previous posts.

111
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

5.2 Advantages

Real life uses of sentiment analysis

Social Media Monitoring, Political Campaigns, Advertising/Marketing, Customer

Review Extraction, Market Research, Mental Health Brand/Influencer’s Reputation
Monitoring are some of the areas where sentiment analysis can be used. All these
mentioned fields or parties can be benefited in accessing their strength and weakness by
analysing their audience’s opinions or from public (Saifee Vohra, 2013).

Figure 95 Distinct use cases of sentiment analysis (Samadhan Engineering, 2021)

For the artificial intelligence or machine trained model to better communicate with
humans, there is need for machines to understand the human sentiments. Thus, the
application of sentiment analysis is not only limited to human uses but also to artificial
intelligence. The tasks where humans are better than machines at this stage can be
replaced by machines themselves with the better sentiment consumption by machines.
There are endless areas where sentiment analysis can be performed as it carries huge
potential use cases. And it is one of the emerging, innovative and applicable area for the
present and future human society (Somers, 2019).

112
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Real life uses of built application:

i. Users can view/add posts about freelancing service and job vacancies related to
social media field.

ii. Data analysts can quickly perform sentiment analysis and extract summary from
the platform for their any sorts of use case.

iii. Marketing professionals can observe current public sentiment about any public
figures and compare among them which can be useful for decision taking about
collaborating with best influencers.

iv. Social media influencers can observe audience’s feedbacks on their content and
publish further content accordingly.

v. Public sentiment regarding any topic can also be useful for news media
professionals. Also, other users can also benefit from this feature.

113
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

5.3 Limitations

Challenges that can arise while performing sentiment analysis are listed below:
(Mohammad, 2016)

Sentiment Analysis is a complex and arduous task in natural language processing.

The challenges of machine learning based sentiment analysis are as follows:

i. Irony and Sarcasm: People uses sarcasm and ironic sentences to express the
meaning. In such cases, the real point they want to make can be one thing but the straight
sentence can mean something different. Even human can find it difficult to understand
sometimes which makes more difficult for trained models to predict the actual meaning.

ii. Context and Polarity: As trained model works within the provided boundaries, it’s
hard to understand the actual sentiment according to the context. Example: the opinion
can be regarding something but the context can be different. Here the calculated
sentiment polarity does to belong to real context resulting less trustworthy analysis.

iii. Stop words: The stop words are better to remove for reducing the numerical
vectorized array of data. This can save memory and better performance but removal of
stop words can alter the meaning of a sentence. Example: In the sentence ‘I do not like
you’. When ‘not’ stop word is removed while pre-processing, the meaning changes. So,
it is challenging to deal with stop words.

Limitations of this built application:

The designed algorithm is not well optimized to perform analysis quickly. There is still
adequate place to improve time and space complexity of the algorithm. So, the platform
is throttled to only take maximum of one thousand data (comment/tweet) at a time.

114
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

5.4 Personal experience and Future work

5.4.1 Self-reflection or personal experience:

Addition of this project on my curriculum vitae is a significant factor that supported

me to receive invitation for a trainee interview in Business Informatics department at
Technical University Dortmund located in Germany. This opportunity was achieved as I
was looking for international opportunities on Data Science. I got the chance to describe
my final year project about the overall integration of the built system. Encouraging
feedback and short mentoring was received during the interview.

Using Flutter to build the frontend part of this project and my past internship on this
framework improved my Flutter knowledge and hands-on experience. This provided me
the confidence and certain level of quality to engage in mobile application to smoothly
transit from academic to professional life.

5.4.2 Here’s how I will be further engaged on this project in future:

Being a data enthusiast from early days of entering technology field, there is
always a keen interest of pursuing Data Science for my post graduate studies. This is an
undergraduate academic work but I will be engaged in this same field even after the
completion of this undergraduate project.

The further aim of mine is to improve space and time complexity of the sentiment analysis
part of this project. The accuracy of the analysis at this level is also not well reliable.
Further research and theoretical findings from multiple scholars will be absorbed to
improvise the quality of this work. And there is a significant possibility to purse this same
topic for my post-graduate thesis work.

115
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Chapter 6: References

Excluding theoretical research for carrying out this project, there is also use of
diverse contents that assisted in executing development work of the project. Attachment
of all the references that were accessed during or before the development of this project
are as follows:

i. LinkedIn learning course for building Flask API by Bruce Van Horn

Figure 96 Snapshot of a LinkedIn learning course for Flask

116
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

ii. Flask official documentation: While performing CRUD operations and making
models for database, Flask’s official documentation was surfed for class, methods using
guide.

Figure 97 Snapshot of Flask documentation web page

iii. Flutter package manager: It is an official package manager where developers can
contribute their work and enormous cross-platform application components are available
here.

Figure 98 Snapshot of Flutter package manager

117
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

iv. Stack overflow community: When there was run-time error building frontend or
backend, stack overflow was surfed to view how developers faced and solved similar
issues in the past. With such reference, the debugging and solving run-time errors were
smooth.

v. Scikit-learn: For the sentiment analysis (Multinomial Naive Bayes algorithm) part
of the project, scikit-learn, a machine learning library has been used.

