File size: 1,730 Bytes
e077409
 
 
 
 
 
2be108a
e077409
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2be108a
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
---
title: News Veracity Checker
emoji: 📰
colorFrom: indigo
colorTo: green
sdk: streamlit
sdk_version: 1.45.1
app_file: app.py
pinned: false
---

# 📰 News Veracity Checker

With the explosive growth of social media, the spread of **fake news** has become a global crisis. According to the MIT Media Lab, **false information spreads six times faster** than true news on platforms like Twitter. Studies also show that nearly **64% of adults have been exposed to fake news online**, often without even realizing it.

That’s why I built this project: an **News Veracity Checker Using AI**. This tool classifies whether a news article or headline is **true or false** and provides a **fact-based explanation using an open-source language model**.

---

## 💡 Features

-**Veracity Score** – Classifies news as true or false using a BERT-based fake news detector
- 📃 **Article Summarizer** – Provides a concise summary of the content using Facebook’s BART model
- 🤖 **AI Explanation** – Uses Google’s Gemma-2B-Instruct LLM to generate fact-based rationales and corrections
- ☁️ **Word Cloud Visualization** – Visual insight into the most frequent keywords
- 🧠 **All models are run locally using Hugging Face Transformers**

---

## 🛠️ Tech Stack

- **Streamlit** – For interactive front-end interface
- **Hugging Face Transformers** – For veracity detection, summarization, and explanation generation
- **PyTorch**
- **Matplotlib & WordCloud** – For visualization
- **Models Used:**
  - `jy46604790/Fake-News-Bert-Detect` – for veracity classification
  - `facebook/bart-large-cnn` – for summarizing long news texts
  - `google/gemma-2b-it` – for AI-generated factual explanation