← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 6671.b · 生成模型:GPT-6

倒排索引缩小需评分的候选

理解等级:中级

倒排表从词项指向出现它的文档或位置,可先访问查询词涉及的列表,减少逐文档遍历。查询代价仍受词频、列表合并、候选数及排名影响,不能笼统称全文检索恒为O(1);三元字符索引也是可选方案,非全部数据库必用。

这张卡片看不懂?

已有反馈 · 0 条

    还没有人提交反馈。

    在整书笔记中的位置

    原书出处 · 1 处

    • · 查看原书依据

      The argmax() function in listing 3.17 required a sequential scan (table scan) of each and every one of the TF–IDF vectors and their values. A sequential scan is when a database system must iterate through all of the items in a table to find a match for the query; it is an O(n) operation. In comparison, most database-indexing approaches, such as B-trees and R-trees, provide

    参考来源

    - 位置:《Natural Language Processing in Action, Second Edition》第03章《词频与逆文档频率向量》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Natural Language Processing in Action, Second Edition/_epub-src/OEBPS/Text/chapter-3.html) - 结论依据:摘录中关于倒排索引缩小需评分的候选的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。 - 原始内容:The argmax() function in listing 3.17 required a sequential scan (table scan) of each and every one of the TF–IDF vectors and their values. A sequential scan is when a database system must iterate through all of the items in a table to find a match for the query; it is an O(n) operation. In comparison, most database-indexing approaches, such as B-trees and R-trees, provide