Fake News Detection Using Machine

Learning
 Introduction
• Fake news has become extremely popular in recent years and is widely
used in social networking. The Internet has enabled an unprecedented
number and speed of news exchanges between users. With the
exponential growth of social media, information spreads among
individuals at an unprecedented amount and speed.
• Verifying the news has effective ways to combat misinformation.
Facilitating online data verification is an effective strategy to deal with
the often overwhelming amount of misinformation.

Aug 14, 2024 Department of Computer Engineering, International Institute of Information Technology, Hinjawadi, Pune-411057 2
 Motivation

Fake news has become a pervasive issue, especially on social

media platforms.
By detecting and classifying fake news we can -
• Improve media literacy
• Protect the public from misinformation
• Facilitate easier fact-checking.

Aug 14, 2024 Department of Computer Engineering, International Institute of Information Technology, Hinjawadi, Pune-411057 3
 Literature Survey
S.N. Title of Paper Journal/ Author Names Summary of Paper (
Conference algorithm/methodology used,
Name/Year of results obtained etc..)
publication

1 Fake News Detection using Machine ELSEVIER 2024 Jumana Jouhar, Anju The study analyzed machine learning algorithms for fake news
Learning and Python Pratap*, Neharin Tijo, detection, with XGBoost being the most effective. It
Meenakshi Mony demonstrated the lowest false positives and true negatives,
minimal overfitting, and excellent generalization.Despite the
literature review's expectations, alternative models surpassed
PAC in this study due to factors such as complexity, handling
non-linear relationships, ensemble learning, regularization
techniques, and dataset characteristics. The research highlights
the need for future research to adapt to the dynamic nature of
fake news dissemination, including social media, and to improve
the model's efficacy.

2 Rapid detection of fake news based on ELSEVIER 2021 Barbara Probierza,∗ , This paper developed a new model to quickly discover fake
machine learning methods Piotr Stefanski ´ a , Jan news based on news titles without analyzing the entire text. It
Kozaka used NLP techniques, machine learning algorithms, single
classifiers, and ensemble methods for prediction. The model
achieved good results after initial analysis, but running time was
a problem. Future work may involve multi-criteria models.

Aug 14, 2024 Department of Computer Engineering, International Institute of Information Technology, Hinjawadi, Pune-411057 4
 Literature Survey
S.N. Title of Paper Journal/ Author Summary of Paper
Conference Names ( algorithm/methodology used,
Name/Year results obtained etc..)
of
publication

3 Detection of fake news from social ELSEVIER 2024 M. Sudhakar * , K.P. Online social networks are spreading fake news, posing
media using support vector machine Kaliyamurthie global problems, affecting the economy, and causing
learning algorithms political panic. To detect and prevent this, a robust
algorithm is needed. Currently, models manually label
tweets as real or fake, but this method takes time. To
address this, four machine learning algorithms were
used, with logistic regression providing 95% accuracy
and SVM achieving 98%. Many social media
companies are working to combat fake news.

Aug 14, 2024 Department of Computer Engineering, International Institute of Information Technology, Hinjawadi, Pune-411057 5
 Summary/findings of Literature Survey

• These research has explored various techniques, including Passive Aggressive Classifier,
Random Forest, Logical Regression, XGBOOST, Naïve Bayes, SVM, Logistic Regression,
and KNN. The study narrowed down the selection to six classifiers: logistic regression,
gradient boosting, random forest, passive aggressive classifier decision tree, and XGBoost.
• The ISOT fake news dataset was chosen for its diverse topics and well-balanced
composition of articles from reputable and unreliable sources.
• Exploratory Data Analysis (EDA) was conducted to identify patterns, relationships, and
anomalies, and visualizations were used to examine subject distribution and identify key
topics.

Aug 14, 2024 Department of Computer Engineering, International Institute of Information Technology, Hinjawadi, Pune-411057 6
 Proposed Methodology

• Fake News Detection System is a machine-learning-based system that

relies solely on its content and enables simple, cost-effective, and
time-efficient news verification.

Aug 14, 2024 Department of Computer Engineering, International Institute of Information Technology, Hinjawadi, Pune-411057 7
 Queries ?

Aug 14, 2024 Department of Computer Engineering, International Institute of Information Technology, Hinjawadi, Pune-411057 8