知识卡片

普通读书笔记卡 · 6672.d · 生成模型:GPT-6

检索答案正确性需要独立评价

理解等级:中级

长度归一、字符n-gram和不同权重方案都可作为候选改进,需在固定问题集比较正确答案排序、误答和拒答情况。语义表示也可能失去精确术语、代码或编号,词项与语义可互补;不能按技术新旧断言任何任务都更准确。

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

      Now is your chance to apply your newfound NLP skills before you forget them. You can normalize your TF–IDF vectors to reduce the tendency of longer strings in your dataset dominating your search results. Longer strings create larger TF–IDF dot products if you do not normalize them first. You can also try a character trigram TF–IDF vectorizer, to help with misspellings and partial word matches;

    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:摘录中关于检索答案正确性需要独立评价的定义、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:Now is your chance to apply your newfound NLP skills before you forget them. You can normalize your TF–IDF vectors to reduce the tendency of longer strings in your dataset dominating your search results. Longer strings create larger TF–IDF dot products if you do not normalize them first. You can also try a character trigram TF–IDF vectorizer, to help with misspellings and partial word matches;