ANALYSIS OF FAKE NEWS DETECTION USING

MACHINE LEARNING
Team Members Project Guide
Amita V (711518104003) Dr. C. Deepa
Head of the Department AI & DS

Hari Priya S (711518104021)

Pranesh R (711518104038)
Shree Varshan .S.R (711518104053)

Overview
● Objective
● Abstract
● Domain
● Introduction
● Literature Survey
● Existing System Disadvantages
● Proposed System
● Proposed System Advantages

Objective:
● The main objective is to detect the fake news, which is a classic text classification
problem with a straightforward proposition.

● It is needed to build a model that can differentiate between Real news and Fake news
using Machine Learning Algorithm

Abstract:
  Internet is one of the important inventions and a large number of persons are its users.
These persons use this for different purposes.
 There are different social media platforms that are accessible to these users. Any user
can make a post or spread the news through these online platforms.
 These platforms do not verify the users or their posts. So some of the users try to
spread fake news through these platforms.
 These fake news can be a propaganda against an individual, society, organization or
political party. A human being is unable to detect all these fake news. So there is a
need for machine learning classifiers that can detect these fake news automatically.
Use of machine learning classifiers for detecting the fake news is described in this
systematic literature review.

Domain:
● Machine learning (ML) is the study of computer algorithms that can improve
automatically through experience and by the use of data.

● Machine learning is a branch of artificial intelligence (AI) and computer science

which focuses on the use of data and algorithms to imitate the way that humans learn,
gradually improving its accuracy

● In this project we use hybrid machine learning(hyper tunning)

Introduction:
 This project aims at learning the detection of pretend news victimization machine
learning program in python. It uses linguistic communication process for sleuthing the
pretend news.

 A model is constructed supported the count vectorizer or a tf-idf matrix word tallies
relative to however usually they're utilized in alternative articles within the dataset
used will facilitate.

 Since this downside may be a quite text classification, implementing a Naive

mathematician classifier and Support vector machine are best as this can be
commonplace for text-based process.
Literature Survey:
S.NO AUTHOR TITLE ALGORITHM DISADVANTAGE
1 W. Vorhies The state of the Detect fake news BERT(Bidirectional Encoder
art(2020) Representations from
Transformers) is large because of
the training structure and corpus. It
is slow to train

2 William Yang liar, liar pants on fire A new benchmark For Hybrid CNNs (Text + All) the
Wang (2017) dataset for fake news Test Accuracy is 0.274 only
detection

3 Granik. M, & Fake news Naive Bayes classifier It assumes that all the features are
Mesyura detection(2017) independent
4 Hunt Allcott and Social media and fake In Journal of This may Change in Public
Matthew Gentz news in the 2016 Economic Perspective Opinion
kow election

Existing System:
● There exists a large body of research on the topic of machine learning methods for
deception detection, most of it has been focusing on classifying online reviews and
publicly available social media posts.

● Particularly since late 2016 during the American Presidential election, the question of
determining 'fake news' has also been the subject of particular attention within the
literature.

Disadvantages of Existing System:

● In existing system they classify the news, only based on the content related
ngrams and shallow part-of-speech.
● It’s hard to prove that a news is fake or not by analysing the tagging.even if the
news is genuine it is detected as fake news
● It seems to be very complicated when a genuine news is published on the social
media
● Failed to classify the valuable news

Proposed System:
● In this project we are going to make use of Natural Language Techniques to overcome the
widespread of false news on the internet.
● In this project we make use of deep learning techniques to determine how the LSTM algorithm
works on the given clip of information which can be given as input to the system either in text
format or image format.
● The approach used in this project is to first train the system and then add the news information for
which one needs to check if its reliable or not reliable as well as print the accuracy of the algorithm
performance on the news clip inserted by the respective reader.
● The dataset used in this process contains 25117 entries collected from kraggle and github , which
goes across 4 columns named as the journalist id which will be provided by their particular
newspaper organisation , the headline of the news which is given as title ranges around 15 to 30
words at most the content of the news which is tagged as text and finally the label that is either 1 or
0 which means that I indicates that the news is true and 0 indicates that the news is false,

Continue reading…
● These datasets are checked and published by journalists who also make sure these information have
high level of authenticity News in these datasets contain news from domains like
sports,politics ,health ,etc.
● After this the training dataset is processed and after processing it the users has to choose which format
of news are they willing to use that is either text format or image format.
● When the reader chooses the text format they can copy the news from the website and paste it on the
prompt displayed on the window and then click ok.
● In case of image format the user has to snip the image from the website and save it in the system then
upload the image and click ok,the image will be converted into a grayscale image and will be displayed
before processing the contents in it.
● On processing the contents in the image will be converted into text After this process is done the
confusion matrix is displayed which contains the true positive value,false positive value,false negative
value and true negative value as in a graph format.
Advantages of Proposed System:
● In proposed system the news is classified according to the text that are continuously
Appeared in a news.
● We apply random forest algorithm in our proposed system based on count vectorizer
and tf-idf vectorizer so the frequently used words are detected and removed and
classified using random forest
● Prediction accuracy will be higher comparing to the existing system.