1 – Recommenders already use vectors of products and users to improve predictions with few events.
2 – Vector databases already deliver predictions with similarity matching.
3 – An example is Algolia‘s new NeuralSearch.
Here are two extracts from their NeuralSearch documentation:
– “NeuralSearch combines the precision of keyword search with the deep understanding of natural language and contextual relevance provided by AI-based vector search. ”
Algolia NeuralSearch is indeed a vector search and vector database.
– “You need to collect at least 1,000 clickedObjectIDsAfterSearch or 100 convertedObjectIDsAfterSearch events within 30 days to activate NeuralSearch. If you have fewer events, you can’t use NeuralSearch.”
Algolia NeuralSearch is also a recommender system, feeding on user events.
But you cannot use Algolia NeuralSearch as a pure vector search, nor as a pure recommender.
This is a an hybrid object of a new kind.