Finance News Intelligence Generation

A comprehensive system for extracting financial intelligence from news sources, analyzing market impact, and generating actionable insights using NLP and knowledge graphs.

🎯 Overview

This project combines Natural Language Processing (NLP), Knowledge Graph technology, and financial data analysis to:

• Extract financial events and entities from news articles
• Perform sentiment analysis on financial news
• Build a knowledge graph of relationships between companies, events, and market impacts
• Generate market impact predictions
• Provide explainable trading signals based on news sentiment

📋 Project Structure

finance-news-intelligence/
├── backend/                    # Python backend (API & Pipeline)
│   ├── config.py              # Configuration management
│   ├── server.py              # FastAPI application
│   ├── pipeline.py            # ETL pipeline
│   ├── database.py            # Database operations
│   ├── datetime_utils.py       # Utility functions
│   │
│   ├── nlp/                    # NLP Module
│   │   ├── ner_model.py       # Named Entity Recognition
│   │   ├── event_extractor.py # Event extraction
│   │   └── aspect_sentiment.py # Aspect-based sentiment analysis
│   │
│   ├── intelligence/           # Intelligence Module
│   │   ├── knowledge_graph.py  # Knowledge graph construction
│   │   ├── query_engine.py     # Graph query engine
│   │   ├── signal_generator.py # Trading signal generation
│   │   └── summarizer.py       # Report summarization
│   │
│   ├── market/                 # Market Analysis Module
│   │   ├── market_data_loader.py # Market data fetching
│   │   ├── price_loader.py     # Price data handling
│   │   └── impact_analysis.py  # Market impact analysis
│   │
│   ├── evaluation/             # Evaluation Module
│   │   ├── accuracy_report.py  # Accuracy metrics
│   │   ├── logic_tester.py     # Logic validation
│   │   └── visual_report.py    # Report visualization
│   │
│   ├── data/                   # Data Directory
│   │   ├── finance.db          # SQLite database
│   │   ├── stocks_universe.csv # Supported stocks list
│   │   ├── raw/                # Raw data storage
│   │   └── yfinance_cache/     # Finance data cache
│   │
│   └── lib/                    # Frontend libraries
│       ├── vis-9.1.2/          # Vis.js network visualization
│       ├── tom-select/         # Select dropdown library
│       └── bindings/           # JavaScript bindings
│
├── frontend/                   # React + Vite Frontend
│   ├── src/
│   │   ├── components/
│   │   │   ├── ExplainableReasoning.jsx
│   │   │   ├── MarketNews.jsx
│   │   │   ├── PortfolioImpact.jsx
│   │   │   └── RippleEffect.jsx
│   │   ├── pages/
│   │   │   ├── Dashboard.jsx
│   │   │   └── Login.jsx
│   │   ├── App.jsx
│   │   ├── api.js
│   │   └── main.jsx
│   ├── public/
│   ├── package.json
│   ├── vite.config.js
│   └── eslint.config.js
│
├── requirements.txt            # Python dependencies
├── README.md                   # This file

🚀 Quick Start

Prerequisites

• Python 3.8+
• Node.js 16+
• SQLite3

Backend Setup

1. Create and activate virtual environment:

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

2. Install dependencies:

pip install -r requirements.txt

3. Initialize database:

# Database will be initialized automatically when the server starts
python backend/server.py

Frontend Setup

1. Install dependencies:

cd frontend
npm install

2. Start development server:

npm run dev

📚 Common Commands

Backend

• Run API Server:

  python backend/server.py

  API will be available at http://localhost:8000

• Run ETL Pipeline:

  python backend/pipeline.py

Frontend

• Development:

  npm run dev

🔧 Core Modules

NLP Module (backend/nlp/)

Handles natural language processing tasks:

• NER Model: Extracts named entities (companies, people, locations)
• Event Extractor: Identifies financial events from text
• Aspect Sentiment: Analyzes sentiment for specific aspects (price, risk, growth)

Intelligence Module (backend/intelligence/)

Generates actionable intelligence:

• Knowledge Graph: Builds relationships between entities and events
• Query Engine: Retrieves relevant information from the graph
• Signal Generator: Creates trading signals based on news impact
• Summarizer: Generates concise reports

Market Module (backend/market/)

Analyzes market data:

• Market Data Loader: Fetches real-time and historical data
• Price Loader: Manages stock price information
• Impact Analysis: Quantifies news impact on prices

Evaluation Module (backend/evaluation/)

Validates and reports results:

• Accuracy Report: Metrics and performance analysis
• Logic Tester: Validates business logic
• Visual Report: Generates dashboard visualizations

📊 API Endpoints

News Analysis

• GET /api/news - Get latest news
• POST /api/news/analyze - Analyze news article
• GET /api/sentiment/{ticker} - Get sentiment for a stock

Knowledge Graph

• GET /api/graph/entities - List all entities
• GET /api/graph/relations/{entity} - Get entity relations
• POST /api/graph/query - Query the knowledge graph

Signals

• GET /api/signals/{ticker} - Get trading signals
• GET /api/signals/portfolio - Portfolio impact analysis

Market Impact

• GET /api/impact/{ticker} - Historical impact analysis
• POST /api/impact/predict - Predict market impact

🔐 Authentication

The system uses JWT-based authentication:

• Login: POST /api/auth/login
• Logout: POST /api/auth/logout
• Verify: GET /api/auth/verify

Credentials are secured using bcrypt hashing.

📈 Features

• Real-time News Processing - Fetches and processes financial news
• Sentiment Analysis - Multi-aspect sentiment scoring
• Entity Recognition - Identifies companies, executives, financial terms
• Knowledge Graph - Visual representation of relationships
• Trading Signals - AI-generated buy/sell recommendations
• Portfolio Impact - Analyze impact on your portfolio
• Explainable AI - Understand reasoning behind signals
• Market Impact Analysis - Quantify news effect on prices

🧪 Testing

Run the logic tester to validate the system:

python backend/evaluation/logic_tester.py

Generate an accuracy report:

python backend/evaluation/accuracy_report.py

📦 Dependencies

Backend

• feedparser - RSS feed parsing
• yfinance - Stock data fetching
• pandas - Data manipulation
• networkx - Graph algorithms
• pyvis - Graph visualization
• fastapi - Web framework
• uvicorn - ASGI server
• passlib - Password hashing
• python-jose - JWT handling
• requests - HTTP client
• matplotlib - Plotting

Frontend

• react - UI framework
• vite - Build tool
• axios - HTTP client

See requirements.txt for full dependencies.

🔄 Data Flow

News Source
    ↓
Feed Parser (feedparser)
    ↓
NLP Pipeline (NER, Event Extraction, Sentiment)
    ↓
Knowledge Graph Construction
    ↓
Market Impact Analysis
    ↓
Signal Generation
    ↓
Dashboard Visualization
    ↓
User

📝 Configuration

Edit backend/config.py to customize:

• Database path
• API port and host
• News sources
• Model parameters
• Update frequency

📄 License

This project is licensed under the MIT License.