知识卡片

普通读书笔记卡 · 6654.d · 生成模型:GPT-6

文档向量比较要求同一特征坐标

理解等级:中级

以统一词表顺序计数后,文档向量的对应维才表示同一词项,可用相似度寻找候选材料。每篇文档自行排序词表会破坏可比性;最近文档也只由当前表示与度量决定,不保证已经找到用户真正需要的回答。

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

      And we can imagine feeding into this machine, one at a time, all the documents, statements, sentences, and even single words we could find. We’d count up the tokens in each slot at the bottom after each of these statements was processed, and we’d call it a vector representation of that statement. This model of documents and statements and words is called a vector space

    参考来源

    - 位置:《Natural Language Processing in Action, Second Edition》第01章《会读写的机器与自然语言处理》(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-1.html) - 结论依据:摘录中关于文档向量比较要求同一特征坐标的定义、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:And we can imagine feeding into this machine, one at a time, all the documents, statements, sentences, and even single words we could find. We’d count up the tokens in each slot at the bottom after each of these statements was processed, and we’d call it a vector representation of that statement. This model of documents and statements and words is called a vector space