Skip to content

news_depoliticizer

A tool that aggregates news from multiple sources, ranks stories by importance, and uses a local LLM to rewrite them without political bias or inflammatory language.

View on GitHub →

Overview

Languages: Python

Documentation

News Depoliticizer

A tool that aggregates news from multiple sources, ranks stories by importance, and uses a local LLM to rewrite them without political bias or inflammatory language.

Overview

News Depoliticizer helps readers consume news without the political spin and inflammatory language that often accompanies today's reporting. The project:

  1. Legally collects news from multiple APIs (no web scraping)
  2. Identifies and groups similar stories across sources
  3. Ranks stories by prominence/importance
  4. Uses a local LLM (via Ollama) to rewrite each story in a neutral, balanced tone
  5. Presents the depoliticized stories through a clean web interface

Features

  • Multi-source aggregation: Collects news from NewsAPI, NY Times API, and The Guardian API
  • Story grouping: Identifies the same stories across different sources
  • Importance ranking: Orders stories based on their prominence across news outlets
  • Bias removal: Rewrites stories to remove political bias and inflammatory language
  • Local processing: All LLM operations run on your local machine, ensuring privacy
  • Clean web interface: Mobile-friendly display of depoliticized news stories

Requirements

  • Python 3.8+
  • Ollama (with llama3 model installed)
  • API keys for news sources (optional, sample data provided)

Installation

1. Clone the repository

git clone https://github.com/yourusername/news-depoliticizer.git
cd news-depoliticizer

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install required packages

pip install requests pandas numpy scikit-learn python-dotenv flask

4. Install Ollama (if not already installed)

Follow the Ollama installation instructions for your platform.

Make sure the llama3 model is installed:

ollama pull llama3

5. Set up API keys (optional)

Create a `.

[View full README on GitHub]