Figure 99 Snapshot of scikit-learn library

118
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Chapter 7: Bibliography

Aqeel, A., 2020. BLoC (Local & Global) State Management with Flutter. [Online]
Available at: https://devmuaz.medium.com/bloc-local-global-state-management-with-
flutter-e8e443b135f8
[Accessed 23 March 2022].
Binita Verma, R. S. T., 2018. Sentiment Analysis Using Lexicon and Machine Learning-
Based Approaches: A Survey. Bhopal, Maulana Azad National Intitute of Technology.
Bloc Library, 2021. Why Bloc?. [Online]
Available at: https://bloclibrary.dev/#/whybloc
[Accessed 2 April 2022].
C9 Apps Desenvolvimento de Software Ltda ME, 2017. TSentiment. s.l.:C9 Apps
Desenvolvimento de Software Ltda ME.
Doshi, S., 2021. Why I Prefer Flutter Over React Native for App Development. Web Dev
Zone, 23 August, pp. 2-4.
GeeksforGeeks, 2021. Saving a machine learning Model. [Online]
Available at: https://www.geeksforgeeks.org/saving-a-machine-learning-model/
[Accessed 10 January 2022].
GeeksforGeeks, 2021. Software Engineering | Quality Characteristics of a good SRS.
[Online]
Available at: https://www.geeksforgeeks.org/software-engineering-quality-
characteristics-of-a-good-srs/
[Accessed 9 December 2021].
Glamazdina, Y., 2021. A Guide to the Agile Software Development Life Cycle (SDLC).
[Online]
Available at: https://brocoders.com/blog/agile-software-development-life-cycle/
[Accessed 26 March 2022].
Hamilton, T., 2022. Integration Testing: What is, Types, Top Down & Bottom Up
Example. [Online]
Available at: https://www.guru99.com/integration-testing.html
[Accessed 28 March 2022].
119
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Jay Khatri, I. K., 2020. Sentiment Analysis - Know The Opinion. s.l.:Jay Khatri.
Jha, V., 2017. Naive Bayes - machine learning algorithm for classification problems.
[Online]
Available at: https://www.techleer.com/articles/200-naive-bayes-machine-learning-
algorithm-for-classification-problems/
[Accessed 07 December 2021].
John Hattie, H. T., 2007. The Power of Feedback, Auckland: University of Auckland.
Khanna, C., 2020. What and why behind fit_transform() and transform() in scikit-learn!.
[Online]
Available at: https://towardsdatascience.com/what-and-why-behind-fit-transform-vs-
transform-in-scikit-learn-78f915cf96fe
Khorikov, V., 2020. Unit Testing: Principles, Practices and Patterns. 1st ed. Shelter
Island: Manning Publications.
Khristich, S., 2020. The Scrum Cycle in Agile Software Development. [Online]
Available at: https://tateeda.com/blog/the-scrum-cycle-in-agile-software-development
[Accessed 17 Febraury 2022].
Linders, B., 2013. Which Questions do you Ask in Retrospectives?. [Online]
Available at: https://www.benlinders.com/2013/which-questions-do-you-ask-in-
retrospectives/
[Accessed 28 March 2022].
LinkedIn, 2021. About LinkedIn. [Online]
Available at: https://about.linkedin.com/
[Accessed 8 December 2021].
Lior Rokach, O. M., 2015. Decision Trees, Tel Aviv: Oded Maimon Department of
Industrial Engineering.
Liu, J., 2016. Data Visualization on Sentiment Analysis. Chicago: Jiaqi Liu.
Ludvig Persson, J. L., 2018. Sarcasm Detection with TensorFlow, Stockholm: School of
Electrical Engineering and Computer Science.
Mohammad, S. M., 2016. Challenges in Sentiment Analysis, Ottawa: National Research
Council Canada.

120
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Mulvenna, J. A., 2015. Improved lexicon-based sentiment analysis for social media
analytics, Belfast: Queen's University Belfast - Research Portal.
Perisic, M., 2014. The Good, The Bad and The Ugly of Agile Software Development.
[Online]
Available at: https://www.linkedin.com/pulse/20141106203747-11592828-the-good-the-
bad-and-the-ugly-of-agile-software-development/
[Accessed 1 April 2022].
Prokopisko, A., 2019. Software development life cycle. [Online]
Available at: https://medium.com/@artjoms/software-development-life-cycle-sdlc-
6155dbfe3cbc
[Accessed 04 April 2022].
Rajat Kathuria, M. K. G. V. K. B. S. K., 2017. Future of Work in a Digital Era: The
Potential and Challenges for Online Freelancing and Microwork in India, New Delhi:
INDIAN COUNCIL FOR RESEARCH ON INTERNATIONAL ECONOMIC RELATIONS.
Saifee Vohra, J. T., 2013. Applications and Challenges for Sentiment Analysis : A
Survey. International Journal of Engineering Research & Technology (IJERT), 2(2), pp.
3-5.
Samadhan Engineering, 2021. Services: Artificial Intelligence Applications. [Online]
Available at: https://www.thesamadhan.com/services/artificial-intelligence-applications
[Accessed 13 December 2021].
Scikit-learn, 2021. User Guide: Naive Bayes. [Online]
Available at: https://scikit-learn.org/stable/modules/naive_bayes.html#multinomial-naive-
bayes
[Accessed 14 Decembwr 2021].
scrum.org, 2020. What is Scrum?. [Online]
Available at: https://www.scrum.org/resources/what-is-scrum
[Accessed 21 November 2021].
Shimon Ullman, T. P., 2014. Unsupervised learning, Massachusetts: Massachusetts
Institute of Technology.
Somers, M., 2019. Emotion AI, explained. Massachusetts: MIT Management Sloan
School.
121
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Sommerville, I., 2011. Software Requirement Documents. In: M. H. Marcia Horton, ed.
SOFTWARE ENGINEERING. Hagerstown: Pearson, pp. 91-98.
Stanford Edu, 2009. Properties of Naive Bayes. [Online]
Available at: https://nlp.stanford.edu/IR-book/html/htmledition/properties-of-naive-bayes-
1.html
[Accessed 18 December 2021].
Star Agile, 2020. Overview of Scrum Phases. [Online]
Available at: https://staragile.com/blog/scrum-phases
[Accessed 6 April 2022].
Walaa Medhat, A. H. H. K., 2014. Sentiment Analysis Algorithms and applications. Ain
Shams Engineering Journal, V(8), pp. 1093-1113.
Young, D. C., 2013. Software Development Methodologies. Huntsville: Alabama
Supercomputer Center.

