cs-chatbot-design

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Chatbot Design

聊天机器人设计

Framework

框架

IRON LAW: Intent First, Response Second

A chatbot must UNDERSTAND what the user wants (intent) before crafting
a response. Building response templates without intent classification
produces a keyword-matching FAQ, not a chatbot.

Flow: User message → Intent classification → Slot extraction → Response

IRON LAW: Intent First, Response Second

聊天机器人必须先理解用户的需求（意图），再生成回复。如果不做意图分类就直接构建回复模板，得到的只是基于关键词匹配的FAQ系统，而非真正的聊天机器人。

流程：用户消息 → 意图分类 → 槽位提取 → 生成回复

Core NLU Pipeline

核心NLU流水线

Stage	What It Does	Example
Intent Classification	Identify what the user wants to do	"What time do you close?" → intent: `check_hours`
Entity/Slot Extraction	Extract key information from the message	"Book a table for 4 on Friday" → slots: {party_size: 4, date: Friday}
Dialogue Management	Decide the next action (ask for missing info, confirm, execute)	Missing slot `time` → ask "What time would you like?"
Response Generation	Produce the reply	"I've booked a table for 4 on Friday at 7pm. See you then!"

阶段	功能	示例
意图分类	识别用户的需求	"你们几点关门？" → 意图： `check_hours`
实体/槽位提取	从消息中提取关键信息	"预订周五的4人桌" → 槽位：{party_size: 4, date: Friday}
对话管理	决定下一步操作（询问缺失信息、确认、执行）	缺失槽位 `time` → 询问：“请问您需要几点的桌位？”
回复生成	生成回复内容	"我已为您预订了周五晚7点的4人桌，期待您的光临！"

Intent Design

意图设计

Start with 10-15 core intents covering 80% of user queries
Each intent needs 10-20 training examples (varied phrasings)
Include a
```
fallback
```
intent for unrecognized inputs
Group related intents:
```
order_status
```
,
```
order_cancel
```
,
```
order_modify
```
under "Order Management"

从10-15个核心意图入手，覆盖80%的用户查询场景
每个意图需要10-20条训练样本（表述方式需多样化）
需包含
```
fallback
```
（兜底）意图，用于处理无法识别的输入
可将相关意图分组：例如将
```
order_status
```
、
```
order_cancel
```
、
```
order_modify
```
归为“订单管理”类别

Dialogue Flow Patterns

对话流程模式

Pattern	When to Use	Example
Single-turn	Simple Q&A, no context needed	"What are your hours?" → respond immediately
Multi-turn (slot filling)	Need multiple pieces of info	"Book a table" → ask party size → ask date → ask time → confirm
Branching	Different paths based on user's answer	"Do you have an account?" → Yes: login flow / No: registration flow
Confirmation	Before executing actions	"I'll cancel order #12345. Is that correct?"
Handoff	Bot can't handle the request	"Let me connect you with a human agent"

模式	适用场景	示例
单轮对话	简单问答，无需上下文	“你们的营业时间是？” → 直接回复
多轮对话（槽位填充）	需要收集多条信息	“预订桌位” → 询问人数 → 询问日期 → 询问时间 → 确认
分支对话	根据用户回答进入不同路径	“您有账户吗？” → 是：登录流程 / 否：注册流程
确认环节	执行操作前需确认	“我将为您取消订单#12345，是否确认？”
人工转接	机器人无法处理请求时	“我将为您转接人工客服”

Response Design Principles

回复设计原则

Acknowledge first: "Got it, you want to check your order status."
Be concise: Answer the question, then stop. Don't add unnecessary information.
Offer next steps: "Is there anything else I can help with?" or suggest related actions.
Use quick replies/buttons: Reduce typing, guide the conversation.
Personality: Define a consistent tone (friendly, professional, casual) and stick to it.

