Technology

GraphRAG

Graph RAG (Graph-based Retrieval-Augmented Generation) is an advanced technique that improves upon the standard Retrieval-Augmented Generation (RAG) approach using a Knowledge Graph.

Used in the following Interventions

Deploy a RAG LLM stack with a Knowledge Graph

Use a Knowledge Graph as the source for information retrieval to provide the LLM with correct, scoped, and contextual information.

Other technologies used

Machine Learning (ML)

Artificial Intelligence (AI)

Large Language Model (LLM) AI

Retrieval-Augmented Generation (RAG) with LLMs

Involved in these Products

Data Graphs

Data Graphs is the fastest speed-to-market of all the knowledge graphs. Go from visually creating your ontology to having a live platform with API, webhooks, and a slick UI, within minutes.

Graph RAG features for Data Graphs - Data Language Summer 2024 Hack Event

Data Graphs

Our summer's hack event is spearheading new RAG Stack features for Data Graphs

Matt Shearer

July 9, 2024

FAQs

How is Graph RAG better than standard RAG?

Graph RAG is better than standard RAG because it provides the LLM with richer information:

Structured entity information: This includes details and properties of entities, not just text chunks.
Relationships between entities: This allows the LLM to understand the bigger picture and connections between concepts.

Overall, Graph RAG enhances the LLM's ability to generate more accurate, contextual, and insightful responses.

Graph RAG vs Standard RAG

Standard RAG: It retrieves relevant information from a knowledge base (like a collection of documents) to feed a large language model (LLM). This can lead to limitations in context and factual accuracy.
Graph RAG: It incorporates a graph database as the knowledge source. This database structures information as a network of interconnected entities and their relationships.