Project Report

on
Fake News Detection

Submitted in partial fulfillment of the requirement for the award of degree of

Master of Computer Application (MCA)

at

University of Mumbai
Submitted by

Sanved S. Vyavahare
Under the guidance of
Prof.Sudeshna Roy

Bharati Vidyapeeth’s
Institute of Management and Information Technology
Sector 8, CBD Belapur
Navi Mumbai
2022 - 2023
 Bharati Vidyapeeth’s
Institute of Management and Information Technology
Navi Mumbai

Certificate of Approval

This is to certify that the Project titled ’ Fake News Detection ’ is suc-
cessfully done by Sanved S. Vyavahare during internship of his course in par-
tial fulfillment of Master’s of Computer Application under the University
of Mumbai, Mumbai, through the Bharati Vidypeeth’s Institute of Management
and Information Technology, Navi Mumbai carried out by him under our guidance
and supervision.

Sign & Date External Examiner

Guide Signature and Date

Principal
Dr.Suhasini Vijaykumar

College Seal
 Declaration

I hereby declare that the project report presented is the result of my own work
completed during my MCA course, with the guidance of Prof.Sudeshna Roy . The
report has not been previously submitted, in part or in whole, to any other insti-
tution for any other degree or diploma from this university or any other university.
The guidelines provided by the university were followed in the report’s prepara-
tion. Any materials, including data, theoretical analysis, text, figures, and images,
from external sources were appropriately credited in the report’s text and referenced
accordingly.

(Signature )

(Sanved S. Vyavahare )

Date
 Acknowledgement

I would like to take this opportunity to express my sincere gratitude to the many
individuals who have contributed to my knowledge and experience throughout the
project and course. While presenting this project, I am thrilled to acknowledge
that its completion was not solely due to my own efforts but also the guidance
provided by our professors. I am especially grateful for the support of our project
guide, Prof.Sudeshna Roy, and the kind assistance of Dr. Suhaisni Vijaykumar,
In charge Principal BVIMIT, and other esteemed faculty members throughout the
entire course. Finally, I extend my deep appreciation to all those who provided their
valuable assistance during the execution of my project.

Date:

Sanved S. Vyavahare
Contents

1 Introduction 1
1.1 Introduction to Project . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Problem Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Objective of the project . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Scope of the project . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 System Study 5
2.1 Existing System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Proposed System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Use Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

3 Analysis & Design 8

3.1 Software Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.2 Hardware Requirements . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.3 Gantt Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.4 Activity Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.5 Sequence Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

4 Testing & Validation 12

4.1 Test case and Report . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

ii
5 User Manual 14
5.1 Explanation of Key functions . . . . . . . . . . . . . . . . . . . . . . 14
5.1.1 Passive Aggressive Classifier . . . . . . . . . . . . . . . . . . . 14
5.1.2 TF-IDF Vectorizer . . . . . . . . . . . . . . . . . . . . . . . . 14
5.2 Method of Implementation . . . . . . . . . . . . . . . . . . . . . . . . 15

6 System Implementation 16
6.0.1 Project Screen Shots . . . . . . . . . . . . . . . . . . . . . . . 16

7 Conclusion 19
7.1 Project Conclusion & Future Enhancement . . . . . . . . . . . . . . . 19

References 21

iii
Chapter 1

Introduction

Fake news, misinformation, and disinformation have become a major issue in today’s
world. With the rise of social media and the ease of sharing information, it has
become increasingly difficult to differentiate between real and fake news. Fake news
can cause harm by misleading people, spreading propaganda, and inciting violence.
It is essential to develop systems that can detect fake news and prevent it from
spreading.
Fake news detection systems use various techniques such as machine learning
algorithms, natural language processing, and data analysis to determine the veracity
of a news article. The system evaluates the credibility of the article by analyzing its
content, source, and other factors. If the article is deemed to be fake or misleading,
the system will flag it for further review by human moderators.
The purpose of this software requirements specification (SRS) is to outline the
functional and non-functional requirements of a fake news detection system. This
SRS will help guide the development of a system that can effectively detect and
prevent the spread of fake news. The system will be designed to be scalable, efficient,
and easy to use for users of all levels of technical expertise.

