Apppublic

NirajGogoi/Multi-Class_News_Authenticity_Detector

sourceHugging Faceupdated 9mo agoView on Hugging Face

0likes

App README

Multi-Class Fake News + Scam + Not-News Detection System

📌 Project Overview

This project is an AI-based Natural Language Processing (NLP) system that classifies input text into four categories:

—REAL News
—FAKE News
—SCAM Messages
—NOT-NEWS (casual text / opinions)

The system uses TF-IDF vectorization and a Logistic Regression model to perform multi-class text classification. A Streamlit web interface is used for real-time prediction.

⚙️ Technologies Used

—Python
—Scikit-learn
—TF-IDF Vectorizer
—Logistic Regression
—Pandas & NumPy
—Streamlit (Web Interface)

📂 Project Files

—app.py → Streamlit application
—lrmodelmulti.jb → Trained multi-class model
—vectorizer_multi.jb → TF-IDF vectorizer
—requirements.txt → Required Python libraries
—README.md → Project documentation

▶️ How to Run Locally

1.Install Dependencies: py -m pip install -r requirements.txt

1.Run Streamlit App: py -m streamlit run app.py

1.Open Browser: http://localhost:8501

🧪 Sample Test Inputs

Government announces new education policy today → REAL Aliens landed in Delhi confirmed → FAKE Your bank account is locked verify immediately → SCAM I love cricket so much → NOT-NEWS

🌐 Deployment

The application can be deployed online using Hugging Face Spaces by uploading:

—app.py
—lrmodelmulti.jb
—vectorizer_multi.jb
—requirements.txt
—README.md

📈 Future Enhancements

—Use Deep Learning models (BERT, LSTM)
—Add URL credibility checking
—Multilingual news detection
—Browser extension integration

🎓 Project Outcome

This system helps users identify misinformation, scams, and non-news content, promoting safer information consumption.