Basic Introduction To Fake News Detection
Uploaded by
IJRASETPublicationsBasic Introduction To Fake News Detection
Uploaded by
IJRASETPublications9 III March 2021
https://doi.org/10.22214/ijraset.2021.33360
International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.429
Volume 9 Issue III Mar 2021- Available at www.ijraset.com
Basic Introduction to Fake News Detection
Swethasubrahmanian P1, Sruthy MR2, Biji KP3
1
Department of Computer Science and Engineering, APJ Abdul Kalam Technological University
2, 3
Assistant Professor Department of Computer Science and Engineering, Nehru collage of Engineering and Research Centre
Pampady
Abstract: The present world is run around a lot of news, but the genunety of news we need to analyse. In this paper, we analyse
what is fake news and what are the methods that is used for the processing of fake news. The fake news classification and
analysis is a big task because it requires huge data and information processing knowledge. This paper provides a basic
understanding of the fake news detection algorithm and also analyse various machine learning approaches.
Keywords: Fake news detection, Machine learning, Data mining.
I. INTRODUCTION
This paper is an introduction of our research topic. Before the talk about a fake news detection methods and algorithm need to
analyse what are fake news, type of fake news how each type of fake one is processed. This paper is a basic introduction of fake
news processing. Consider the example of covid 19 spread, Sometimes, the news that spread through various social media and web
site also have negative shade that means may or may not be true. At the same time we also need to valuate two to three article for
finding the gen unity of news that we heard. We know technologies are become advanced day by day, so, we need to update ourself
with it. Also, the news and information that we here have many faces. Throughout this paper We are guiding you the step towards
towards fake news detection strategies and methods. We can analyse the topic in four section. The first section provide what is fake
news and its scientific definition and examples. The second section provides type of fake news, the third section explains the
methods, and algorithms.
II. FAKE NEWS ANALYSIS
A. What Is Fake News
As a beginner, I need to move through the topic step by step manner. So from my observation the fake news is nothing but someone
or something created a false data or information regarding something in order to mislead the reader for their needs.
B. Type Of Fake News
Fake news is misleading information presented in the form of news with the aim of destroying the reputation of a person or an
organization. There is a huge verity of fake news, that depend on the mode of transmission and nature of news.
1) One type of fake news is the ability to fool and may be misinterpreted as a fact. It does not necessarily cause negative effect
may be just for an entertainment purpose.
2) Another one is a false content made by someone without any detailed study of that topic. It may be due to the unawareness of
that person, that person may be a journalist or a social media influencer. That may or may not be focus on profit.
3) For example, sometimes on social media like YouTube we may read a headline or thumbnail with some news but while
clicking it says something different from the heading.
4) Next one is purposeful doing something which cause misleading the information Next one is spreading false content along with
real one which destroy the intention of real one.
C. Fake News Detection Method
Based on the study of fake news detection method i found that fake news can be detected in different perspectives.
a) Depending Upon the knowledge: It focuses on the knowledge inside the content whether it is true or not. The analysis is based
on the fact checking basis. We can further classify it as automatic (by human, expert)and manual(by various algorithms).
b) Based on Syntax: We can use syntax analysis method for fake news detection. That means Write the sentences into parse tree
and further processing
c) Based on path- This model analyses the news on media by checking the path which spreading. The main feature of this method
is build a time series with neural network(CNN/RNN). It also uses clustering methods like agglomeration with KNN.
d) Predictive Model Based Method: This method is based on the probability of outcome. The common predictive algorithms are
fall under machine learning and deep learning.
©IJRASET: All Rights are Reserved 874
International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.429
Volume 9 Issue III Mar 2021- Available at www.ijraset.com
III. ALGORITHMS FOR FAKE NEWS DETECTION
A. Logistic Regression
Logistic regression come under the regression model. It is a predictive model based technique. LR analyses the relationship between
dependent variable, one independent variable and series of dependent variable. We can use one or more dependent or independent
variable in order to predict outcome of some result. In shot a logistic regression is a classification method it will analyse and predict
the value either true or not, that is a binary model. It is based on probability prediction and classification. For example whether the
transaction successful or not in this case, we can use logistic regression model. There are three type of logistic regression model,
binary ordinal, multinomial.
