Predicting cyber bullying on t witter using machine learning

PREDICTING CYBER
BULLYING ON TWITTER
USING MACHINE LEARNING
U N D E R T H E G U I D A N C E O F ,
D R . E . I L A V A R A S A N , M . T E C H . , P H . D , P R O F E S S O R
P O N D I C H E R R Y E N G I N E E R I N G C O L L E G E
T E A M M E M B E R S :
Z A H I D S H A B A N M I R ( 2 0 C S 3 3 9 )
J O H N P A U L A ( 2 0 C S 3 3 1 )

AGENDA
Introduction
Problem Definition
Objectives of the proposed system
Literature Review
Proposed System
Current Status of the project
Hardware and Software requirements
Conclusion
References

INTRODUCTION
In today’s world which has been made smaller by technology,
new age problems have been born.
No doubt technology has a lot of benefits; however, it also comes
with a negative side. It has given birth to cyberbullying.
To put it simply, cyberbullying refers to the misuse of information
technology with the intention to harass others. Subsequently,
cyberbullying comes in various forms.
It doesn’t necessarily mean hacking someone’s profiles or posing
to be someone else.

INTRODUCTION
It also includes posting negative comments about somebody
or spreading rumors to defame someone.
As everyone is caught up on the social network, it makes it
very easy for anyone to misuse this access.
For this reason prediction of cyberbullying has become area
of research.

PROBLEM DEFINITION
In the last few years, online communication has shifted toward user-
driven technologies, such as SM websites, blogs, online virtual
communities, and online sharing platforms.
New forms of aggression and violence emerge exclusively online. The
dramatic increase in negative human behavior on SM, with high
increments in aggressive behavior, presents a new challenge.
Consequently, developing a cyber bullying prediction model that
detects aggressive behavior that is related to the security of human
beings is more important than developing a prediction model for
aggressive behavior related to the security of machines.

OBJECTIVES OF THE PROPOSED SYSTEM:
To propose our project NLP (Natural Language Processing) and
Machine learning techniques for Cyberbullying prediction in social
media.
•By using NLP, get lot of features like sentiment analysis, pos-
tagging, noun phrase extraction, etc

Merits

OBJECTIVES OF THE PROPOSED SYSTEM:
A word can appear as a verb in a sentence, whereas in other sentences, it can
appear as an adverb. Also, based on the sentence structure, a word can be in
the present, past, or past participle form.
If we consider including all these words instead of stemming these words into
their root form as part of our model, then dictionary size will be huge, and also,
the model will perform a massive amount of unnecessary calculation, which is
an inefficient use of the memory.
So, ultimately the model will take more time during the training process. To
avoid all these, we perform stemming operations to convert each word into its
root form.

LIMITATIONS OF THE EXISTING SYSTEM/RESEARCH
ISSUES:
Cyberbullying prediction based on Content-based features of
various machine learning methods.
Increasing over fitting risk when the number of observations is
insufficient
Over fitting is not reduced.
The given paper uses only text message as input.
Demerits:

LIMITATIONS OF THE EXISTING SYSTEM/RESEARCH
ISSUES:
Training patterns cannot be updated dynamically .It has to be
manually done every time
The model takes more time during the training process to
convert the each words to itsroot form.
The given model does not deal with the real time problem where
now days usage of words has modified people tend to use emoji
and emoticons rather than typing words in such cases the
proposed model does not have all the above mentioned
features.

Existing System
Collect input tweets from the tweeter.
Preprocessing of input comments – remove
duplicates, stop words, punctuations.
Train the model using SVM and hyperparamater
tunning is also done.
Visualize the aspects of the given model.
1.
2.
3.
4.

THE PROPOSED SYSTEM
1.Collect input(text,audio) tweets from the tweeter.
2.Coversion of emoji and audio to text is done.
3.Preprocessing of input comments – remove duplicates,
stop words, punctuations.
4.Train the model using RNN and hyperparamater tunning is
also done.
5.Visualize the aspects of the given model.

