In this presentation, we will discuss two contemporary methods for representing human knowledge (or memory).
We present a step-by-step Question-Answer task that integrates a Large Language Model (LLM) with a Knowledge Graph (KG) for a specific domain.
The KG provides the LLM with specific context (such as a hospital or company), enabling the LLM to generate detailed and accurate domain-specific Outputs, without hallucinations.
To achieve this integration, we will cover fundamental concepts related to Knowledge Graphs, Large Language Models, and the UMLS database.
Finally, we will show how Python tools can be used to integrate the LLM and KG into a local application.
VISTA Lab