B. Support Vector Machine Learning
Support vector is a machine learning approach. It follows supervised machine learning model. The SVM algorithm, we imagine data
item are point n-dimensional space it represents value of a particular coordinate. The classification is based on hyperplane. Hyper-
plane that differentiates the two classes very well.
C. Multy Level Perceptron
Another classification algorithms used for the fake news detection are multilayer perceptron. Neural network based algorithm is
more suitable for complicated tasks or for the complicated data classification. Neural network is a group of neurons associated with
their weight. The multilayer perceptron classifies data based on MLP model that contains input layer and output layer in between
number of hidden layers for filtering and classification. Other than input node uses nonlinear activation function. It uses back-
propagation method for training neural network, run under supervised machine learning method.
D. KNN
The KNN(k-nearest Neighbours)is a simple algorithm compare to other algorithms we discussed earlier. It is an easy way to
implement a supervised machine learning method. We can uses it for both classification and regression. In these cases, the input
contain the k closest training examples in data set. The output is based on nature of KNN, whether it is for regression or for
classification. KNN will calculate the euclidean distance of each(k number of neighbours)and then choose the k nearest neighbour
as per the euclidean distance. Then it tries to count the number of data in each section (category). Finally, it allocates the new data
points to that category based on the maximum number of neighbours.
E. Random Forest
Random forest is run under the supervised learning algorithm. it's almost like the choice tree. it's trained by the bagging method. The
bagging method is that a mixture of learning models increases the general results of a system. used for both classification and
regression. Random forest may be a way of averaging multiple deep decision trees, trained on different parts of an equivalent
training set, for reducing the variance. Bagging cause better model performance because it decreases the variance of the model, with
none change in bias. this suggests that while the predictions of one tree is very sensitive to noise on It's training set, the typical of
the many trees aren't , as long because the trees aren't correlated. Simply training many trees on one training set would give strongly
correlated trees bootstrap sampling may be a way of de-correlating the trees by showing them different training sets.
F. LSTM
LSTM is the abbreviation of long short term memory. It will overcome the limitation of RNN. The main disadvantage of RNN is
vanishing gradient problem. The RNN is used for short sequence and short time period, it is not suitable for long sequence and long
time period. In order to overcome vanishing gradient problem LSTM is used. There are manly three part for LSTM forget gate,
output gate, input gate. The LSTM process data in sequence manner. LSTM chain contains four neural networks and different
memory blocks. Some information are may not be useful, so, they are removed by forget gate. Adding useful information into cell
by input gate. The final step is filter out useful information from the current cell , and presented as an output is by output gate.
IV. CONCLUSIONS
This paper is a basic introduction of fake news detection methods.i analyses what is fake news from different perspectives, what are
the methods used for fake news analysis ans finally come up with various machine learning algorithms used for fake news detection.
©IJRASET: All Rights are Reserved 875
International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.429
Volume 9 Issue III Mar 2021- Available at www.ijraset.com
REFERENCES
[1] Akshay Jain” fake news detection” Department of Electronics and Communication Engineering Maulana Azad National Institute of Technology Bhopal, India
[2] Fake News Detection on Social Media: A Systematic Survey Mohamed K. Elhadad ,Kin Fun Li Fayez Gebali Dept. of Elec. and Comp. Eng. University of
Victoria Victoria, Canada [email protected]
[3] Shivam B. Parikh, Vikram Patil and Pradeep K. Atrey “On the Origin, Proliferation and Tone of Fake News “
[4] Hemant Kumar Gianey , Rishabh Choudhary “Comprehensive Review On Supervised Machine Learning Algorithms “
[5] Abdullah-All-Tanvir ,Ehesas Mia Mahir, Saima Akhter, Mohammad Rezwanul Huq “Detecting Fake News using Machine Learning and Deep Learning
Algorithms” Dept. of Computer Science and Engineering
