知识卡片

普通读书笔记卡 · 6659.b · 生成模型:GPT-6

归一化可能提高召回并降低区分力

理解等级:中级

把不同拼写、词形或同义线索合并,可能找回此前漏掉的文档,也可能引入不相关匹配。应按相关文档计算召回,不是返回结果越多就越好;结果中不相关所占比例是1减精确率,不能误称总体假阳性率。

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

      By generalizing your model to work with text that has odd capitalization, case normalization can reduce overfitting for your machine learning pipeline. Case normalization is particularly useful for a search engine. For search, normalization increases the number of matches found for a particular query. This is often called the recall performance metric for a search engine (or any other classification model).30

    参考来源

    - 位置:《Natural Language Processing in Action, Second Edition》第02章《思维的词元》(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-2.html) - 结论依据:摘录中关于归一化可能提高召回并降低区分力的定义、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:By generalizing your model to work with text that has odd capitalization, case normalization can reduce overfitting for your machine learning pipeline. Case normalization is particularly useful for a search engine. For search, normalization increases the number of matches found for a particular query. This is often called the recall performance metric for a search engine (or any other classification model).30