122
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Chapter 8: Appendix

Appendix 1: Additional Diagrams/Figures

i. Gantt chart

Figure 100 Gantt Chart of the project work division

123
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

ii. Work breakdown structure

Figure 101 Work breakdown structure of overall project

124
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

iii. Milestone chart

Figure 102 Milestone chart of overall project

125
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

iv. Layers in the designed sentiment analysis algorithm

Figure 103 Layers showing how sentiment analysis works

126
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

v. Product Backlog

Figure 104 Product backlog design

127
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

vi. Sprint Backlogs

Figure 105 Sprint 1 Backlog

128
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Figure 106 Sprint 2 Backlog

129
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Figure 107 Sprint 3 Backlog

130
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Figure 108 Sprint 4 Backlog

131
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Figure 109 Sprint 5 Backlog

132
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Figure 110 Sprint 6 Backlog

133
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Appendix 2: Reasons to use consumed libraries/framework

i. Using BLOC for state management

Business Logic Component (BLOC) is recommended by Google itself for any state
management related tasks in Flutter application. There is a need to use state
management for quality applications where managing state becomes significant action.
There is a need of tracking the current state of the application to navigate through multiple
user interfaces and passing data using their models. Multiple developers can work in the
same project as workloads are divided into multiple modules. The interface design codes
and logical codes are separately written in BLOC pattern (Bloc Library, 2021).

ii. Using Flask over fully featured framework like Django

For this project, we need backend to develop only API excluding any views like
web pages. If Django or other fully fetched framework was used, it would only cause extra
space which is not a good practise. And Django provides most of the work done by itself
with its REST framework. Using such pre-built method, there would be less learning
opportunities. Instead, I found it better to write own functions while using Flask as it is a
microframework and there is a need to import other modules as per need.

iii. Multinomial Naive Bayes algorithm over available package like TextBlob

Packages like TextBlob, Vader has the functionality to calculate polarity by itself.
There would be no learning opportunities on supervised machine learning if the project
was fully depended on such libraries. So, the plan was made to break down analysis into
spam, sarcasm labels. Labelled data models were used from Kaggle (data model source)
and using Multinomial Naïve Bayes of Ski-kit learn, the labels of testing datasets were
performed. This provided me the learning opportunity and certain level of experience on
supervised learning.
134
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Appendix 3: Software Requirement Specification

Project: Sentimento
“Social media assisting platform with sentiment analysis”

Introduction

The SRS document is a non-technical report clarifying the features/capabilities of

the software, system or application. It contains detailed information on functional and non-
functional requirements of the project. This document is prepared identifying and
evaluating the client’s requirement. If the project is not being built for specific client, the
survey feedbacks of the potential users of the system is used to prepare the software
requirement specification document (Sommerville, 2011).

The qualities of a clear and in-depth SRS document are as follows (GeeksforGeeks,
2021):
i. The requirements stated by client or feedbacks from surveys are clearly addressed
in the document.

135
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

ii. All the must have functional and non-functional requirements are properly
attached.
iii. Preciseness over ambiguousness is another essential quality for a good SRS
document.
iv. It is better for a SRS document to be easily modifiable as requirements of a project
in SCRUM of Agile are prone to have changes or improvements in the further
iterations of the project.
v. The SRS document is also made for client or non-technical viewers. It is a good
practise to make this document easily understandable with use of simpler
language and terms.

136
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Purpose

Primary objective of this SRS document is to provide substantial information about

the functional and non-functional requirements of the Sentimento platform. This document
clearly highlights must include features of the platform to function. The other tool and
technologies requirements for the development of the project is also defined in this
document.

The core purpose of this mobile application is to provide a user-friendly platform where
non-technical and technical (with understanding of Data Analytics) users can perform
sentiment analysis within same interface.

Intended Audiences of this document

This document is primarily intended for the development team of the Sentimento
project. Developers from algorithm team, backend and frontend including other team
members engaged in technical and non-technical areas are the key readers/viewers of
this document. The client of the project or potential users can also read this document to
verify their demands are included in the document.

Why this document for the developer team?

Viewing all the features to include in the platform, developers can identify and plan
their moves accordingly. The vocal hearing from the client is not consistent and official
but this document with client’s approval provides clear vision on what to do and how to
implement those essential features in the platform. This document can also be a proof of
record of tasks to carry out for both the parties (client and development team).

137
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Contents

User’s need:
From the feedbacks of pre-survey, a mobile application with easy-to-use interfaces
is to be developed where they can just provide video URL or topic of tweet. And just with
this little information, users should be able to do their sentiment analysis task. The
demand of a social media related job vacancies posting/viewing feature is also
considered to include in the system.

Assumptions and dependencies:

The development work of this project is first divided into four sections and carried
out sequentially. The four sections are as follows:

i. Sentiment analysis functionality

Two functions are to be made to extract comments of a chosen video of YouTube and
tweets using each of the respective third-party documentation. After that, a function to
clean the data is made and using Multinomial Naive Bayes functionality of scikit-learn and
training data models, sentiment analysis module is developed.

ii. Backend development

Flask framework will be used to make the application programming interface (API) of the
system. The CRUD operations and sentiment analysis module will be integrated with this
part of the development and it interacts with the frontend part of the system to complete
the overall usefulness of the system.

138
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

iii. Frontend development

Flutter, cross-platform application framework is to be used to effectively developed all the

features of the designed platform. For state management task, Business Logic
Component (BLOC) will be used. The packages from the Flutter community will be utilized
to perform basic functionalities quickly and efficiently. The folder structure will be
maintained as recognized by Flutter developer community. The user interfaces will be
broken down into smaller components to make the particular code easily accessible when
needed to make any changes.

iv. Hosting and taking to production level

Heroku, a cloud hosting service is to be used to host the Python backend. The free
account with zero payment to make service will be utilized to make the platform’s API live.

