Closed K-Schubert closed 2 months ago
Setup a document retrieval endpoint in backend.
Do a semantic similarity match (cosine similarity) on indexed documents in postgres vector DB.
Takes as input a user query such as: {"query": query} and returns a response such as {"contextDocs": docs, "sourceUrl": url}.
Will be called from the frontend when a user inputs a query that doesn't match an autocomplete suggestion.
Setup a document retrieval endpoint in backend.
Do a semantic similarity match (cosine similarity) on indexed documents in postgres vector DB.
Takes as input a user query such as: {"query": query} and returns a response such as {"contextDocs": docs, "sourceUrl": url}.
Will be called from the frontend when a user inputs a query that doesn't match an autocomplete suggestion.