Checking for existing embeddings. Filter specific queries

Pinecone:
Checking for Existing Embeddings:
To avoid creating duplicate embeddings, you can implement a system to check for existing embeddings before generating new ones. Here’s a potential approach:
a) Create a unique identifier for each loan case file. This could be a combination of the filename and a hash of the file contents. It will be your index.

  1. File-Specific Queries:
    To enable file-specific querying, you can use Pinecone’s metadata filtering capabilities.
    a) When storing embeddings, include metadata that identifies the specific file. This could be the file name or the unique identifier we discussed earlier.
    b) When querying Pinecone, use metadata filters to restrict the search to a specific file.

How implement above things

Hi @anusha.gudipati. Can you share where this suggestion came from? I couldn’t find anything like it in our docs and would like to have more context around what you’re trying to do.