An AI-powered RAG system designed to combat media bias by automatically generating a neutral, fact-based summary of any news article.
The system takes a URL to a news article and orchestrates a multi-step retrieval and generation process:
-
Neutral Query Generation
An LLM analyzes the article's title and description to extract core keywords and form a bias-free search query. -
Multi-Source Retrieval
The agent uses this query to retrieve comprehensive context from three distinct sources:- AskNews API: Gathers different perspectives from a variety of mainstream news outlets.
- Reddit: Collects public opinion, discussions, and alternative viewpoints on the topic.
- Original Article RAG: Retrieves specific facts and details from the source text itself.
-
Synthesis and Neutralization
All gathered information is fed into a final LLM, which synthesizes the diverse perspectives and facts into a single, cohesive, and unbiased news article.
It retains key details while stripping away any biased language or framing.
-
Agentic Framework
Built on LangChain, the agent intelligently plans and executes a multi-step workflow. -
Comprehensive Context
Gathers information from mainstream media, public forums, and the original source for a holistic view. -
Bias Removal
Specifically engineered to identify and neutralize politically or emotionally charged language. -
Extensible Architecture
The modular design allows for easy integration of new data sources (e.g., academic journals, government reports).