Generating graph databases using large language models

A computer-implemented method for generating a graph database comprising:

• processing user input and stored metadata using a first processor configured by code to generate a prompt for input to a large language model (LLM);
• dynamically supplementing the prompt using stored metadata from a relational database as well as a vectorized metadata index in a fused prompt building process;
• using a second processor executing the LLM:
• extracting graph metadata from the prompt;
• generating a knowledge graph based on the graph metadata and using a knowledge graph platform that uses mapped extracted entities;
• using a third processor:
• generating a graph data model including both structured and unstructured data using the knowledge graph from both structured and unstructured data sources;
• and
• generating a graph on a graphic platform using the graph data model,
• wherein the step of generating the graph model includes:
• refining metadata obtained in response to a user query to reduce a number of tokens for input to an LLM;
• identifying parts of speech in the user query input,
• detecting entities, relationships, and attributes in the refined metadata using an LLM, and
• mapping extracted entities to file headers;
• filtering the detected entities, relationships, and attributes based on a data domain model and metadata production standards.