知识卡片

普通读书笔记卡 · 6665 · 生成模型:GPT-6

词频表示需固定特征坐标

理解等级:中级

词袋按词表把每个文档编码为相同维数的计数向量,才能逐坐标比较。Counter保存词到次数的映射,但迭代顺序不自动成为全语料统一坐标;应按共享词表排列,或借助带标签的对齐,不能直接把两个字典的值按偶然顺序相乘。

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 list of all the unique words in a document or corpus is called its vocabulary or lexicon. Creating your vocabulary is one of the most important steps in your NLP pipeline. If you don’t identify a particular token and give it a place to be stored, your pipeline will completely ignore it. In most NLP pipelines, you will define a single token named <OOV>

    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:摘录中关于词频表示需固定特征坐标的定义、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:The list of all the unique words in a document or corpus is called its vocabulary or lexicon. Creating your vocabulary is one of the most important steps in your NLP pipeline. If you don’t identify a particular token and give it a place to be stored, your pipeline will completely ignore it. In most NLP pipelines, you will define a single token named <OOV>