Fake News & Misinformation Classifier Tool

A modular Python-based application that analyzes news headlines and articles to determine their credibility using source reputation, sentiment analysis, and clickbait detection.

---

Project Overview

This system is designed to simulate a real-world fake news detection workflow. It demonstrates NLP integration, weighted scoring algorithms, and modular OOP application design.

The project focuses on:

• Structured GUI development using Streamlit
• SQLite database management for source reputation and history
• Natural Language Processing using TextBlob
• Object-Oriented Programming (Abstraction, Inheritance, Polymorphism, Encapsulation)
• Modular file organization across separate packages

---

Key Features

• Credibility scoring using a weighted formula: 0.40 × source + 0.35 × sentiment + 0.25 × keywords
• Source reputation database with 18+ pre-seeded news outlets (Dawn, BBC, Reuters, etc.)
• Sentiment analysis using TextBlob — detects emotional/biased tone
• Clickbait detection via regex (ALL CAPS, !!!, sensational phrases)
• Persistent history log of all past checks stored in SQLite
• Custom exception handling — no raw tracebacks ever shown in GUI
• Clean color-coded verdict: 🟢 Likely Real · 🟡 Suspicious · 🔴 Likely Fake

---

Tech Stack

• Python
• Streamlit (GUI)
• SQLite (Database)
• TextBlob (Sentiment Analysis)
• Pandas (Data Display)
• Regular Expressions (Clickbait Detection)

---

Project Structure

fake_news_classifier/
│
├── main.py
├── requirements.txt
│
├── models/
│   ├── article.py
│   ├── source.py
│   ├── content_checker.py
│   └── credibility_score.py
│
├── analysis/
│   ├── sentiment.py
│   └── scoring.py
│
├── database/
│   ├── db_manager.py
│   └── schema.sql
│
├── data/
│   └── keywords.json
│
└── utils/
    └── exceptions.py

• main.py → Streamlit entry point and all 3 GUI screens
• models/content_checker.py → Abstract class + HeadlineChecker + FullTextChecker
• models/credibility_score.py → CredibilityScore base + WeightedScore (formula)
• models/article.py → NewsArticle class with input validation
• models/source.py → Source class with encapsulated reputation score
• analysis/sentiment.py → TextBlob wrapper converting tone to 0–100 score
• analysis/scoring.py → Keyword/clickbait scorer using regex + JSON lists
• database/db_manager.py → Centralized SQLite CRUD operations
• database/schema.sql → Table definitions + 18 seeded news sources
• data/keywords.json → Clickbait phrase and sensational word lists
• utils/exceptions.py → 4 custom exception classes

---

Installation & Usage

1. Clone the repository:

git clone https://github.com/M3M397/fake-news-classifier.git

2. Install dependencies:

pip install -r requirements.txt

3. Download TextBlob language data (run once):

python -m textblob.download_corpora

4. Run the application:

streamlit run main.py

Open your browser at http://localhost:8501

---

Purpose of This Project

This project demonstrates my ability to:

• Design modular Python applications using OOP principles
• Integrate relational databases with application logic
• Apply NLP techniques for real-world text classification
• Build functional, modern GUI applications
• Handle errors gracefully using custom exception hierarchies

---