The mobile application is estimated to publish on Google Play Store if the developed
application at the end fulfils all the policy requirements of publishing application on Play
Store.

139
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Requirements

Functional requirements:

i. Primary Features

Req.ID Requirement Description

FR. 1 User can register to system with unique and valid registration detail
System Requirement
SR. 1 User can provide their correct registration information
SR. 2 Backend responds with registration success or fail according to
user provided information

FR. 2 User can login to system with their credentials used on registration
System Requirement
SR. 3 User can provide their correct login credentials
SR. 4 Backend responds with login fail in case of invalid information or
else user will be forwarded to dashboard

FR. 3 User can perform sentiment analysis on YouTube comments and tweets
System Requirement
SR. 5 User can provide YouTube video URL or topic of tweet to perform
sentiment analysis
SR. 6 Backend responds with sentiment report if valid input is given by
user.

140
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

FR. 4 User can view open job vacancies and skill offering from other users of
platform
System Requirement
SR. 7 User can choose either job vacancies or freelancing services to
view
SR. 8 Backend responds with currently available vacancies posted by
other users of platform

FR. 5 User can post social media related job vacancies and their skill offering in the
platform
System Requirement
SR. 9 User can provide information about their vacancy or freelancing
skill.
SR. 10 Backend stores the data in the database that is accessible by
other users later

FR. 6 User can view their detailed profile

System Requirement
SR. 11 Profile viewing REST API request can go from frontend to
backend with user’s authentication detail
SR. 12 Backend responds with user’s profile data if authentic request is
provided.

FR. 7 User can save and view their past sentiment analysis reports
System Requirement
SR. 13 User can have the option to save sentiment report for further
uses.
SR. 14 Backend stores the report in the database.

141
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

ii. Authentication
Bearer authentication tokens will be used by the system to authenticate user’s request to
backend.

iii. Reporting/Documentation requirements

Use case diagram, system architecture, flowchart and similar diagrams are designed
before the start of development of the project.

Non-functional requirements:

i. Performance: To reduce the memory usage and shorten sentiment analysis

algorithm processing time, the machine learning model is to be serialized into pickle file.

ii. Scalability: The backend of the project is to be hosted in cloud for larger scale
accessibility of the project.

iii. Security: For user’s security reasons, the platform should not store third party
access tokens. Considering it, the tokens will be locally saved in user’s device.

142
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Approval

The platform is designed and developed to fit all the potential users like data
analyst, marketing and news media professionals and social media related job
seeker/provider. It clarifies that this project is not made for any particular client. The
platform is developed keeping in mind to fulfil the professional needs of potential users.
The user interface and database architecture are structured for all the general users of
the platform.

This project is designed to suit the interest of multiple or large-scale users, there is no
any particular client of this project to approve this software requirement specification.
Instead, our supervisors of the project have approved the proposed system to carry out
the development works of this project.

143
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Appendix 4: User Feedbacks

i. Google Play Store Reviews

Figure 111 User feedback from Google Play Store reviews

144
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

ii. User Feedback Survey Report

145
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

146
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

147
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Appendix 5: Weekly/Increment task information for development

(15th Increment)
Week20: Published Flutter application to Google Play Store.

(14th Increment)
Week19: Database migrated from SQLite to PostgreSQL, Deployed API on Heroku, made
live and stable

(13th Increment)
Week18: Vacancy viewing/adding/deleting UI feature and their bloc, fixes widget pixel
oversizing issue in smaller width devices

(12th Increment)
Week17: YouTube, Twitter, manual sentiment analysis UI, report saving feature and their
bloc

(11th Increment)
Week16: Profile, Home Page Menus, Keys Manage and their bloc

(10th Increment)
Week15: Login/Signup Screens with auth BLOC and API request handler for auth task

(9th Increment)
Week14: Implemented sentiment report and vacancy (saving/viewing) backend API part

148
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

(8th Increment)
Week13: User Login, JWT authentication to allow only authenticated users to perform
sentiment analysis, YouTube/Twitter sentiment algorithm linked

(7th Increment)
Week12: Flask backend setup and done up to user registration

(6th Increment)
Week11: Data integration part code using YouTube, Twitter API documentation

(5th Increment)
Week10: Implemented pickle (serializing trained ML model) concept in my system that
increases calculation speed

(4th Increment)
Week9: Made a working sentiment analysis model that works by providing both training,
testing data

(3rd Increment)
Week8: Worked on data pre-processing like tokenizing, stemming, stop words part

(2nd Increment)
Week7: Started to write code for development work after initial level research and learning
NLTK, TextBlob

(1st Increment)
Week6: Initial UI prototype is made ready to implement it properly later after completion
of backend work

Week1-5: Research, interim report, brainstorming on development work, gathering

development work guidance source
149
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Appendix 6: Essential code snippets

i. Backend

from flask import Flask

import os
from flask_jwt_extended import JWTManager
from src.register import register_blueprint
from src.login import login_blueprint
from src.report import report_blueprint
from src.profile import profile_blueprint
from src.vacancy import vacancy_blueprint
from src.youtube import youtube_blueprint
from src.twitter import twitter_blueprint
from src.manual import manual_blueprint
from src.database import db

def create_app(test_config=None):
app = Flask(__name__, instance_relative_config=True)

if test_config is None:
app.config.from_mapping(
SECRET_KEY=os.environ.get("SECRET_KEY"),
SQLALCHEMY_DATABASE_URI=os.environ.get("DATABASE_URI"),
SQLALCHEMY_TRACK_MODIFICATIONS=False,
JWT_SECRET_KEY=os.environ.get('JWT_SECRET_KEY')
)

else:
app.config.from_mapping(test_config)

app.register_blueprint(register_blueprint)
app.register_blueprint(login_blueprint)
app.register_blueprint(report_blueprint)
app.register_blueprint(profile_blueprint)
app.register_blueprint(vacancy_blueprint)
app.register_blueprint(youtube_blueprint)
app.register_blueprint(twitter_blueprint)
app.register_blueprint(manual_blueprint)

