Replies: 1 comment 1 reply
-
You'll want to look at the index for adjusting what data is stored; it also contains most of the logic for updating page metadata when a page changes. The markdown importer handles parsing markdown files on a background web worker and would probably be where you do any heavyweight embedding. For exposing this, you'd either want to add it to the JS api (easier) or to the query language (several more steps involved). |
Beta Was this translation helpful? Give feedback.
1 reply
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hello,
I was thinking about adding semantic search capabilities to the dataview API. Would be great if someone can point me to the right files to understand how indexing works in dataview and any caveats to consider. I especially would like to understand how the index is updated. My hypothesis is that the combinaton of structured data with the flexibility of semmantic search will be necessarily superior to both.
My thinking at the moment is basically to generate embeddings for every page where a embeddings are generated for itemized level of within dataview (page, section, list, task, etc). Would also like to open a discussion on how to extend the query API for semantic search capabilities.
Anyways, open to discussion.
Beta Was this translation helpful? Give feedback.
All reactions