Document query based on vector nearest neighbor search

The invention discloses a technical scheme of document query based on vector nearest neighbor search. A vector approximate matching retrieval technology is applied to a search engine; the query content and the webpage document are respectively subjected to semantic vectorization, and the webpage document close to the query content is obtained by utilizing a vector approximate matching retrieval mode, so that the limitation of a symbol matching retrieval mode can be broken through, and a retrieval service capable of better mastering the intention of a user is provided. 本文公开的基于向量最近邻搜索的文档查询的技术方案，将向量近似匹配检索技术应用到搜索引擎中，通过将查询内容和网页文档分别进行语义向量化后，利用向量近似匹配检索的方式，获取与查询内容接近的网页文档，从而能够突破符号匹配的检索方式的限制，提供能够更好地把握用户意图的检索服务。