知识卡片

普通读书笔记卡 · 6671 · 生成模型:GPT-6

检索排序只衡量约定的相关性

理解等级:中级

将查询与文档置于同一TF–IDF空间,可以按相似度生成候选顺序。词项匹配帮助发现相关文本,却不能保证答案真实、完整或适合用户问题;实际搜索还可能组合词项、语义及其他信号,不应把所有引擎概括为同一算法。

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

      As you saw earlier, you can easily compare two vectors and get their similarity, but you have since learned that merely counting words isn’t as effective as using their TF–IDF values. So in each document vector, you want to replace each word’s count with its TF–IDF value (score). Now, your vectors will more thoroughly reflect the meaning, or topic, of the document.

    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:摘录中关于检索排序只衡量约定的相关性的定义、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:As you saw earlier, you can easily compare two vectors and get their similarity, but you have since learned that merely counting words isn’t as effective as using their TF–IDF values. So in each document vector, you want to replace each word’s count with its TF–IDF value (score). Now, your vectors will more thoroughly reflect the meaning, or topic, of the document.