Knowledge Graph of Thoughts (KGoT)

Architecture Overview | Setup Guide | Quick Start | Citations

This is the official implementation of Affordable AI Assistants with Knowledge Graph of Thoughts.

Knowledge Graph of Thoughts (KGoT) is an innovative AI assistant architecture that integrates LLM reasoning with dynamically constructed knowledge graphs (KGs). KGoT extracts and structures task-relevant knowledge into a dynamic KG representation, iteratively enhanced through external tools such as math solvers, web crawlers, and Python scripts. Such structured representation of task-relevant knowledge enables low-cost models to solve complex tasks effectively.

Architecture Overview

The KGoT system is designed as a modular and flexible framework that consists of three main components: the Controller, the Graph Store, and the Integrated Tools, each playing a critical role in the task-solving process.

• The Controller component offers fine-grained control over the customizable parameters in the KGoT pipeline and orchestrates the KG-based reasoning procedure.
• The Graph Store component provides a modular interface for supporting various Knowledge Graph Backends. We initially support Neo4j, NetworkX and RDF4J.
• The Integrated Tools component allows for flexible and extensible Tool Usage and enables the multi-modal reasoning capabilities of the framework.

Setup Guide

In order to use this framework, you need to have a working installation of Python 3.10 or newer.

Installing KGoT

Before running the installation, make sure to activate your Python environment (if any) beforehand.

git clone https://github.com/spcl/knowledge-graph-of-thoughts.git
cd knowledge-graph-of-thoughts/
pip install -e .
playwright install

Configuring API Keys and Models

To get started make a copy of the following template files inside the kgot directory:

• kgot/config_llms.template.json
• kgot/config_tools.template.json

Then rename them as follows:

• config_llms.template.json → config_llms.json
• config_tools.template.json → config_tools.json

Please update the API keys, if necessary, for the language models you intend to use in the kgot/config_llms.json file. You can also add new models by incorporating their information into the JSON file. The object key is the language model identifier used in KGoT, and the various attributes contain the information needed to run the model.

Local models are expected to be hosted using Ollama. KGoT assumes that the model is accessible at the default Ollama...