Dense search
GPT5.icon
Dense Retrieval is a search method that transforms queries and documents into **dense vectors (embedding)** in a learned model, and finds similarities by vector similarity (cosine/inner product).
It is more resistant to putting in other words and semantic proximity than sparse search (e.g., BM25), which focuses on keyword matching.
---
This page is auto-translated from /nishio/デンスな検索 using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I'm very happy to spread my thought to non-Japanese readers.