Pramana

Epistemic Reasoning Engine for AI Systems Based on Navya-Nyaya Logic

Pramana is a research project that builds an epistemic reasoning engine for AI systems. It teaches LLMs to apply the 2,500-year-old Indian epistemological methodology of Navya-Nyaya logic to solve logical problems systematically, rather than relying on probabilistic pattern-matching.

Overview

Unlike standard chain-of-thought reasoning, Pramana enforces a structured 6-phase Nyaya methodology:

1. Samshaya (Doubt Analysis) - Classify the type of uncertainty/ambiguity
2. Pramana (Evidence Sources) - Identify valid knowledge sources (Pratyaksha/perception, Anumana/inference, Upamana/comparison, Shabda/testimony)
3. Pancha Avayava (5-Member Syllogism) - Construct formal argument with Pratijna (thesis), Hetu (reason), Udaharana (universal example), Upanaya (application), Nigamana (conclusion)
4. Tarka (Counterfactual Testing) - Use reductio ad absurdum to verify conclusions
5. Hetvabhasa (Fallacy Detection) - Check for reasoning errors
6. Nirnaya (Ascertainment) - Reach definitive conclusion or explicitly state insufficient evidence

Installation

# Clone the repository
git clone https://github.com/your-org/pramana.git
cd pramana

# Install with uv (recommended)
uv sync --dev

# Or install with pip
pip install -e ".[dev]"

Development

# Run tests
uv run pytest

# Run linter
uv run ruff check

# Run type checker
uv run mypy src

# Format code
uv run ruff format

Project Structure

pramana/
├── src/pramana/
│   ├── domain/          # Core business logic
│   ├── application/     # Use cases and orchestration
│   ├── infrastructure/  # External integrations
│   ├── config/          # Configuration management
│   └── cli/             # Command-line interface
├── tests/               # Test suite
├── configs/             # YAML configuration files
├── data/                # Training and evaluation data
└── docs/                # Documentation

License

MIT License - see LICENSE for details.