hi , i want to find out how should i use command to query for non embedded data rows in pinecone. im using langchain and openai embedding for RAG
vectorstore.similarity_search(query)
however this command for embedded rows only . i want to ask for non embedded rows also.
im making chatbot
By non-embedded rows I’m assuming you mean non-embedded metadata that is stored alongside your vectors?
Metadata is not stored as a separate row in Pinecone and is not queryable by itself.
Pinecone is purpose-built exclusively for semantic search across embedded/vectorized data. While performing a semantic search, you can ask Pinecone to include the associated metadata in its response by setting the includeMetadata=True parameter in the query. Langchain abstractions may be hiding this detail from you, or setting it by default.
If querying non-embedded data is your use case (i.e. non-semantic search), I’d suggest using a separate/more traditional database more suited to that type of query.
There is no reason to limit yourself to using only Pinecone.
Hope this helps,
Silas