Find answers from the community

Updated 2 months ago

Csv

I converted to sqlite as it is structured and have tried the structured data tutorial.
It seems to work, but not the way I expected. I think I might change the field names to be more unique and less generic.

I'm searching through a taxonomy, looking for the catagory which is best, based on a description. the taxonomy has a description too.

1 comment

LLogan M

You could just parse each row in the csv as a document object, which might also work well with a vector index (just might have to increase the top k if the rows are small, default top k is 2)

Plain Text

doc= Document(text=<text string from row>)
....
index.as_query_engine(similarity_top_k=4)

Add a reply