# decorator for index page of the backend

@app.route("/")
150
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

def index():
return {
"Registration": "/register/ [POST]",
"Login": "/login/ [POST]",
"User profile": "/profile/ [GET]",
"YouTube comments analysis": "/red/ [POST]",
"Tweets analysis": "/blue/ [POST]",
"User reports": "/report/ [GET/POST]",
"Vacancy": "/vacancy/ [GET/POST/DELETE] &
/vacancy/all?filter=freelancer or vacancy [GET]",
"Sentence/Paragraph polarity": "/manual/ [POST]"
}

db.app = app
db.init_app(app)

JWTManager(app)
# with app.app_context():
# db.create_all()
return app

The above code file is the backend logic that listens the requests received from API

from flask_sqlalchemy import SQLAlchemy

from datetime import datetime

db = SQLAlchemy()

class User(db.Model):
id = db.Column(db.Integer, primary_key=True)
name = db.Column(db.String(55), nullable=False)
email = db.Column(db.String(80), unique=True, nullable=False)
password = db.Column(db.Text(), nullable=False)
purpose = db.Column(db.String(30), nullable=False)
analysis_count = db.Column(db.Integer, default=0)
vacancy_count = db.Column(db.Integer, default=0)
reports = db.relationship('Report', backref="user")
vacancies = db.relationship('Vacancy', backref="user")

def __init__(self, **kwargs):

super().__init__(**kwargs)

def __repr__(self) -> str:

151
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

return 'User>>> {self.email}'

class Report(db.Model):
id = db.Column(db.Integer, primary_key=True)
user_id = db.Column(db.Integer, db.ForeignKey('user.id'))
analysis_datetime = db.Column(db.DateTime, default=datetime.now())
platform = db.Column(db.String(20), nullable=False)
data = db.Column(db.String(200), nullable=False)
data_count = db.Column(db.Integer, nullable=False)
spam_count = db.Column(db.Integer, nullable=False)
sarcasm_count = db.Column(db.Integer, nullable=False)
positive_count = db.Column(db.Integer, nullable=False)
negative_count = db.Column(db.Integer, nullable=False)
neutral_count = db.Column(db.Integer, nullable=False)
overall_polarity = db.Column(db.String(6), nullable=False)

def __init__(self, **kwargs):

super().__init__(**kwargs)

def __repr__(self) -> str:

return 'Report>> {self.data}'

class Vacancy(db.Model):
id = db.Column(db.Integer, primary_key=True)
user_id = db.Column(db.Integer, db.ForeignKey('user.id'))
title = db.Column(db.String(50), nullable=False)
contact_email = db.Column(db.String(80), nullable=False)
contact_number = db.Column(db.String(20), nullable=False)
description = db.Column(db.String(600), nullable=False)
work_hr = db.Column(db.Integer, nullable=False)
wage_hr = db.Column(db.Text, nullable=False)
is_negotiable = db.Column(db.Boolean, default=True, nullable=False)
is_vacancy = db.Column(db.Boolean, default=True, nullable=False)

def __init__(self, **kwargs):

super().__init__(**kwargs)

def __repr__(self) -> str:

return 'Vacanct>> {self.title}'

The above code file contains all the three models of the system

152
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

from flask import Blueprint

from flask_jwt_extended import jwt_required, get_jwt_identity
from src.database import User
from src.functions.field_validators import success_false
from src.senti_bot import notifyProfileView

profile_blueprint = Blueprint('profile', __name__)

@profile_blueprint.route('/profile/', methods=['GET'])
@jwt_required()
def profile():
try:
user_email = get_jwt_identity()
current_user = User.query.filter_by(email=user_email).first()
return {
"success": "true",
"name": current_user.name,
"email": current_user.email,
"purpose": current_user.purpose,
"analysis_count": str(current_user.analysis_count),
"vacancy_count": str(current_user.vacancy_count)}
except:
return success_false()

The above code file handles the user’s profile viewing request

from flask import Blueprint, request

from werkzeug.security import generate_password_hash
from src.database import User, db
from src.functions.field_validators import validate_email, success_false
from flask_jwt_extended import create_access_token

from src.senti_bot import notifyRegisterToAdmin, notifyRegisterTry

register_blueprint = Blueprint('register', __name__)

@register_blueprint.route('/register/', methods=['POST'])
def register():
request_data = request.get_json()
try:
name = request_data['name']
email = request_data['email']

153
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

password = request_data['password']
purpose = request_data['purpose']

# to carry further process only if data in post requests are valid

# first will check if any unwanted format of data is provided
if len(name) > 55:
return success_false(msg="Name should be less than 55chars")
if len(name) < 5:
return success_false(msg="Name should be at least 5chars")
if len(password) < 6:
return success_false(msg="Password should be at least 6chars")
if len(email) > 80:
return success_false(msg="Email should be less than 80chars")
if len(purpose) > 30:
return success_false(msg="Purpose should be less than 30chars")
if not validate_email(email):
return success_false(msg="Please enter a valid email address")

# now moving to database querying if provided registration data format is

correct
if User.query.filter_by(email=email).first() is not None:
# if user with this email already exist
return success_false(msg="This email is already in use")

# if registration request is valid, moving to registration stage from

here
hashed_password = generate_password_hash(password)
user_to_register = User(name=name, email=email,
password=hashed_password, purpose=purpose)
# inserting user data to user table
db.session.add(user_to_register)
db.session.commit() # saving database instance

access_token = create_access_token(
identity=email, expires_delta=False)

# responding user that user is now registered

return {"success": "true", "msg": "User is now registered", "email":
email, "name": name, "access_token": access_token}

except:
return success_false()

