Hello.
Context: We have prospects asking to combine App Search & Semantic Search.
What is the best strategy?
Specific Questions:
Is there a curation like functionality in Semantic Search, already available today or in a roadmap?
Incase this is not present/on the roadmap, what is the best way to foresee this ourselves?
We had the following idea to implement this ourselves if this was not present and were wondering if this is a good enough solution?
We currently have an API that handles the search.
This API would first send a request to an index filled with the curations, and matches the query, example
GET currations-index/_search
{
"query": {
"match": {
"query": "I am a search query"
}
}
}
This would return an array of ids.
After with these ID's we would get the embeddings from the inference endpoint.
We then would send a pinned query with a knn query as the body of the organic field.
This would in theory have the same functionality as how a curation would work ( at least for promoting the field, not hiding one)
Would a rule query be better here and put all the "curations" inside a query ruleset ?
Any insights are appreciated.
Kind regards,
Chenko