PROJECT PROPOSAL

Submitted in the Prescribed Proforma

Fake News Detection Using Machine

Learning Algorithms
Used for Detecting MisInformation
Submitted to
Coign Consultants Pvt.Ltd
by

Dr. V.Chandrashekar

Host Institution

MALLA REDDY COLLEGE OF ENGINEERING & TECHNOLOGY

(Autonomous Institution – UGC, Govt. of India)
Sponsored by CMR Educational Society
(Affiliated to JNTU, Hyderabad, Approved by AICTE – Accredited by NBA & naac – ‘A’ Grade - I ISO 9001:2008 Certified)
Maisammaguda, Dhulapally, (Post Via Hakimpet),Secunderabad - 500 100.Telangana State, India.
Contact Number:040-23792146/64634237, E-Mail ID: [email protected], website: www.mrcet.ac.in
 “Used for Detecting Misinformation”
PROJECT SUMMARY AND BUSINESS PLAN

Proposed Work

(A1) Fake News Detection Using Machine Learning Algorithms

“Used for Detecting Misinformation”

(A2) Introduction

In recent years, the rapid growth of the internet and social media platforms has drastically
transformed the way information is consumed and shared. While this has brought about numerous
benefits in terms of access to information, it has also given rise to a significant challenge: the
spread of fake news. Fake news refers to misinformation or deliberately fabricated content
designed to deceive or mislead people. This type of false information can be found across various
platforms, including social media, blogs, news websites, and even messaging apps.

The consequences of fake news can be far-reaching. It has the potential to influence public
opinion, sway elections, manipulate markets, and even incite social unrest. The sheer volume and
speed at which news spreads online make it increasingly difficult for individuals and even
traditional fact-checking organizations to keep up. As a result, detecting and combating fake news
has become a critical task for researchers, journalists, and policymakers.

Machine learning (ML), a subset of artificial intelligence, has emerged as a powerful tool for
addressing this challenge. Machine learning algorithms can be trained to automatically identify
patterns in news articles and classify them as either real or fake, often with high levels of
accuracy. These algorithms use advanced statistical techniques to analyze vast amounts of data
and learn from historical examples of fake and real news.

The process of fake news detection involves several key components, including data collection,
text Preprocessing, feature extraction, model selection, and evaluation. Through various
supervised and unsupervised learning methods, such as decision trees, support vector machines
(SVM), and deep learning models like convolutional neural networks (CNN) and transformers
(e.g., BERT), machine learning algorithms can detect subtle linguistic and contextual cues that
distinguish fake news from credible sources.

Given the growing importance of fake news detection, this field has garnered significant attention
in recent years. By utilizing machine learning techniques, researchers aim to create automated
systems that can accurately identify and mitigate the spread of false information in real-time,
improving the quality of information available to the public.

In this context, this study will explore the application of machine learning algorithms in detecting
fake news, analyzing the challenges, techniques, and future opportunities for advancing the field.

As fake news continues to evolve, research in this area is ongoing to develop more sophisticated
and explainable AI models. The integration of blockchain technology for verifying sources,
reinforcement learning for adaptive detection, and multimodal analysis incorporating images and
videos represents the future of fake news detection.
 (A3) If the idea involves use of existing intellectual property, give details of the owner
of the same and arrangements of sourcing the innovation and terms of its
commercialization
The system involves the use of existing intellectual property (IP), it is essential to identify
the owners of the technology, the arrangements for sourcing the innovation, and the
terms of commercialization. Below are the key aspects related to IP rights and
commercialization strategies.