The above code file handles the user’s register request

154
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

from flask import Blueprint, request

from flask_jwt_extended import jwt_required, get_jwt_identity
from src.database import User, db
from src.algo.extract_data import youtube_comments_extract
from src.algo.Sentimento import Sentimento
from src.senti_bot import notifyYouTube

youtube_blueprint = Blueprint('youtube', __name__)

@youtube_blueprint.route('/red/', methods=['POST'])
@jwt_required()
def youtube_sentiment():
request_data = request.get_json()
try:
user_email = get_jwt_identity()
current_user = User.query.filter_by(email=user_email).first()
video_id = request_data['video_id']
api_key = request_data['api_key']
approx_comments = 0 # declaring outside so that it can be used at this
space
try:
approx_comments = int(request_data['approx_comments'])
except:
return {"success": "false", "msg": "Approx comments should be
numeric"}

if (len(video_id) < 5 or len(video_id) > 50):

return {"success": "false", "msg": "Please provide valid video
source"}
if (len(api_key) < 5 or len(api_key) > 250):
return {"success": "false", "msg": "Please provide valid key/coupan"}
if ((approx_comments < 20) or (approx_comments > 1000)):
return {"success": "false", "msg": "Comments between 20 to 1000 can
only be set at the moment"}

# after post req validation, moving to youtube sentiment analysis

youtube_data = youtube_comments_extract(
video_id, api_key, approx_comments)

# making object of sentiment result

sentiment = Sentimento(youtube_data[0])

155
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

# after getting sentiment result, need to increase user's analysis count

record

current_user.analysis_count += 1
db.session.commit()

# after adding the user's analysis count, need to return sentiment result
to user
return {
"success": "true",
"msg": "Analysed given data succesfully",
"title": youtube_data[2],
"thumbnail": youtube_data[1],
"total_data": sentiment.data_count(),
"spam_count": sentiment.layer_count()[0],
"sarcasm_count": sentiment.layer_count()[1],
"positive_count": sentiment.overall_polarity()["positive_count"],
"negative_count": sentiment.overall_polarity()["negative_count"],
"neutral_count": sentiment.overall_polarity()["neutral_count"],
"overall_polarity":
round((sentiment.overall_polarity()["overall_polarity"]), 3)
}

except:
return {"success": "false", "msg": "Invalid API request"}

The above code file handles user’s YouTube comment sentiment analysis request

156
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

ii. Frontend

import 'package:flutter/material.dart';
import 'package:flutter_neumorphic/flutter_neumorphic.dart';
import 'package:flutter_secure_storage/flutter_secure_storage.dart';
import 'package:sentimento/screens/analysis_screen/comments_screen.dart';
import 'package:sentimento/screens/analysis_screen/manual_analysis.dart';
import 'package:sentimento/screens/analysis_screen/tweets_screen.dart';
import 'package:sentimento/screens/authscreen_controller.dart';
import 'package:sentimento/screens/getstarted_screen/getstarted_screen.dart';
import 'package:sentimento/screens/home_screen/home_screen.dart';
import 'package:sentimento/screens/login_screen/login_screen.dart';
import 'package:sentimento/screens/mynetwork_screen/post_service.dart';
import 'package:sentimento/screens/mynetwork_screen/post_vacancy.dart';
import 'package:sentimento/screens/mynetwork_screen/user_posts.dart';
import 'package:sentimento/screens/mynetwork_screen/view_freelancers.dart';
import 'package:sentimento/screens/mynetwork_screen/view_vacancies.dart';
import 'package:sentimento/screens/profile_screen/profile_screen.dart';
import 'package:sentimento/screens/register_screen/register_screen.dart';
import 'package:sentimento/screens/reports_screen/reports_screen.dart';
import 'package:sentimento/utilities/routes.dart';

FlutterSecureStorage? secureStorage;

void main() {
//making single instance so that it can be used again by other methods from
multiple pages
secureStorage = const FlutterSecureStorage();

runApp(MaterialApp(
debugShowCheckedModeBanner: false,
theme: ThemeData(primarySwatch: Colors.teal, brightness: Brightness.light),
home: const AuthScreenController(),
routes: {
AppRoutes.getstartedscreen: (context) => const GetStartedScreen(),
AppRoutes.loginscreen: (context) => const LoginScreen(),
AppRoutes.registerscreen: (context) => const RegisterScreen(),
AppRoutes.homescreen: (context) => const HomeScreen(),
AppRoutes.profilescreen: (context) => const ProfileScreen(),
AppRoutes.commentsAnalysis: (context) => const CommentsScreen(),
AppRoutes.tweetsAnalysis: (context) => const TweetsScreen(),
AppRoutes.manualAnalysis: (context) => const ManualAnalysis(),
AppRoutes.reports: (context) => const ReportsScreen(),
157
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

AppRoutes.myPosts: (context) => const UserPosts(),

AppRoutes.viewVacancy: (context) => const ViewVacancies(),
AppRoutes.viewFreelancer: (context) => const ViewFreelancers(),
AppRoutes.postVacancy: (context) => const PostVacancy(),
AppRoutes.postSkill: (context) => const PostService()
},
));
}

Code content of main.dart file

Models used in built Flutter application

Used state management contents

158
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

Screens used in the Flutter application

import 'dart:convert';
import 'package:http/http.dart' as http;
import 'package:sentimento/utilities/securestorage.dart';

const String domain = "https://sentimentofyp.herokuapp.com";

const requestHeader = <String, String>{

'Content-Type': 'application/json; charset=UTF-8',
};

//to get response from backend (POST request)

getResponse(endPoint, header, body) async {
final response =
await http.post(Uri.parse(endPoint), headers: header, body: body);
return response;
}

