LLM Logic Demos

A collection of demonstrations exploring the synergy between Large Language Models (LLMs) and various formal logic and reasoning systems. This repository provides code examples to illustrate how to integrate LLMs with techniques like logic programming, knowledge graph traversal, and advanced reasoning frameworks to create more robust and explainable applications.

The goal is to move beyond simple prompt-based LLM interactions by grounding the models in structured, symbolic logic to enhance reasoning capabilities and reduce hallucinations.

---

Repository Structure

The repository is organized into distinct folders, each containing a specific set of logic and LLM integration demos:

Directory	Description
Backward Chaining Engine	Demonstrations of implementing or utilizing a backward chaining inference engine, often used in expert systems, and how an LLM can interact with or formulate the initial goals and rules.
LangChain and LangGraph	Examples utilizing the powerful LangChain and LangGraph frameworks to build complex, multi-step LLM agents. Demos here may show how to construct conversational agents or reasoning pipelines that incorporate external knowledge and logical steps.
datalog	Code demonstrating the use of Datalog, a declarative logic programming language, for knowledge representation and querying, potentially using an LLM to generate Datalog queries or interpret results.
prolog	Demos focused on Prolog, a prominent logic programming language. This section explores how to bridge LLM outputs (e.g., extracting facts or rules) with a Prolog engine to perform sophisticated symbolic reasoning.
requirements.txt	A list of necessary Python packages required to run the demonstrations.

---

Getting Started

Prerequisites

• Python (3.9+)
• API keys for the LLM services used in the demos (e.g., OpenAI, Anthropic, etc.), which you may need to configure as environment variables.

Installation

1. Clone the repository:

   git clone [https://github.com/sakisaki-dev/LLM-Logic-Demos.git](https://github.com/sakisaki-dev/LLM-Logic-Demos.git)
   cd LLM-Logic-Demos

2. Set up the environment: It is recommended to use a virtual environment:

   python -m venv venv
   source venv/bin/activate  # On Windows, use `venv\Scripts\activate`

3. Install dependencies:

   pip install -r requirements.txt

Usage

Navigate into any of the subdirectories and follow the instructions or run the Python scripts/Jupyter notebooks provided within. Each folder is a self-contained demonstration of a specific logic-LLM integration pattern.

For example, to explore the Prolog demos:

cd prolog
python your_prolog_demo.py # (or run the Jupyter notebook)