Pg Vector database

0
Hello,I am currently working on RAG for chatbot, which requires the PGVector module. I want to know if PGVector only works with PostgreSQL, or can we also use MSSQL?Also, in the database URL, how does it recognize which knowledge base to use? Or any other method to retrive data from database and give output for chatbot assistant
asked
2 answers
0

Hi Pankaj,

 

the PgVector Knowledge module primarily works with PostgreSQL databases because it makes use of the PgVector extension that was specifically created for Postgres. The main advantage is the storing of vectors and distance calculation that is needed to perform similarity searches (finding nearest knowledge chunks based on a given input).

 

The URL you pass in the configuration is typically the endpoint of your database server, while your individual knowledge bases are tables inside of that database. The name of the table needs to be specified when creating a Connection object which is needed in every operation (for example Insert or Retrieve).

 

Here is some general information about setting up a Vector Database: https://docs.mendix.com/appstore/modules/genai/reference-guide/external-connectors/pgvector-setup/

And the GenAI Showcase App demonstrates RAG in a few examples while the Support Assistant Starter App shows how to use RAG in a chat bot interface.

I hope that helps you.

 

Best regards

Liam

answered
0

Hi Pankaj, I responded to your question in this thread:

https://community.mendix.com/link/space/app-development/questions/143106

 

answered