//to get response from JWT protected POST request

getPostResponse(endPoint, body) async {
String? token = await secureStorage?.read(key: 'access-token');
var headerwithToken = <String, String>{
'Content-Type': 'application/json; charset=UTF-8',
'Authorization': 'Bearer $token',
};
final response = await http.post(Uri.parse(endPoint),
159
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

headers: headerwithToken, body: body);

return response;
}

//to get (GET request)

getGETResponse(endPoint) async {
String? token = await secureStorage?.read(key: 'access-token');
var headerwithToken = <String, String>{
'Content-Type': 'application/json; charset=UTF-8',
'Authorization': 'Bearer $token',
};
final response =
await http.get(Uri.parse(endPoint), headers: headerwithToken);
return response;
}

//to handle post/vacancy delete request of user

getDELETEResponse(endPoint) async {
String? token = await secureStorage?.read(key: 'access-token');
var headerwithToken = <String, String>{
'Content-Type': 'application/json; charset=UTF-8',
'Authorization': 'Bearer $token',
};
final response =
await http.delete(Uri.parse(endPoint), headers: headerwithToken);
return response;
}

//to respond back to bloc

blocResponder(response) {
if (response.statusCode == 200) {
var decodedResponse = json.decode(response.body);
return decodedResponse;
}
}

import 'package:flutter/material.dart';
import 'package:flutter_bloc/flutter_bloc.dart';
import 'package:sentimento/bloc/auth_bloc/auth_bloc.dart';
import 'package:sentimento/screens/getstarted_screen/getstarted_screen.dart';
import 'package:sentimento/screens/home_screen/home_screen.dart';
import 'package:sentimento/widgets/minor_ui_parts.dart';
160
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

class AuthScreenController extends StatelessWidget {

const AuthScreenController({Key? key}) : super(key: key);

@override
Widget build(BuildContext context) {
return BlocProvider(
create: (context) => AuthBloc(LoadingState())..add(SessionCheckerEvent()),
child: BlocBuilder<AuthBloc, AuthState>(
builder: (context, state) {
if (state is LoggedInState) {
return const HomeScreen();
} else if (state is MessageState) {
manualToastMsg(state.message);
} else if (state is LoggedOutState) {
return const GetStartedScreen();
}
return showLoading(context);
},
),
);
}
}

Dart functions to communicate with backend API’s

import 'package:flutter/material.dart';
import 'package:flutter_bloc/flutter_bloc.dart';
import 'package:sentimento/bloc/auth_bloc/auth_bloc.dart';
import 'package:sentimento/screens/getstarted_screen/getstarted_screen.dart';
import 'package:sentimento/screens/home_screen/home_screen.dart';
import 'package:sentimento/widgets/minor_ui_parts.dart';

class AuthScreenController extends StatelessWidget {

const AuthScreenController({Key? key}) : super(key: key);

@override
Widget build(BuildContext context) {
return BlocProvider(
create: (context) => AuthBloc(LoadingState())..add(SessionCheckerEvent()),
child: BlocBuilder<AuthBloc, AuthState>(
builder: (context, state) {
if (state is LoggedInState) {
return const HomeScreen();
161
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

} else if (state is MessageState) {

manualToastMsg(state.message);
} else if (state is LoggedOutState) {
return const GetStartedScreen();
}
return showLoading(context);
},
),
);
}
}

Dart file that controls the navigation when application is opened.

class TextFieldValidator {
static emailValidator(email) {
bool emailValid = RegExp(
r"^[a-zA-Z0-9.a-zA-Z0-9.!#$%&'*+-/=?^_`{|}~]+@[a-zA-Z0-9]+\.[a-zA-
Z]+")
.hasMatch(email);
if (emailValid) {
return null;
} else {
return "Enter valid email address";
}
}

static passwordValidator(password) {
if (password!.isEmpty || password.length < 6) {
return "Password length must be atleast 6";
}

return null;
}

static nameValidator(name) {
if (name!.isEmpty) {
return "Name cannot be empty";
}
if (name.length < 5) {
return "Name should be more than 5chars";
}
if (name.length > 55) {
return "Name should be less than 55chars";
162
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

}
}

static keyValidator(key) {
if (key!.isEmpty) {
return "Key/coupan cannot be empty";
}
if (key.length < 5) {
return "Key/coupan should be more than 5chars";
}
if (key.length > 250) {
return "Key/coupan should be less than 250chars";
}
}

static dataSourceValidator(source, platform) {

if (platform == "youtube") {
if (source!.isEmpty) {
return "Video url cannot be empty";
}
if (source.length < 5) {
return "Url cannot be this small";
}
if (source.length > 250) {
return "Url cannot be longer than 250chars";
}
}
if (platform == "twitter") {
if (source!.isEmpty) {
return "Tweet topic cannot be empty";
}
if (source.length < 2) {
return "Tweet topic cannot be this small";
}
if (source.length > 250) {
return "Tweet topic be longer than 250chars";
}
}
}

static numOfDataValidator(dataCount) {
try {
int approx = int.parse(dataCount);
if ((approx < 20) || (approx > 1000)) {
return "Range of 20 to 1000 is only allowed for now.";
163
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

}
} catch (e) {
return "Enter numeric values only";
}
}

static roleValidator(role) {
if ((role.length < 3) || (role.length > 50)) {
return "Length of role should be between 3 to 50";
}
}

static phoneNumberValidator(number) {
if ((number.length < 5) || (number.length > 20)) {
return "Length of phone number should be between 5 to 20";
}
}

static descriptionValidator(desc) {
if (desc.length < 15) {
return "Description should be atleast 15chars in length";
}
}

static workHrValidator(hr) {
try {
int hrs = int.parse(hr);
if ((hrs < 1) || (hrs > 8)) {
return "Working hour should be in range of 1 to 8";
}
} catch (e) {
return "Enter numeric values only";
}
}

static wageValidator(wage) {
if ((wage.length < 2) || (wage.length > 15)) {
return "Invalid wage/hour format";
}
}
}