1. Intellectual Property Ownership & Existing Technologies

Several components of fake news detection may involve pre-existing intellectual
property, including:
1. Machine Learning Algorithms & NLP Models
 Algorithms like Naïve Bayes, Support Vector Machines (SVM), Random
Forest, LSTMs, and Transformers are open-source and widely used in
research.
 Pre-trained NLP models (e.g., BERT, GPT, RoBERTa) are developed by
companies like Google (TensorFlow), OpenAI, Meta (PyTorch), and
Hugging Face. These models have open-source licenses, but commercial
use might require permissions or licensing.
2. Datasets for Model Training
 Datasets such as LIAR, FakeNewsNet, Kaggle’s Fake News Dataset are
commonly used for training machine learning models. These datasets
have varying licenses (e.g., CC-BY, GPL, or Apache License), and
commercial usage should comply with their terms.
3. Software Libraries & Cloud Services
 TensorFlow, PyTorch, Scikit-learn, NLTK, and Spacy are open-source
libraries commonly used for fake news detection.
 Cloud-based AI solutions from Google Cloud, AWS, or Microsoft Azure
may have proprietary rights and require subscription-based or usage-
based licensing.

2. Arrangements for Sourcing the Innovation

If the project integrates existing ML models, datasets, or software libraries, the sourcing
can be managed through:
1. Open-Source Licensing
 Utilize models and tools under permissive licenses such as Apache 2.0,
MIT, or BSD, ensuring compliance with usage terms.
 If modifying an open-source model (e.g., fine-tuning a BERT model), check
if attribution or redistribution conditions apply.
2. Paid API & Cloud-Based AI Services
 If using commercial AI services (e.g., Google NLP API, OpenAI API),
licensing fees or subscription costs apply.
 Data storage and processing may require agreements with cloud providers
like AWS, Google Cloud, or Microsoft Azure.
 3. Collaborations & Partnerships
 Partnering with universities, research institutions, or tech companies can
provide access to exclusive datasets or proprietary AI models.
 Licensing agreements may be needed if third-party models are
incorporated into the project.
3. Terms of Commercialization
To ensure smooth commercialization while respecting IP laws, the following strategies
can be applied:
1. Licensing Agreements
 If proprietary datasets, APIs, or AI models are used, negotiate commercial
licensing agreements with the owners.
 Open-source models with restrictions on commercialization (e.g., GNU
General Public License) must be checked to avoid legal issues.
2. Patent Filing for Unique Innovations
 If the project develops a novel fake news detection algorithm or an
improved AI framework, patenting it under IP laws can protect the
innovation.
 Filing a patent or copyright in relevant jurisdictions (e.g., India’s Intellectual
Property Office (IPO), US Patent Office (USPTO)) can provide market
exclusivity.
3. Revenue Model Considerations
 The system can be monetized via subscription-based services, API
access, enterprise licensing, or SaaS (Software-as-a-Service) models.
 If using external APIs or cloud-based solutions, factor in royalty payments,
revenue-sharing agreements, or platform fees.
4. Compliance with Data Privacy Laws
 If handling user data, ensure compliance with regulations like GDPR
(Europe), CCPA (California), or India’s Digital Personal Data Protection Act
(DPDPA).

A4) Specify the potential areas of application in industry/ market

 Social-Media platforms
 Political campaigns and Elections
 Financial markets
 Government and law Enforcement

A5) Specify newness/uniqueness of the innovation (better

performance/new features / improvements)
The proposed Fake News Detection System introduces several innovative features and
improvements over existing solutions, making it more accurate, efficient, and scalable.
Below are the key aspects that define the newness and uniqueness of the innovation :

1. Improved Performance with Hybrid Machine Learning Models

 Unlike traditional models that rely on single classifiers (e.g., Naïve Bayes, SVM,
or Random Forest), this system employs a hybrid model combining multiple
techniques.
 It integrates deep learning (LSTM, Transformers like BERT or RoBERTa) with
traditional NLP models for higher accuracy.
  The model uses ensemble learning, where multiple classifiers work together to
reduce false positives and improve reliability.

2. Advanced Natural Language Processing (NLP) Techniques

 Context-Aware Analysis: Instead of just keyword matching, the system
understands the context and intent of news articles.
 Sentiment & Emotion Detection: Identifies sensationalism or emotionally
manipulative language used in fake news.
 Stance Detection: Compares the stance of a given article with verified sources to
check for inconsistencies.
 Named Entity Recognition (NER): Identifies and verifies critical entities like
people, organizations, and locations to detect misinformation.

