OriginTrail dRAG DeSci Example

This repository provides an API for natural-language querying the Decentralized Science Knowledge Graph using a combination of LLM-powered SPARQL queries and vector similarity search.

Overview

The dRAG DeSci Example allows you to:

• Ask natural language questions about scientific papers and research
• Automatically generate SPARQL queries to extract relevant metadata, terms, titles and IDs from the DeSci Knowledge Graph
• Enhance results by querying a vector database containing paper summaries
• Receive natural language answers along with the Knowledge Assets used to generate the response

Key Features

• LLM-powered query generation and natural language responses
• Integration with OriginTrail Decentralized Knowledge Graph
• Vector similarity search for enhanced information retrieval
• Returns answers and Knowledge Assets of science papers used to generate the response

Getting Started

1. Check SETUP.md for detailed setup instructions
2. See .env.example for required environment variables configuration

Prerequisites

• Node.js
• MySQL database
• Access to OriginTrail DKG node
• Vector database (Milvus)
• API keys for LLM services

For detailed setup instructions and configuration, please refer to the setup guide.