1.1 Introduction to Project

These days’ fake news is creating different issues to fabricated news and plan propa-
ganda in some outlets. Fake news and lack of trust in the media are growing prob-
lems with huge complications in our society. Fake News Detection is a model which

1
 Fake News Detection

prevents the spreading of misleading news stories that come from non-reputable
sources. In this project, it is seeking to create a model that can predict that a given
article is fake or real.

1.2 Problem Definition

The problem of fake news is widespread and complex. The proliferation of social
media platforms and the ease of sharing information has made it increasingly dif-
ficult to distinguish between real and fake news. Fake news can be created and
spread intentionally or unintentionally, and it can cause harm by misleading people,
spreading propaganda, and inciting violence.
The problem of fake news detection involves developing a system that can ac-
curately identify fake news and prevent it from spreading. This requires the use of
advanced technologies such as machine learning algorithms, natural language pro-
cessing, and data analysis. The system must be able to analyze the content, source,
and other factors of a news article to determine its credibility.
The problem of fake news detection is further complicated by the fact that fake
news can take many forms, from misleading headlines to fabricated stories. The
system must be able to detect all forms of fake news and differentiate them from
real news.
The consequences of failing to detect fake news can be severe, including social
unrest, political polarization, and the erosion of trust in institutions. Therefore,
developing an effective fake news detection system is critical for maintaining the
integrity of news and information in the digital age.

1.3 Objective of the project

The primary objective of developing a fake news detection system is to accurately
identify fake news and prevent it from spreading. To achieve this objective, the
following specific objectives can be outlined:
Develop an efficient and scalable system that can process a large volume of news
articles in real-time.

BVIMIT Page 2
 Fake News Detection

Use advanced technologies such as machine learning algorithms, natural language

processing, and data analysis to accurately identify fake news.
Develop a reliable and robust system that can differentiate between real and fake
news across various languages, cultures, and sources.
Design a user-friendly system that can be used by individuals with different levels
of technical expertise.
Develop a system that can identify the source of the fake news and track its
spread across various social media platforms.
Create a system that can flag potentially fake news articles for review by human
moderators and improve the accuracy of the system’s detection algorithms through
continuous feedback and updates.
Implement a system that complies with ethical and legal standards, including
data privacy, security, and transparency.
By achieving these objectives, a fake news detection system can effectively iden-
tify and prevent the spread of fake news, promoting the integrity of news and infor-
mation in the digital age.

1.4 Scope of the project

The scope of a fake news detection system includes the following:
The system will be designed to analyze news articles from various sources, in-
cluding online news outlets, social media platforms, and other sources.
The system will use advanced technologies such as machine learning algorithms,
natural language processing, and data analysis to analyze the content, source, and
other factors of a news article to determine its credibility.
The system will be able to differentiate between various types of fake news,
including fabricated stories, misleading headlines, and propaganda.
The system will be able to analyze news articles in various languages and across
different cultures and sources.
The system will be able to track the spread of fake news across various social
media platforms and identify the source of the fake news.
The system will flag potentially fake news articles for review by human moder-
ators to improve the accuracy of the system’s detection algorithms.

BVIMIT Page 3
 Fake News Detection

The system will comply with ethical and legal standards, including data privacy,
security, and transparency.
The scope of the fake news detection system will be limited to identifying fake
news and preventing its spread. The system will not be responsible for determining
the truth or accuracy of news articles, and it will not make judgments about the
quality or bias of news sources. The system will only provide an indication of the
credibility of a news article based on its analysis of the content, source, and other
factors.

