You are on page 1of 23

Fake News Prediction of

Text and Image with


Report Generation
Guide: By:
Dr.G.Arun Sampaul Thomas Associate K. Kavya
professor HT NO :19671D5803
JBIET JBIET
Content

Abstract 
Hardware requirements

Objectives 
Workflow

Existing system 
Scop

Proposed system 
Limitations

Approach 
System architechure

Software requirements 
Uml diagram
Abstract
• Internet news consumption is increasing day by day, due to
its rapid spread of information and its easy access

• which lead people to consume news rapidly without the


knowledge of whether the news is false or true.

• it leads to the wide spread of false news which leads to the


negative impacts on society.
• We will build machine learning model which identifies the
text and images, whether the news is false or true.

• Ones the news is predicted fake, a report is redirected to the


authorized website (cybercrime department) to take the
immediate necessary actions required to stop these news
from spreading.
Objectives

To study on how Machine Learning can classify the data
collected into two categories.

To develop a Support Vector Machine classification
model based on collected data from social media.

To evaluatethe accuracy of developed model.
Existing System
• we have Compare the news with authorized news
channels/Websites , of whether the news is false or true.
• A comprehensive overview of existing security vulnerabilities.
• Critical analysis of the state-of-the-art mitigation techniques and their
pros and cons
Existing System


The existing methodology maybe sufficient for detecting plagiarism
of images when the source and suspected image have not been rotated
by a large margin, but in case of rotational changes the existing
methodology will fail .

The proposed methodology will ensure that even if the image is
rotated plagiarism is detected if it has occurred or if an attack of
rotational change has been made rotational change has been made.
Proposed system

We will build machine learning model which identifies the
text and images, whether the news is false or true.

Analysis of new cyber attack patterns in emerging
technologies .

Potential future research directions in cyber security.
Approach
• This model works on processing the text and images together by
providing an interactive Application Interface (API)

• Text by applying the model Logistic regression classifier.

• Image by applying self-consistency algorithm. The natural language


tool kit (NLTK) model is used for these implementation through
python
System requirements
• Software Requirements
• Python Anaconda spider IDE
• Python Machine learning Libararies (Logistic regression classifier, NLP), numpy,
pandas , scrikt-learn , Keras ..etc

• Hardware Requirements
• Operation System : Windows/Mac/Linux
• Memory RAM 1GB , Hard disk 20GB
Scope

Social Media. 
1

Training SVM algorithm.

Data recognition.

Classification of data.
Limitations

Collecting data from social
media.

Classifying text from single
post.
Data model flow

Text Features of News Verification

The research paper now a days present the prediction of
fake-news using URL and tweet based text features.

Image Features of News Verification

Nowadays images in cyberspace are popular in
forms of posts, memes etc. along with the
description of text. Image features of news
prediction still exists in basic level of research.

DATASET AND METHODS

Description of Dataset


The Cyberspace news prediction of multimedia content is fairly
a new problem. Therefore there are only less number of data sets
that are public.

Usually the news prediction can be classified based on the
accuracy of news as completely-true, completely-false,
obscure-true, obscure-false.

Methods

There are various models involved in implementation of each
module starting from data pre-processing to classification of news.
System Architecure

Fig.1. System Architecture


UML DIAGRAM

Use case diagram


New User Registration
New User Login and Testing
Prediction of Text
Testing the image
Image tested as to give grafical representation
THANKS YOU

You might also like