Athena's Asylum 🤖📚

A context-aware AI assistant that combines document understanding, knowledge graphs, and witty personality to deliver intelligent responses.

Features ✨

• PDF Context Processing: Extracts and analyzes text from PDF documents
• Knowledge Graph Integration: Stores entities and relationships in Neo4j
• Vector Search: FAISS-based semantic search for document context
• Hybrid Reasoning: Combines LLM capabilities with structured knowledge
• Personality-infused Responses: Witty and charming interaction style

Technologies 🛠️

• Natural Language Processing: spaCy (en_core_web_sm)
• Knowledge Graph: Neo4j
• LLMs: Anthropic Claude 3 Sonnet & OpenAI Embeddings
• Vector Store: FAISS
• Document Processing: PyPDF, LangChain

Installation ⚙️

1. Clone repository:

   git clone https://github.com/yourusername/athenas-asylum.git
   cd athenas-asylum

2. Install dependencies:

   pip install -r requirements.txt
   python -m spacy download en_core_web_sm

3. Set up services:

• Install and run Neo4j locally
• Obtain API keys for Anthropic and OpenAI

Configuration 🔧

1. Replace hardcoded credentials in:

• llm_client.py (Anthropic & OpenAI API keys)
• vector_store.py (OpenAI API key)
• main.py (Neo4j connection details)

2. Recommended security practice:

   # Replace hardcoded keys with environment variables:
   os.getenv("ANTHROPIC_API_KEY")
   os.getenv("OPENAI_API_KEY")

3. Place your PDF document in the project root as r.pdf or modify path in main.py

Usage 🚀

python main.py

Example Interaction:

You: What's the main topic of the document?
Assistant: The document focuses on AI ethics - though I must say, 
          your choice of questions is as sharp as Athena's owl! 🦉

Project Structure 🗂️

File	Purpose	Key Components
main.py	Entry point for application flow and user interaction	- Initializes all components - Chat loop logic - Context aggregation
knowledge_graph.py	Manages Neo4j knowledge graph operations	- Entity extraction with spaCy - Cypher query execution - Relationship mapping
llm_client.py	Handles AI model interactions and response generation	- Claude-3 Sonnet integration - OpenAI embeddings - Personality system prompts
pdf_processor.py	Processes PDF documents for text extraction and preparation	- PyPDF text extraction - Text cleaning - Recursive chunking
vector_store.py	Implements semantic search capabilities	- FAISS vector database - OpenAI embeddings integration - Context retrieval
requirements.txt	Lists all project dependencies	- NLP libraries - Database drivers - AI service SDKs

Development Requirements 🔨

Core Prerequisites

• Python 3.9+ with pip package manager
• Neo4j Desktop (v4.4+) running locally
• 4GB+ RAM for optimal performance

API Access

• ✅ Valid Anthropic API key (Claude-3 access)
• ✅ OpenAI API key (text-embedding-3-small model)
• Active internet connection for LLM services

Recommended Setup

1. Create virtual environment:

   python -m venv athena_env
   source athena_env/bin/activate  # Linux/Mac

2. Configure Neo4j:

   neo4j start
   # Set password to 'your_password' or update in main.py

License & Attribution 📜

MIT License - Full text available in LICENSE file

Third-Party Services

• Anthropic Claude-3 Sonnet (Terms)

• OpenAI Embeddings (Policies)

• Neo4j Community Edition (License)

Hot Tips 🔥

1. For complex PDFs: Pre-process with OCR using pdf_processor.py extensions

2. Visualize knowledge graph: Access Neo4j Browser at http://localhost:7474

3. Try hybrid queries: "Explain Section 2.3 using both document context and knowledge graph entities"

"Wisdom begins in wonder" - Socrates (Athena's favorite philosopher) 🏛️