← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 6664 · 生成模型:GPT-6

朴素贝叶斯分类并非词典打分复制

理解等级:中级

多项朴素贝叶斯从各类的词项频数估计条件概率,按模型假设组合得到类别分数。系数通常对应类条件对数概率,先验与平滑也参与判断;学习得到参数不代表每项都客观最优,更不同于人工极性词典的单分数。

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

      A naive Bayes model tries to find keywords in a set of documents that are predictive of your target (output) variable. When your target variable is the sentiment you are trying to predict, the model will find words that predict that sentiment. The nice thing about a naive Bayes model is that the internal coefficients will map words or tokens to scores just like VADER

    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:摘录中关于朴素贝叶斯分类并非词典打分复制的定义、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:A naive Bayes model tries to find keywords in a set of documents that are predictive of your target (output) variable. When your target variable is the sentiment you are trying to predict, the model will find words that predict that sentiment. The nice thing about a naive Bayes model is that the internal coefficients will map words or tokens to scores just like VADER