Financial Insight - AI-Powered Document Analysis Platform

Financial Insight is a comprehensive platform designed to extract actionable intelligence from financial and legal documents. It leverages advanced AI models for Named Entity Recognition (NER), Sentiment Analysis, Clause Extraction, and Contextual Question Answering (RAG).

🚀 Key Features

1. Named Entity Recognition (NER)

• Automatic Extraction: Identifies key entities such as Organizations, Dates, Money, Locations, and more.
• Visualization: Interactive highlighting of entities within the document text.
• Analytics: Entity distribution charts and frequency analysis.
• Powered By: spaCy (en_core_web_sm) and custom-trained models.

2. Sentiment Analysis

• Financial Context: specifically tuned for financial texts using FinBERT.
• Granular Analysis: Sentence-level sentiment scoring (Positive, Negative, Neutral).
• Visual Heatmap: Color-coded document view to instantly spot sentiment trends.
• Powered By: ProsusAI/finbert via Hugging Face Transformers.

3. Clause Extraction

• Legal Intelligence: Automatically extracts and categorizes critical legal clauses (e.g., Liability, Termination, Payment).
• Risk Profiling: Categorizes clauses by risk level (High, Medium, Low).
• Powered By: Groq (llama-3.1-8b-instant) for high-speed, accurate extraction.

4. RAG Chatbot (Contextual AI Guide)

• Interactive Q&A: Chat with your documents to get precise answers.
• Context Aware: Uses Retrieval-Augmented Generation to ground answers in document content.
• Powered By:
  • LLM: Groq (llama-3.1-8b-instant)
  • Embeddings: Google Gemini (text-embedding-004)
  • Vector DB: Pinecone (with in-memory fallback)

---

🛠️ Tech Stack

Backend

• Framework: FastAPI (Python)
• AI/ML: PyTorch, Transformers, spaCy, LangChain
• LLM Providers: Groq, Google Gemini
• Vector Database: Pinecone
• Document Processing: PyPDF, python-docx

Frontend

• Framework: React + Vite
• Styling: Tailwind CSS
• UI Components: Radix UI, Lucide React
• Visualization: Recharts, Framer Motion

---

📋 Prerequisites

• Python 3.9+
• Node.js 16+
• API Keys:
  • GROQ_API_KEY (for RAG and Clause Extraction)
  • GOOGLE_API_KEY (for Embeddings)
  • PINECONE_API_KEY (Optional, for production vector storage)

---

⚡ Installation & Setup

1. Backend Setup

Navigate to the backend directory:

cd "B3-Developing-Named-Entity-Recognition-NER-Models-for-Financial-Data-Extraction--backend"

Create a virtual environment (recommended):

python -m venv venv
# Windows
./venv/Scripts/activate
# Linux/Mac
source venv/bin/activate

Install dependencies:

pip install -r requirements.txt

Configuration: Create a .env file in the backend directory with your API keys:

GROQ_API_KEY=your_groq_key
GOOGLE_API_KEY=your_google_key
PINECONE_API_KEY=your_pinecone_key (optional)

Run the server:

python app.py

The backend will start on http://localhost:8001

2. Frontend Setup

Navigate to the frontend directory:

cd "Financial-Insight---AI-Powered-Document-Analysis-Platform"

Install dependencies:

npm install

Run the development server:

npm run dev

The frontend will start on http://localhost:5173 (typically)

---

📖 Usage Guide

1. Upload: Go to the "Upload" page and select a PDF, DOCX, or TXT file.
2. Analyze: The system will automatically process the document.
3. View Results:
   • Dashboard: See high-level metrics.
   • NER: Explore extracted entities.
   • Sentiment: View sentiment distribution and heatmap.
   • Clauses: Review extracted legal clauses and risk profile.
4. Chat: Switch to the "RAG" tab to ask specific questions about the uploaded document.

---

📄 API Documentation

Full API documentation is available when the backend is running at:

• Swagger UI: http://localhost:8001/docs
• ReDoc: http://localhost:8001/redoc

---

Built with ❤️ for Financial Insight