Functions to validate text fields

164
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

import 'dart:convert';
import 'api_universal.dart';

const String loginEndPoint = "$domain/login/";

const String registerEndPoint = "$domain/register/";

Future requestProfile() async {

return blocResponder(await getGETResponse("$domain/profile/"));
}

Future requestLogin(String email, String password) async {

var requestBody =
jsonEncode(<String, String>{"email": email, "password": password});
return blocResponder(
await getResponse(loginEndPoint, requestHeader, requestBody));
}

Future requestRegister(name, email, password, purpose) async {

var requestBody = jsonEncode(<String, String>{
"name": name,
"purpose": purpose,
"email": email,
"password": password
});
return blocResponder(
await getResponse(registerEndPoint, requestHeader, requestBody));
}

Asynchronous functions to perform login, register http request

165
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

import 'dart:convert';
import 'package:sentimento/models/post_model.dart';
import 'package:sentimento/widgets/minor_ui_parts.dart';

import 'api_universal.dart';

const String vacancyEndPoint = "$domain/vacancy/";

const String deletePostEndPoint = "$domain/vacancy/?id=";
const String vacanciesEndPoint = "$domain/vacancy/all/?filter=vacancy";
const String freelancersEndPoint = "$domain/vacancy/all/?filter=freelancer";

Future requestAllPosts(filter) async {

if (filter == "vacancy") {
var response = blocResponder(await getGETResponse(vacanciesEndPoint));
return PostModel.fromJson(response);
} else {
var response = blocResponder(await getGETResponse(freelancersEndPoint));
return PostModel.fromJson(response);
}
}

Future requestUserPosts() async {

var response = blocResponder(await getGETResponse(vacancyEndPoint));
return PostModel.fromJson(response);
}

Future addNewPost(Vacancy post) async {

var requestBody = jsonEncode(<String, dynamic>{
"title": post.title,
"contact_email": post.contactEmail,
"contact_number": post.contactNumber,
"description": post.description,
"work_hr": post.workHr,
"wage_hr": post.wageHr,
"is_negotiable": post.isNegotiable,
"is_vacancy": post.isVacancy
});
var response =
blocResponder(await getPostResponse(vacancyEndPoint, requestBody));
manualToastMsg(response["msg"]);

return response;
}

Future deleteSelectedPost(postId) async {

166
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

var deleteResponse =
blocResponder(await getDELETEResponse("$deletePostEndPoint$postId"));
manualToastMsg(deleteResponse["msg"]);

return deleteResponse;
}

Asynchronous functions to perform vacancy related http requests

import 'dart:async';
import 'package:bloc/bloc.dart';
import 'package:equatable/equatable.dart';
import 'package:sentimento/api/auth_requests.dart';
import 'package:sentimento/utilities/securestorage.dart';
part 'auth_event.dart';
part 'auth_state.dart';

class AuthBloc extends Bloc<AuthEvent, AuthState> {

AuthBloc(LoadingState loadingState) : super(LoadingState()) {
on<SessionCheckerEvent>(_sessionChecker);
on<LoginClickedEvent>(_loginClick);
on<RegisterClickedEvent>(_registerClick);
on<InitialEvent>(_initialEvent);
}

FutureOr<void> _initialEvent(
InitialEvent initialEvent, Emitter<AuthState> emit) {
emit.call(LoggedOutState());
}

FutureOr<void> _sessionChecker(
SessionCheckerEvent sessionClickedEvent, Emitter<AuthState> emit) async {
emit.call(LoadingState());
//checks if bearer token is saved or not
bool userLoggedIn = await isLoggedIn();
if (userLoggedIn) {
emit.call(LoggedInState());
} else if (!userLoggedIn) {
emit.call(LoggedOutState());
}
}

FutureOr<void> _loginClick(
LoginClickedEvent loginClickedEvent, Emitter<AuthState> emit) async {
167
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

emit.call(LoadingState());
try {
var loginResponse = await requestLogin(
loginClickedEvent.email, loginClickedEvent.password);
authValidatior(loginResponse, emit);
} catch (e) {
emit.call(MessageState(message: "No internet or try again"));
}
}

FutureOr<void> _registerClick(RegisterClickedEvent registerClickedEvent,

Emitter<AuthState> emit) async {
emit.call(LoadingState());
try {
var registerResponse = await requestRegister(
registerClickedEvent.fullName,
registerClickedEvent.email,
registerClickedEvent.password,
registerClickedEvent.purpose);
authValidatior(registerResponse, emit);
} catch (e) {
emit.call(MessageState(message: "No internet or try again"));
}
}

void authValidatior(backendResponse, Emitter<AuthState> emit) async {

if (backendResponse["success"] == "true") {
//when login request is valid, will save token and take user to home screen
saveUserInfo(backendResponse["access_token"], backendResponse["name"],
backendResponse["email"]);
emit.call(MessageState(message: backendResponse["msg"]));
//await Future.delayed(const Duration(milliseconds: 500));
emit.call(LoggedInState());
} else {
emit.call(MessageState(message: backendResponse["msg"]));
// emit.call(LoggedOutState());
}
}
}

User authentication related state management

168
PRASHANT GHIMIRE
CS6PO5NI FINAL YEAR PROJECT

class AppRoutes {
static String getstartedscreen = "/getstarted";
static String loginscreen = "/login";
static String registerscreen = "/register";
static String splashscreen = "/splash";
static String homescreen = "/home";
static String profilescreen = "/profile";
static String commentsAnalysis = "/comments";
static String tweetsAnalysis = "/tweets";
static String manualAnalysis = "/manual";
static String analysisResult = "/result";
static String reports = "/reports";
static String postVacancy = "/post/vacancy";
static String postSkill = "/post/skill";
static String viewVacancy = "/view/vacancies";
static String viewFreelancer = "/view/freelancers";
static String myPosts = "/myposts";
}

Maintained static routes to access easily

169
PRASHANT GHIMIRE