Agentic RAG with Knowledge Graphs for Complex Multi-Hop Reasoning in Real-World Applications

TL;DR

INRAExplorer is an agentic RAG system that leverages a knowledge graph and multi-tool architecture to perform complex, multi-hop reasoning and targeted retrieval in scientific data, improving knowledge interaction in specialized domains.

Contribution

It introduces INRAExplorer, a novel agentic RAG system that integrates a knowledge graph with LLMs for advanced multi-hop reasoning and comprehensive data retrieval in scientific research.

Findings

Enables iterative, targeted queries for exhaustive datasets

Performs multi-hop reasoning over complex entity relationships

Delivers structured, comprehensive answers in specialized fields

Abstract

Conventional Retrieval-Augmented Generation (RAG) systems enhance Large Language Models (LLMs) but often fall short on complex queries, delivering limited, extractive answers and struggling with multiple targeted retrievals or navigating intricate entity relationships. This is a critical gap in knowledge-intensive domains. We introduce INRAExplorer, an agentic RAG system for exploring the scientific data of INRAE (France's National Research Institute for Agriculture, Food and Environment). INRAExplorer employs an LLM-based agent with a multi-tool architecture to dynamically engage a rich knowledge base, through a comprehensive knowledge graph derived from open access INRAE publications. This design empowers INRAExplorer to conduct iterative, targeted queries, retrieve exhaustive datasets (e.g., all publications by an author), perform multi-hop reasoning, and deliver structured, comprehensive answers. INRAExplorer serves as a concrete illustration of enhancing knowledge interaction in specialized fields.