知识卡片
分类评估应看各标签及聚合方式
意图分类可报告准确率、精确率、召回与F1,多标签还需明确macro、micro和阈值。单一总体分数会掩盖稀少关键意图;按混淆类型检查更利于决定补数据还是改类别定义。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
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
As intent-based chatbots are built on top of a prediction model, we can adopt some of the metrics you’ve met in this book. Remember the accuracy and F1 score we introduced in chapter 4? As a quick reminder, for a binary classifier, accuracy is the ratio of correct predictions out of all the predictions. And F1 score is a harmonic mean of precision and recall,
参考来源
- 位置:《Natural Language Processing in Action, Second Edition》第12章《对话引擎与会话设计》(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-12.html)
- 结论依据:摘录中关于分类评估应看各标签及聚合方式的定义、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:As intent-based chatbots are built on top of a prediction model, we can adopt some of the metrics you’ve met in this book. Remember the accuracy and F1 score we introduced in chapter 4? As a quick reminder, for a binary classifier, accuracy is the ratio of correct predictions out of all the predictions. And F1 score is a harmonic mean of precision and recall,