3. Deepfake & Multimedia Fake News Detection

 Unlike most fake news detectors that focus only on text, this system analyzes
images, videos, and audio content.
 Uses AI-based deepfake detection to identify manipulated videos or AI-generated
synthetic media.
 Integrates reverse image search & metadata analysis to verify the authenticity of
news-related visuals.

4. Real-Time Fact-Checking & Source Verification

 Automatically cross-references articles with verified fact-checking databases (e.g.,
PolitiFact, Snopes, Google Fact Check).
 Uses blockchain technology to ensure data integrity and verify sources.
 Integrates web scraping and real-time news aggregation to compare multiple
sources.

5. Scalable API & Cloud-Based Deployment

 Offers a lightweight and scalable API that can be integrated into social media
platforms, news websites, and government monitoring systems.
 Cloud-based architecture ensures real-time processing of large datasets, making
it highly scalable for enterprises.
 Multi-platform support: Works across web applications, mobile apps, and browser
extensions for easy accessibility.

6. Adaptive Learning with Reinforcement Learning & User Feedback

 The system continuously learns from new fake news patterns using reinforcement
learning.
 Users can report and flag content, which the model uses to improve detection
accuracy over time.
 Bias Detection Mechanism: Identifies political or ideological bias in news articles
and presents a balanced view.

7. Compliance with Global Regulations & Ethical AI Principles

 Ensures compliance with GDPR, India’s IT Rules 2021, EU Digital Services Act,
and other global regulations on misinformation.
 Uses explainable AI (XAI) to provide transparency on why an article was
classified as fake or real.
 Privacy-focused: Avoids unnecessary data collection and ensures ethical AI use.

 A6) what is already available in market? What is the market
potential? What value it would add in market
Several solutions are already available, developed by tech giants, fact-checking
organizations, and AI researchers. Some notable existing solutions include:
A. AI-Based Fake News Detection Tools
 Google Fact Check Tools – Aggregates fact-checks from credible sources.
 Facebook & Twitter AI-Based Content Moderation – Uses machine learning to
flag misinformation.
 Logically AI – AI-powered misinformation detection and analytics platform.
 Microsoft NewsGuard – Uses human and AI-based evaluations for website
credibility ratings.
B. Fact-Checking Organizations
 PolitiFact & Snopes – Human-driven fact-checking for political news.
 BBC Reality Check & Reuters Fact Check – Verify misleading news reports.
C. Open-Source Fake News Detection Models
 LIAR Dataset & FakeNewsNet – Used in research for training ML models.
 Fake News Detection APIs (e.g., AdVerif.ai, BERT-based models) – NLP-
based detection solutions.
2. Market Potential for Fake News Detection
A. Growing Demand & Industry Adoption
 The global fake news detection market is expected to grow significantly due to
rising misinformation threats.
 Governments worldwide are enforcing misinformation laws (e.g., EU’s Digital
Services Act, India’s IT Rules 2021).
 Media organizations, social media platforms, and governments are actively
investing in AI-powered misinformation detection.
B. Potential Market Size & Sectors
 Social Media & Tech Platforms: Facebook, Twitter, YouTube need scalable fake
news detection tools.
 News & Media: Press agencies and journalists require AI-based fact-checking
assistance.
 Government & National Security: Used to detect political propaganda and
misinformation campaigns.
 Corporate & Finance: Preventing stock market manipulation through fake news.
 Healthcare: Identifying false medical claims and vaccine misinformation.
3. Value Addition of the Proposed Innovation
A. Key Advantages Over Existing Systems
 Multimodal Detection: Unlike most solutions focused on text, this system detects
fake images, videos, and deepfakes.
 Hybrid AI Approach: Combines NLP, deep learning, and blockchain-based
verification.
 Real-Time Fact-Checking: Cross-checks with verified sources automatically.
 Bias & Sentiment Analysis: Identifies emotional manipulation and political
bias in news.
 Scalability: Cloud-based API integration allows for wider adoption across
industries.
B. Business & Social Impact
 Restores Trust in Digital Information: Helps fight misinformation, protecting
democracy and public discourse.
 Enhances Brand & Reputation Protection: Corporations can detect and
mitigate false narratives.
B.Current Development Status of Innovation
B1) what is the current development status of the innovation/product or
service offerings? (Whether still an idea or ready to launch)
The present development of the idea is in design stage.