BVIMIT Page 4
Chapter 2

System Study

2.1 Existing System

Limited accuracy: Existing systems for fake news detection often have limited accu-
racy due to the complexity of the problem. Fake news can take many forms, and it
is challenging to develop a system that can detect all types of fake news accurately.
Limited scalability: Some existing systems for fake news detection are not scal-
able and may not be able to handle a large volume of news articles in real-time.
This can limit their effectiveness in detecting fake news as it spreads rapidly across
social media platforms.
Limited language support: Many existing systems for fake news detection are
designed to analyze news articles in a specific language, limiting their ability to
detect fake news in other languages.
Limited cultural context: Some existing systems for fake news detection may not
be able to analyze news articles within the cultural context of a particular region
or community. This can limit their ability to detect fake news that is tailored to
specific cultural norms or beliefs.
Bias: Existing systems for fake news detection may contain bias due to the
training data used to develop the algorithms. If the training data is biased, the
system may also be biased in detecting fake news.
Cost: Developing and implementing an effective fake news detection system can
be costly, making it difficult for smaller organizations to invest in such systems.
Limited human oversight: Existing systems for fake news detection may rely too

5
 Fake News Detection

heavily on algorithms and automation and may not provide enough human oversight
to ensure accuracy and fairness.
Overall, existing systems for fake news detection have several limitations that
must be addressed to improve their accuracy and effectiveness in detecting fake
news.

2.2 Proposed System

There are several proposed systems for fake news detection, which use various tech-
nologies and techniques to analyze the content, source, and other factors of a news
article to determine its credibility. Some of the proposed systems for fake news
detection are:
Machine Learning Based Systems: These systems use machine learning algo-
rithms to analyze news articles and learn from historical data to identify patterns
and characteristics of fake news. These systems are trained on a large dataset of
labeled news articles, and the algorithms use this data to develop a model that can
accurately identify fake news.
Natural Language Processing Based Systems: These systems use natural lan-
guage processing techniques to analyze the language and tone of a news article to
identify patterns that are indicative of fake news. These systems can analyze the
semantics, syntax, and sentiment of a news article to determine its credibility.
Social Network Analysis Based Systems: These systems use social network anal-
ysis techniques to analyze the spread of news articles across various social media
platforms to identify the source of fake news and track its spread. These systems
can analyze the network structure and activity of social media users to detect pat-
terns of fake news dissemination.
Hybrid Systems: These systems combine multiple techniques, such as machine
learning, natural language processing, and social network analysis, to develop a
comprehensive approach to fake news detection. By combining these techniques,
these systems can provide more accurate and reliable detection of fake news.
The proposed systems for fake news detection are continually evolving, and new
techniques and technologies are being developed to improve their accuracy and reli-
ability. The selection of the system will depend on the specific requirements of the

BVIMIT Page 6
 Fake News Detection

application and the available resources.

2.3 Use Cases

Fake news detection has become an essential tool for media organizations, social
media platforms, and other entities that rely on accurate and reliable news reporting.
Some of the use cases for fake news detection are:
Media Organizations: Media organizations can use fake news detection to verify
the authenticity of news articles before publishing them. This can help prevent the
spread of fake news and improve the credibility of the organization.
Social Media Platforms: Social media platforms can use fake news detection to
flag potentially fake news articles for review by human moderators. This can help
prevent the spread of fake news and limit its impact on society.
Fact-checking Organizations: Fact-checking organizations can use fake news de-
tection to verify the accuracy of news articles and provide fact-checking services to
media organizations and social media platforms.
Government Agencies: Government agencies can use fake news detection to iden-
tify and track the spread of fake news related to important issues, such as public
health or national security.
Educational Institutions: Educational institutions can use fake news detection
as a teaching tool to educate students about the importance of critical thinking and
media literacy.
Corporate Entities: Corporate entities can use fake news detection to monitor
the spread of fake news related to their brand or industry and take appropriate
measures to protect their reputation.
Individuals: Individuals can use fake news detection to verify the authenticity
of news articles they come across on social media or other sources and prevent the
spread of fake news.
Overall, fake news detection has a broad range of use cases across various indus-
tries and sectors, highlighting its importance in today’s media landscape.

