Boost search results with ai vector results

To explain my setup better. I have a pinecone setup and when I try to search a keyword "γενεθλια" Vector DB Explorer ρετθρν σοme results as you can see. But on my search result of drupal api not included. Any help on this? Have I setup something wrong?

Hi! Might be the 0,2 instead of 0.2 for relevance threshold

Otherwise can you confirm main View for search is Search API Database? For SOLR there is 📌 Check that SOLR 'boost' of results finds results that are not found by traditional SOLR search Active open which means SOLR for now only boosts results that are also found by main View whereas database version also finds results that are not otherwise found by the database search alone

Thanks for reply. I have change to 0.2 but still no results. My view is search api. A block for search and a view results for presentation.

Check at https://viosarp.gr/el/search?search_api_fulltext=%CE%93%CE%B5%CE%BD%CE%A...

From 1:06 in the video here I do a walk through of setup https://www.drupal.org/project/ai/issues/3478581 🌱 #2 Drupal AI Online Meetup Active
At 1:40 I get into combining search api database and ai search

Hope that helps!

Thanks for your reply.
I have follow your instruction and I have create a new view that return results from AI search index of my pinecone.

My question is how to edit my existing view of search results to include (prepend) results from ai index. I have already enable "Boost Database by AI Search" at processors of my standard index. But no results. If any documentation how to activate AI boost search will be helpful.

I try to find a solution viewing your youtube video.

My question is: the view must be the index of pinecone or can be the index of normal search api and at this to add boost search api? Check image to explain it better.

For #3 I have try to change to 0.2 from 0,2 but with no success. Despite I write 0.2 it saves 0,2. Check image screenshot.

One more question as I try to find what I do wrong.

When I do the search API index, at Vector Database indexing options, can I use only type of "Rendered HTMNL output" when I add fields?
Can I boost my search index with ai boost if the structure of entities results are different?

Sorry not very around at the moment but maybe first step is enabled ai api Explorer to make sure you're getting results in the vector database test section there.

Not necessary to have the indexed fields match at all, essentially the boost:

1) programmatically runs the free text search against the vector database to fetch entity IDs only

2) the does something like this pseudo query into database:
"WHERE
(keywords LIKE" %normal database search%" OR item id IN (IDs retrieved from vector database)
AND (other filters & exposed filters
ORDER BY
Vector database retrieved item IDs DESC,
Relevance from database search DESC,
Any other orders you have applied DESC
"

You can e.g. in the boost plugin code add a dvm to check item IDs are found in (1) and that should be exactly what the AI API Explorer submodule shows to make sure that part is working okay

Hope that helps keep you going on debugging