// the find
wangshusen/SearchEngine
搜索引擎原理
A Chinese-language course on search engine internals, covering the full stack from relevance modeling and BERT-based ranking to inverted indexes, vector recall, and query rewriting. It's lecture slides and a PDF book, not code — this is educational material for engineers who want to understand how production search systems actually work. Primarily useful if you read Chinese.
The scope is unusually complete for a free resource: it covers the full retrieval pipeline from tokenization through multi-stage ranking and query recommendation, not just IR basics. The BERT relevance modeling section goes into both inference and training, which most intro materials skip. Having both slides and a companion book PDF in one place is convenient. The YouTube/Bilibili dual-hosting means the lectures are actually accessible.
It's entirely in Chinese with no translations, which cuts the potential audience significantly. There's no code — no reference implementations, no notebooks, no datasets — so you can't run anything or validate your understanding against real outputs. Several sections in the outline (query processing, ranking fusion) have no slides linked yet, suggesting the course was never finished. Last updated April 2024, so the LLM-era retrieval landscape (RAG, hybrid search, re-rankers) is largely absent.