知识卡片
消息日志要记录可重建的上下文
保留输入、当前节点、选中意图及动作结果,有助复现误解和设计缺口。日志字段应与排查目标对应,避免把全部私密历史无差别复制到每条记录;仅有最终机器人回复无法判定前序决策是否正确。
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
For the message history, you can record conversations in a MessageLog table. You will need this to be able to analyze what your users are saying to your chatbot. And you can use this message log as a source of examples to label with intents so that you can periodically retrain your intent recognition system. Each user session represents a path through your conversation graph.
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:摘录中关于消息日志要记录可重建的上下文的定义、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:For the message history, you can record conversations in a MessageLog table. You will need this to be able to analyze what your users are saying to your chatbot. And you can use this message log as a source of examples to label with intents so that you can periodically retrain your intent recognition system. Each user session represents a path through your conversation graph.