Skip to content

Metadata Loaders

When a Solution Kit is deployed, there are many tasks that must be run to load subsets of the solution-kit-metadata.json file that must be loaded into TigerGraph so that other tools can use this data.

Here are some tasks that we will need:

  1. Questions Loaded into Vector Store - each question should be run through a process that creates an embedding and this embedding is loaded into a vector store.

  2. Persona Startup Questions - when a chatbot starts up, it will display a short list of sample questions that are used frequently or that demonstrate a specific capability. A list of all questions that have personal-based startup flagged will need to be loaded into TigerGraph as a Question vertex and a GSQL query such as getStartupQuestionForPersona(PERSONA) will be callable by a front end. This will return a JSON array with the question string and the script to call.