先确认需求：“好的，我了解您想查询订单状态。”
简洁明了：回答问题后即停止，避免冗余信息。
提供后续选项：“请问还有其他需要帮助的吗？”或推荐相关操作。
使用快捷回复/按钮：减少用户输入，引导对话方向。
保持统一人设：定义一致的语气（友好、专业、随性）并严格遵循。

Metrics

指标

Metric	Definition	Target
Intent accuracy	% correctly classified intents	> 85%
Containment rate	% resolved without human handoff	> 60-70%
CSAT	Customer satisfaction score	> 4.0/5
Fallback rate	% triggering fallback/unknown intent	< 15%
Resolution time	Average time to resolve	< 2 minutes

指标	定义	目标值
意图识别准确率	正确分类的意图占比	> 85%
自主解决率	无需人工转接即可解决的请求占比	> 60-70%
客户满意度（CSAT）	用户满意度评分	> 4.0/5
兜底触发率	触发兜底/未知意图的请求占比	< 15%
问题解决时长	平均解决时间	< 2分钟

Output Format

输出格式

markdown

undefined

markdown

undefined

Chatbot Design: {Use Case}

聊天机器人设计：{使用场景}

Intent Catalog

意图目录

Intent	Description	Example Utterances	Priority
{intent}	{what it means}	"{example 1}", "{example 2}"	H/M/L

意图	描述	示例表述	优先级
{intent}	{意图含义}	"{示例1}", "{示例2}"	高/中/低

Dialogue Flows

对话流程

{Flow Name}

{流程名称}

User: {trigger utterance}
Bot: {response + slot question if needed}
User: {provides info}
Bot: {confirmation or action}

用户：{触发语句}
机器人：{回复 + 如需槽位则提问}
用户：{提供信息}
机器人：{确认或执行操作}

Fallback Strategy

兜底策略

After 1 miss: rephrase + suggest options
After 2 misses: offer human handoff

首次无法识别：重新表述问题 + 提供选项
两次无法识别：转接人工客服

Metrics Targets

指标目标

Metric	Target
Intent accuracy	> {X%}
Containment	> {X%}

undefined

指标	目标值
意图识别准确率	> {X%}
自主解决率	> {X%}

undefined

Gotchas

注意事项

Users don't follow your flow: People type in unexpected ways, change topics mid-conversation, and give incomplete information. Design for messiness, not just the happy path.
Fallback is your most important intent: A good fallback ("I'm not sure I understood. Did you mean X, Y, or Z?") is better than a bad guess.
LLM-powered bots still need guardrails: Using GPT/Claude for response generation? Add intent classification as a first layer to route and constrain, preventing hallucination and off-topic responses.
Test with real users, not team members: Your team knows how the bot works and phrases things "correctly." Real users don't. Test with 10+ real users before launch.
Conversation logs are gold: Review conversation logs weekly. Failed conversations reveal missing intents, confusing flows, and training data gaps.

用户不会严格遵循流程：用户的输入方式可能超出预期，会中途切换话题或提供不完整信息。设计时需考虑各种异常情况，而非仅针对理想流程。
兜底意图是重中之重：一个好的兜底回复（“我没太理解您的意思，您是指X、Y还是Z？”）比错误猜测更重要。
大语言模型驱动的机器人仍需约束：使用GPT/Claude生成回复时，需先添加意图分类层进行路由和约束，防止生成幻觉内容或偏离主题的回复。
请真实用户测试，而非内部团队：内部人员了解机器人的使用逻辑，会用“正确”的方式提问，但真实用户不会。上线前请至少10名真实用户参与测试。
对话日志是宝贵资源：每周复盘对话日志。失败的对话会暴露缺失的意图、易混淆的流程和训练数据的漏洞。

References

参考资料

For NLU training data best practices, see
```
references/nlu-training.md
```
For LINE/Messenger platform integration, see the ecom-conversational skill

关于NLU训练数据的最佳实践，请查看
```
references/nlu-training.md
```
关于LINE/Messenger平台集成，请参考电商对话技能文档