Can I add multiple embeddings into one record and run similarity searches on multiple embeddings?

Same as the title, I have been using Azure AI Search as a vector database of which columns are content and summary. and I have been running similarity searches on both the content field and summary field at the same time. does Pinecone offer such feature as well?

Hi @jason.kang, we do not currently support this functionality. I’ll relay your interest to our product team and transfer this post to the Feature Requests section of our forums, which will allow other members to upvote.

Thank you! looking forward to it

It’s not exactly the same thing, but you might want to look into using our Hybrid Search.

Also, you can embed both the ‘content’ and the ‘summary’ columns as separate vectors in Pinecone and then use metadata to identify which is which and link them back to the same source record…then when you query, you will be querying against both of them at the same time and getting back whichever results are most relevant. Or you could use the metadata filter to do two queries: one for ‘content’ and one for ‘summary’

1 Like