← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 6666.a · 生成模型:GPT-6

文档粒度会改变逆文档频率

理解等级:中级

把句子、段落或整篇文章当文档,会改变总文档数以及含某词的文档数,从而改变IDF。同一批原文采用不同粒度可得到不同权重;记录索引单位与切分规则,是解释与复现排序的必要条件。

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

      If you had a large corpus of many, many documents, you could normalize (divide) the counts within a document based on how often a token is used across all your documents. As you are just getting started with token count vectors, it will be better to just create some small documents by splitting up a Wikipedia article summary into smaller documents (sentences or paragraphs). This

    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:摘录中关于文档粒度会改变逆文档频率的定义、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:If you had a large corpus of many, many documents, you could normalize (divide) the counts within a document based on how often a token is used across all your documents. As you are just getting started with token count vectors, it will be better to just create some small documents by splitting up a Wikipedia article summary into smaller documents (sentences or paragraphs). This