Hi, I’ve just started looking into Weaviate and I am wondering if there are best practices for interacting with tabular data and semantic search. For example, this might be something like fetching a row of data based on a semantic match in a particular column. I guess this can be done by using two step search - one to find the row and then another to return the row. It is still not clear to me how Weaviate combines vectors and objects ,so I thought this might be a use case where I could get a better understanding of Weaviate’s advantages? Thanks
Hi @markNZed ! Welcome to our community
That’s an interesting question. While you can pass your tabular data in a unstructured way, so it create the vectors and allow similarity search, retrieving an exact column and row would be more complicated.
You could try chunking the row into each object, and maybe generate the response using the generative module?
can you show me eaxmples?
I don’t have an example but the basic idea of chunking the data into rows seems a good approach.