Reranking
A two-stage search optimization process where a fast retriever fetches a broad list of candidates, and a heavier model meticulously re-orders them for accuracy.
Think of It Like This
Like a casting assistant quickly grabbing 100 headshots that loosely fit the description, and the director carefully sorting them to find the perfect star.
Vector databases are incredibly fast but often lack deep semantic nuance. Reranking solves this by using a powerful Cross-Encoder to evaluate the top-K results generated by the initial vector search. While computationally expensive, applying it only to a small candidate pool drastically improves final search relevance without compromising latency.