用户满意度调查

本指南向你展示如何创建调查流来收集用户对其对话体验的反馈。你可以通过在流末尾使用 link 步骤（而不是 END 步骤）来实现此目的。

注意

此解决方案依赖于 Rasa Pro v3.9 及更高版本中的功能。

添加调查流

首先，创建一个调查流，其中的流守卫始终为 false。这确保我们的流只能从另一个活动流中启动。

feedback.yml

flows:
  collect_feedback:
    description: Collect user satisfaction feedback
    if: False
    steps:
      - collect: user_satisfaction

将相应的槽和话语添加到领域中：

domain.yml

slots:
  user_satisfaction:
    type: float

responses:
  utter_ask_user_satisfaction:
    - text: |
        On a scale from 1 (not at all satisfied) to 5 (completely satisfied),
        how satisfied are you with this conversation?
      buttons:
        - payload: "/SetSlots(user_satisfaction=1)"
          title: "1 (not satisfied)"
        - payload: "/SetSlots(user_satisfaction=2)"
          title: "2 (somewhat unsatisfied)"
        - payload: "/SetSlots(user_satisfaction=3)"
          title: "3 (neutral)"
        - payload: "/SetSlots(user_satisfaction=4)"
          title: "4 (somewhat satisfied)"
        - payload: "/SetSlots(user_satisfaction=5)"
          title: "5 (very satisfied)"

最后，添加 link 步骤以启动调查流。在此示例中，我们从 CALM 模板对话机器人开始，并将链接添加到联系人流列表中：

list_contacts.yml

flows:
  list_contacts:
    name: list your contacts
    description: show your contact list
    steps:
      - action: list_contacts
        next:
          - if: "slots.contacts_list"
            then:
              - action: utter_list_contacts
              - link: collect_feedback
          - else:
              - action: utter_no_contacts
              - link: collect_feedback

训练对话机器人并使用 rasa inspect 启动它。向机器人发送消息 "list my contacts" 以触发响应。