Hello,
I was wondering if Neo4j could consider both the following:
- Pre-filtering vector queries (I already saw posts about this topic)
- enable
queryNodes
to return results ordered by score ascending (lower score first)
An example use case could be: Recommending activities within a specific geolocation. The recommendations would be a sum of top X similar and top X dissimilar activities compared to what the user already perform in the past.
Technically, we would create a cypher query prefiltering activities based on the user location and filtering out the ones the user already done in that area. Then, from the result, run a vector search ordered by score descending for to find the most similar activities to what the user already did and combine those result with the same queryNodes search but ordered by lowest score first to offer the top activities the most innovant compared to what the user already did.
I believe this is not possible today, could those 2 new abilities be considered?
Thank you