PopNews Pulse Feed

A personalized AI-powered news aggregator that fetches, summarizes, and fact-checks news articles.

Setup Instructions

Prerequisites

• Python 3.8+
• Node.js 16+
• npm or yarn

Installation

1. Install Python dependencies:

pip install flask flask-cors requests

News Analysis System Workflow

Overview

This system uses LangGraph, Flask, and external AI services to fetch, analyze, and fact-check news articles.

API Endpoints

Endpoint	Method	Description
/api/news	GET	Search news articles by topic
/api/analyze	POST	Analyze a single article's credibility
/api/summary	POST	Generate bullet-point summaries
/api/fake-news-detection	POST	Detect if an article contains fake news

Core Agent Workflow

The main processing workflow consists of five sequential steps:

1. fetch_news():

   • Retrieves articles from GNews API based on search topic
   • Transforms articles into a standardized format
   • Handles errors from the API

2. detect_fake_news():

   • Uses Gemini API to analyze article authenticity
   • Generates a fake news score (0-1)
   • Provides explanations for the score

3. generate_summary():

   • Creates structured bullet-point summaries
   • Focuses on key insights, context, and significance
   • Uses Gemini API for natural language processing

4. fact_check():

   • Performs detailed credibility analysis
   • Evaluates source verification and content issues
   • Provides reliability points and misinformation warnings

5. finalize_results():

   • Compiles all analysis data into a complete response
   • Marks the process as completed

External Dependencies

• GNews API: Source of news articles using API key GNEWS_API_KEY
• Gemini AI API: Google's generative AI model providing analysis capabilities
  • Uses model: gemini-1.5-flash
  • Accessed via API key GEMINI_API_KEY

Data Flow

1. User submits request via one of the API endpoints
2. For full topic searches:
   • process_topic() orchestrates the complete workflow
   • All five agent functions run sequentially
3. For single article analysis:
   • The appropriate agent function is called directly
   • Results are returned to the user
4. External API calls are made as needed
5. Results are formatted as JSON and returned to the user

Error Handling

• Each agent function contains error handling to ensure graceful failure
• Default values are provided when external APIs fail
• Errors are logged and returned to the client