ARCHITECTURE OF PROPOSED SYSTEM

INPUT DATA
Text
Text with emoji’s
Audio file
Emoji’s are converted
into text
After removal of noise
audio files are
converted into text.

DATA PREPROCESSING
•Tokenization(chopping it up into pieces)
•Removing Stop Word(filtering of words)
•Stemming and Lemmitazation (generation of root words)
•GLoVe(Global Vectors for Word Representation)

RNN
Recurrent Neural Network

• Deep learning
•Sequential data
•Output from previous step
are fed as input to the current
step.
•Important feature of RNN is
Hidden state, which
remembers some information
about a sequence.

Current Status of the project
•Collected the Datasets
•Working with the datasets

Input and Output
Example for cyber bullying or non- cyber bullying:

Cyber bullying
What.. you are an idiot?.
Non- cyber bullying
That person is a hero to us all.

Hardware and Software requirements
Environment:
•Software Requirements:
•Cloud Tool: Anaconda
•Operating system: 64 Bit OS (Windows 7/ 8 / 10) / Linux
•Scripting : Python
•IDE : Jupyter notebook,Google colab or Spyder.
Hardware Requirements:
•RAM: Minimum 4GB
•ROM: 20 GB minimum

CONCLUSION
Considering that social media has become the most common way to communicate,
cyberbullying is the most common method of bullying.
It is important to delete cyberbullying because it is plaguing every user, including
our next generation .
The situation with cyber bullying is getting only worse, as the incidents of
depression and suicide are only increasing.In this we use NLP (Natural Language
Processing) and Machine learning techniques for Cyberbullying prediction in social
media.
We believe that the current study will provide crucial details on and new directions
in the field of detecting aggressive human behavior, including cyberbullying
detection in online social networking sites.

Reference
[1]. Mohammed ali al-garadi, Mohammad rashid hussain , Nawsher khan , Ghulam murtaza,
Henry friday nweke, Ihsan ali , Ghulam mujtaba, Haruna chiroma , Hasan ali khattak , and
Abdullah gani .“Predicting Cyberbullying on Social Media in the Big Data Era Using Machine
Learning Algorithms: Review of Literature and Open Challenges. vol.7, pp. 70701 - 70718, 22
May 2019,doi.10.1109/ACCESS.2019.2918354.
[2]. Waleed Mugahed Al-Rahmi ; Noraffandy Yahaya ; Mahdi M. Alamri,” A Model of Factors
Affecting Cyber Bullying Behaviors Among University Students. vol. 7, pp. 153417– 153431,
October. 2019,doi. 10.1109/ACCESS.2018.2881292.
[3].Norah Basheer Alotaibi :” Cyber Bullying and the Expected Consequences on the Students’
Academic Achievement”. vol. 7, pp.153417 - 153431, 16 October 2019, doi.
10.1109/ACCESS.2019.2947163.

Reference
[4]. Pete Burnap1 and Matthew L Williams,” Us and them: identifying cyber hate on Twitter
across multiple protected characteristics”, Burnap and Williams ,. EPJ Data Science vol. 5,
Article number: 11, 23 March 2016,doi. 10.1140/epjds/s13688-016-0072-6.
•
[5]. Egundo Moisés Toapanta, John Angelo Recalde Monar, Luis Mafla Gallegos,”Prototype to
Perform Audit in Social Networks to Determine Cyberbullying”. 01 October
2020,doi.10.1109/WorldS450073.2020.9210335.
•
[6]. Saloni Wade1, Maithili Parulekar2, Prof. Kumud Wasnik,” Cyber Bullying Detection on
Twitter Mining” International research journal of engineering and technology (IRJET) vol. 07,
issue: 07 , July 2020.

Predicting cyber bullying on t witter using machine learning

More Related Content

What's hot (20)

Similar to Predicting cyber bullying on t witter using machine learning (20)

Recently uploaded (20)

Predicting cyber bullying on t witter using machine learning