B2) Idea readiness level

TRL 2 : Technology concept and/or application formulated

B3) Specify the time period in months required for innovation to be

completely developed for field testing/ ready for intended end user?
The total time period will take up to 12 to 15 months.

C.Financial Requirements

C1) Do you have a business plan for taking innovation from lab to
market? (Attach business plan in your own format)
We are interested in taking the prototype to market however currently
there is no business plan as it depends on the prototype.

C2) what level of funding is required for making innovation market ready?
Rs. 5. 04 Lakhs for the prototype; Further funds required to setup a manufacturing
plant and DFM services.

C3) Please give activity wise breakup (Activities include design / prototype
development / lab / bench scale production/ professional services/hiring staff / trials
/ test marketing / miscellaneous)

S. No. Particular Amount (INR)

1 work order & Analysis 75,000.00
2 Modelling & Design with Code 1,00,000.00
3 Testing & deployment process with final reports 75,000.00
Total Amount
(Rupees Two Lacks Fifty Thousand only) 2,50,000.00

A) Are there any team members/partners and mentors/ guides in your

innovative project. If so give name and complete contact address with
phone and email)

Dr. M. MOHAMMED MOHAIDEEN

Professor
Department of Aeronaut ical Engineering
Maisammaguda, Dulapally
Post Secunderabad 500100
Mobile: +91 7373369629
Email: [email protected]
B)Information on Patents filed/granted (if any)
NONE

C)Any awards or recognition related to the innovation

NONE

D) Please include any further information that you wish to communicate to us to

help us in judging your application

We have experience in research and developing evolutionary algorithms,

which are part of evolutionary learning’s in problem solving. The ultimate
success of the venture depends on management to develop innovative
product and to cost effectively deliver it to large and receptive market. We
have caliber professionals as mentors who create immense synergy in our
venture.

Profile of PI

Dr. M. Mohammed Mohaideen is working as a Professor in the

Department of Aeronautical Engineering. He obtained his B.E (Mech.), M.E
(Aero.) and Ph.D. (Aero). degrees from Anna University, Chennai and M.B.A from
Madurai Kamaraj University, Tamil Nadu, India. He is having 15 years of
teaching experience in Engineering subjects to U.G and P.G students. He has
been granted four international patents in Government of Australia, India, South
Africa and in Federal Republic of Germany, furthermore one Indian patent had
been published. He has published more than 25 research publications, three
book chapters to his credit and guided many PG and UG student’s projects. He is
the Recognized supervisor in Gujarat Technical University, Ahmedabad, Gujarat
He has also been delivered various expert lectures as the resource person in the
various domains of Aeronautical Engineering. He is the Reviewer of Elsevier and
Hindawi publication journals. He has organized various workshops, faculty
development programmes, seminars and conferences.
Dr. Mohammed is a Life member of Solar Energy society of India, a Life
member of Indian Society of Technical Education, a Member of American Society
of Mechanical Engineers and Innovation Ambassador of Institutions Innovation
Council, MRHD Innovation cell, India. He is also an Academic council member of
Noorul Islam Deemed University for Board of studies. He received best faculty
award and best paper award from IJCE and IOP journal respectively. His
research areas of expertise are Propulsion, Aircraft Performance, Nano
composites in Aerospace application and aircraft structures.