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