suggest me resources that help me to modify supabase match_documents function for accurate data retrival.
suggest me resources that help me to modify supabase match_documents function for accurate data retrival.
It is not analyzing all the chunks associated with the meeting transcript.
For example, if meeting A has 30 rows in supabase, it retrieves only 10-15 rows from that meeting & rest of the rows from other meetings. Is this something that can be solved? I have set limit to 50. I can’t pass meeting title to retrive data about that meeting only due to other requirements.
It is already optimal and no modification is necessary.
it shouldn’t be analyzing chunks, it finds the data chunks whose vector embeddings are most similar to the vectorized input text.
If found chunks are irrelevant, then chunking strategy could be the problem or insufficient context is provided while chunking, or the expectations are that vector database works in the same way as a regular database.
Depends on the content, size and structure. Sometimes the size allows not to chunk, other times there are distinct parts to it. There is no one answer to your question.