BVIMIT Page 7
Chapter 3

Analysis & Design

3.1 Software Requirements

ˆ Windows 8 and above
ˆ Matplotlib library for data visualization.
ˆ Pandas library for data manipulation and analysis.
ˆ Scikit-learn (sklearn) library for machine learning algorithms and tools.
ˆ NumPy library for numerical computing and array operations.

3.2 Hardware Requirements

ˆ RAM - 8 GB
ˆ CPU: Intel Core i5 or higher.
ˆ Storage: At least 1GB of free storage for the project files and datasets.

8
 Fake News Detection

3.3 Gantt Chart

A Gantt chart was created to plan and schedule the development of the Fake News
Detection Model. The chart outlines the various tasks and milestones involved in
the development process, including data collection, preprocessing, model training,
and tool development. The Gantt chart was used to ensure that the development
process is on track and to identify potential delays and bottlenecks.

Figure 3.1: Gantt Chart

BVIMIT Page 9
 Fake News Detection

3.4 Activity Diagram

Figure 3.2: Activity Diagram

BVIMIT Page 10
 Fake News Detection

3.5 Sequence Diagram

Figure 3.3: Sequence Diagram

BVIMIT Page 11
Chapter 4

Testing & Validation

4.1 Test case and Report

Test Case 1: Valid News Article - Input a news article that is confirmed to be real,
and check if the system correctly classifies it as real.

Test Case 2: Fake News Article - Input a news article that is confirmed to be
fake, and check if the system correctly classifies it as fake.

Test Case 3: Ambiguous News Article - Input a news article that is ambiguous
or difficult to classify, and check if the system provides an appropriate output.

Test Case 4: Cross-validation - Use cross-validation to evaluate the performance

of the system on a large dataset with known labels. Compare the accuracy, preci-
sion, recall, and F1-score of the system with other state-of-the-art methods.

Test Case 5: Adversarial Examples - Generate adversarial examples to test the

robustness of the system against attacks. Adversarial examples are inputs that are
specifically crafted to fool the system into making incorrect predictions.

Test Case 6: Edge Cases - Test the system’s performance on edge cases such
as short news articles, long news articles, news articles with misspelled words or
grammatical errors, etc.

12
 Fake News Detection

Test Reports:
After conducting each test case, a report should be generated to document the
results of the test. The report should include the following: Test Case Description:
A brief description of the test case. Input Data: The input news article used for the
test.
Expected Output: The expected output of the system for the given input.
Actual Output: The actual output of the system for the given input.
Evaluation Metrics: The evaluation metrics used to evaluate the performance of
the system such as accuracy, precision, recall, and F1-score.
Analysis: An analysis of the test results and any observations or insights gained
from the test.
Conclusion: A conclusion summarizing the test results and the overall perfor-
mance of the system.
By conducting a variety of test cases and generating detailed test reports, the
performance and reliability of the fake news detection system can be assessed and
improved over time.

BVIMIT Page 13
Chapter 5

User Manual

5.1 Explanation of Key functions

5.1.1 Passive Aggressive Classifier
Passive-Aggressive algorithms are generally used for large-scale learning. It is one
of the online-learning algorithms. In online machine learning algorithms, the input
data comes in sequential order and the machine learning model is updated sequen-
tially, as opposed to conventional batch learning, where the entire training dataset
is used at once.
Passive: If the prediction is correct, keep the model and do not make any changes.
i.e., the data in the example is not enough to cause any changes in the model.
Aggressive: If the prediction is incorrect, make changes to the model. i.e., some
change to the model may correct it.

5.1.2 TF-IDF Vectorizer

Tf-IDF stands for term frequency-inverse document frequency. Transforms text to
feature vectors that can be used as input to estimator vocabulary Is a dictionary
that converts each token (word) to feature index in the matrix, each unique token
gets a feature index.

