Skip to content

In a digital age where truth is buried beneath waves of misinformation, VERIFIKA stands as the last sentinel, cutting through deception with relentless precision.

License

Notifications You must be signed in to change notification settings

AlwaysRead/Verifika

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

25 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

VERIFIKA – AI-Powered Fake News Detection WebApp


Vision

In an era where misinformation spreads faster than truth, VERIFIKA aims to be the digital guardian of truth. By leveraging AI-powered fact-checking, we strive to create a world where individuals can access unbiased, credible, and verified news in real-time.


System Flowchart

User Input (News Article Text)  
        ↓  
Preprocessing (Text Cleaning, Tokenization, Stopword Removal)  
        ↓  
Feature Extraction (Sentiment, Political Bias, Clickbait Probability)  
        ↓  
AI Model Analysis (Logistic Regression for Fake News Classification)  
        ↓  
Credibility Score Generation
        ↓  
Result Output (Fake/Real News Classification + Additional Analysis)  

Dataset

  • Source: Kaggle Fake and Real News Dataset
  • Size:
    • Fake.csv (23,502 fake news articles)
    • True.csv (21,417 true news articles)
  • Features:
    • Headline & Body Text
    • Source Reliability
    • Sentiment & Clickbait Score
    • Political Bias Analysis

Features

Real-Time Fact-Checking – Instantly verify online news and social media posts.
Fake News Detection – AI-powered model to classify news as Real or Fake.
Sentiment & Bias Analysis – Understand the emotional and political tilt of the article.
Clickbait Detector – Identify misleading and sensationalist headlines.
Credibility Score – Evaluate the trustworthiness of a source before sharing.
WebApp Interface – Accessible via a user-friendly web application.


Technologies Used

  • Frontend: React.js & CSS [Github Repo], Vercel (Frontend Hosting)
  • Backend: Python, Render (Backend Hosting)
  • Model Training: Scikit-learn, NLTK, Logistic Regression
  • Data Storage & Management: GitHub

Model Training & Selection

  • Preprocessing: Tokenization, Stopword Removal, TF-IDF
  • Models Used:
    • Logistic Regression (Best-performing model, achieving 98% accuracy)
    • Decision Tree Classifier
    • Gradient Boosting Classifier
    • Random Forest Classifier
  • Feature Engineering:
    • Sentiment Analysis using NLTK
    • Political Bias & Clickbait Score Calculation
  • Training Process:
    • Data Cleaning & Splitting (Train-Test: 80-20)
    • Evaluation using Precision, Recall, and F1 Score

Real-Time Prediction

  1. User Inputs an Article or News Headlines
  2. Preprocessed & Extracted Features Sent to ML Model
  3. AI Model Analyzes & Returns:
    • Fake/Real Classification
    • Sentiment & Bias Insights
    • Credibility Score
  4. Results Displayed on WebApp

Future Scope

Improved Accuracy – Upgrade to Deep Learning (Transformers, BERT, LSTMs)
Multi-Language Support – Expand to detect misinformation across languages
Blockchain Integration – Immutable records for verified news sources
Mobile App & Browser Extension – For seamless user experience
Social Media Integration – Fact-check trending topics in real-time


Sample

image

image

image


Contact

Chinmoy Saikia
Nichol Das


🔗 VERIFIKA – Because the Truth Deserves to Be Heard.

About

In a digital age where truth is buried beneath waves of misinformation, VERIFIKA stands as the last sentinel, cutting through deception with relentless precision.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%