14
 Fake News Detection

5.2 Method of Implementation

There are several methods of implementing a fake news detection system, each with
its advantages and disadvantages. The following are some of the commonly used
methods:
Rule-Based Approach: A rule-based approach involves defining a set of rules
based on which news articles are classified as fake or real. These rules are often
based on common characteristics of fake news such as sensational headlines, emo-
tional language, and lack of sources. While this approach is relatively simple and
transparent, it may not be effective in detecting sophisticated fake news that can
mimic real news.
Supervised Learning Approach: A supervised learning approach involves train-
ing a machine learning model using a labeled dataset of news articles. The model
learns to classify news articles based on their features such as word frequency, sen-
timent analysis, and more. While this approach can be highly effective in detecting
fake news, it requires a large and diverse labeled dataset for training and may be
susceptible to overfitting.
Unsupervised Learning Approach: An unsupervised learning approach involves
clustering news articles based on their features without any prior labels. This ap-
proach is useful in identifying patterns and anomalies in the data but may not be
effective in distinguishing between fake and real news.
Hybrid Approach: A hybrid approach combines the rule-based and machine
learning-based methods to achieve greater accuracy and robustness. The rule-based
approach can be used to filter out obvious cases of fake news, and the machine
learning-based approach can be used to classify more subtle cases.
The method of implementation for a fake news detection system depends on
several factors, such as the available data, the resources and time constraints, and
the required accuracy and reliability. In general, a hybrid approach that combines
both rule-based and machine learning-based methods can provide the best balance
of accuracy, transparency, and robustness.

BVIMIT Page 15
Chapter 6

System Implementation

6.0.1 Project Screen Shots

16
 Fake News Detection

Figure 6.1: Before Predicting

Figure 6.2: Predicting Fake News

BVIMIT Page 17
 Fake News Detection

Figure 6.3: Predicting Real News

BVIMIT Page 18
Chapter 7

Conclusion

7.1 Project Conclusion & Future Enhancement

In conclusion, the development of a fake news detection system is an important and
challenging task that requires a combination of machine learning, natural language
processing, and data analytics techniques. The system should be able to accurately
identify fake news articles while minimizing false positives and false negatives. The
system can be deployed on various platforms, including social media platforms,
news websites, and search engines, to help combat the spread of fake news and
misinformation.
In terms of future enhancements, there are several areas of improvement that
can be explored, such as:
Incorporating more data sources and types of data, including images, videos,
and social media posts, to improve the accuracy of the system.
Developing more advanced natural language processing techniques, such as sen-
timent analysis and emotion detection, to better understand the context and tone
of the news articles.
Using explainable artificial intelligence (AI) techniques to increase transparency
and interpretability of the system’s decision-making process.
Incorporating user feedback and engagement to continually improve the system’s
performance and to adapt to new types of fake news and misinformation.
Integrating the fake news detection system with other tools and platforms, such
as fact-checking services, to provide a more comprehensive solution for combating

19
 Fake News Detection

fake news and misinformation.

Overall, the development and improvement of fake news detection systems will
continue to be an important area of research and development in the field of artificial
intelligence and data analytics, as it has the potential to have a significant impact
on society’s ability to distinguish truth from fiction in the news media.

BVIMIT Page 20
References

[1] J. Mantas. Methodologies in pattern recognition and image analysis-a brief

survey. Pattern Recognition, 1987.

[2] Samuel Foucher. An evaluation of medical imaging. In IGARSS, 2009.

[3] Sen Lee and Eric Foucher. Radar Imaging and Applications. CRC Press, 2009.

[4] A. Zurada, S. Zage, and E. Nezry. Frequency division FDM. In Remote Sensing
Symposium, 1992. IAPRSS ’98., pages 80 –82, 1992.

[5] Shitole Sanjay. Digital image processing. Recognition, 1988.

[6] Y S Rao. Microwave imaging. Imaging, 1992.

21