对话式界面

尚无可操作的示例；配图为编辑性说明，用于展示概念而非实际产品截图。在具体产品中落地前，仍需验证：系统的能力边界说明是否真实反映其能力、自由文本的歧义处理效果、生成内容的可核实机制，以及历史记录的可编辑与可重试性。

{
  "states": [
    "empty",
    "composing",
    "awaiting-response",
    "responded",
    "error"
  ],
  "a11y": [
    "消息流对屏幕阅读器是可读的活动区域，新消息会被播报",
    "建议的操作按钮和输入框都可用键盘单独聚焦和激活"
  ],
  "motion": "新消息出现不依赖动画即可被感知"
}

{
  "id": "conversational-interface",
  "type": "interaction",
  "name": {
    "zh": "对话式界面",
    "en": "Conversational Interface"
  },
  "aliases": [
    "聊天式界面",
    "对话界面",
    "Chatbot Interface"
  ],
  "granularity": "flow",
  "intents": [
    "state-intent",
    "discover-capabilities"
  ],
  "tags": [
    "对话",
    "聊天",
    "自然语言",
    "语音助手",
    "轮流对话"
  ],
  "definition": "用自然语言的轮流对话组织交互：用户陈述意图，而不是操作可见的控件；常见于聊天窗口与语音助手。",
  "boundary": "与渐进披露不在同一条轴线上，二者不是替代关系，渐进披露管理的是可见控件的呈现时机，对话式界面管理的是整个交互的组织方式；对话式的外观本身不代表系统具备理解或执行能力；自由文本输入让用户无法像看到菜单那样看到系统能做什么，这项可发现性成本需要在界面里显式弥补。",
  "signature": [
    "以轮次交替的消息流组织交互",
    "用户用自由文本或语音表达意图，而非选择固定控件",
    "系统的能力边界与局限在对话开始处或旁侧可见"
  ],
  "when": [
    "任务的意图容易用一两句话说清，但对应的操作路径因人而异，不适合固定成菜单",
    "语音或免手操作场景，用户无法或不便查看图形控件"
  ],
  "when_not": [
    "任务有明确、少量的固定选项，图形控件能一眼呈现且更快操作",
    "用户需要精确比较多个结构化数据点（如价格、参数表）",
    "系统尚未验证能可靠理解自由文本输入，却让对话成为唯一入口"
  ],
  "tradeoff": "省去了记忆菜单结构的负担，让表达更自由，但用户必须自己回忆或试探系统能做什么，且系统要处理自由文本中的歧义、拼写变化与跨轮次的上下文延续。",
  "comparison": {
    "focus": "交互以自然语言轮次组织，而非以可见控件的操作组织",
    "mechanism": "用户陈述意图，系统以对话轮次响应，历史可回看",
    "cost": "能力边界的可发现性、自由文本的歧义处理、生成内容的可信核实"
  },
  "sources": [
    {
      "id": "nng-chatbots",
      "title": "NN/g · Chatbots",
      "url": "https://www.nngroup.com/articles/chatbots/",
      "claim": "聊天机器人常因用户不把聊天当作独立的交互入口而发现性差，且自由文本输入难以处理拼写变化、歧义表达与跨轮次的上下文延续。",
      "checkedAt": "2026-09-15",
      "archive": {
        "lookupUrl": "https://web.archive.org/web/20260915*/https://www.nngroup.com/articles/chatbots/",
        "status": "available",
        "checkedAt": "2026-09-16T04:51:22.273Z",
        "jobId": "spn2-17ba5b3970d427f8e7fba298095436a101c6e7fc",
        "url": "https://web.archive.org/web/20260916045127/https://www.nngroup.com/articles/chatbots/",
        "timestamp": "20260916045127"
      }
    },
    {
      "id": "ms-hax-guidelines",
      "title": "Microsoft HAX Toolkit · Guidelines for Human-AI Interaction",
      "url": "https://www.microsoft.com/en-us/haxtoolkit/library/",
      "claim": "第 G1 条建议让用户清楚系统能做什么，第 G8 条建议支持高效忽略不想要的服务，第 G9 条建议在系统出错时支持高效编辑或纠正。",
      "checkedAt": "2026-09-15",
      "archive": {
        "lookupUrl": "https://web.archive.org/web/20260915*/https://www.microsoft.com/en-us/haxtoolkit/library/",
        "status": "available",
        "checkedAt": "2026-09-16T05:43:24.829Z",
        "jobId": "spn2-7cb34684fb905780ab4c35f6e41f0ec7680b9b7e",
        "url": "https://web.archive.org/web/20260916054347/https://www.microsoft.com/en-us/haxtoolkit/library/",
        "timestamp": "20260916054347"
      }
    },
    {
      "id": "nng-ai-chat-not-answer",
      "title": "NN/g · AI Chat Is Not (Always) the Answer",
      "url": "https://www.nngroup.com/articles/ai-chat-not-the-answer/",
      "claim": "采用对话式外观本身并不能替代对用户真实需求与系统实际能力的判断，团队应先确认对话是否真正解决问题。",
      "checkedAt": "2026-09-15",
      "archive": {
        "lookupUrl": "https://web.archive.org/web/20260915*/https://www.nngroup.com/articles/ai-chat-not-the-answer/",
        "status": "available",
        "checkedAt": "2026-09-16T05:43:44.651Z",
        "jobId": "spn2-9e367b19532dc846bcbffd77ac0675c638fa340b",
        "url": "https://web.archive.org/web/20260916054409/https://www.nngroup.com/articles/ai-chat-not-the-answer/",
        "timestamp": "20260916054409"
      }
    }
  ],
  "relations": [
    {
      "target": "recognition",
      "kind": "conflicts_when",
      "condition": "系统只提供空白输入框、不给出可点击的选项或范例",
      "reason": "自由文本要求用户凭记忆想起该怎么问，这与识别优于回忆的方向相反。",
      "basis": "editorial"
    },
    {
      "target": "progressive-disclosure",
      "kind": "related",
      "condition": "两者都用于管理信息或功能的呈现方式",
      "reason": "渐进披露分阶段呈现可见控件，对话式界面用自然语言轮次组织整个交互，处在不同的呈现轴上，不是彼此的替代方案。",
      "basis": "editorial"
    },
    {
      "target": "modal-dialog",
      "kind": "alternative",
      "condition": "需要用户完成一个边界清晰、选项固定的独立任务",
      "reason": "模态对话框用可见控件呈现固定选项，比让用户在自由文本里描述同一个任务更直接、更少歧义。",
      "basis": "editorial"
    }
  ],
  "rules": [
    {
      "when": "用户第一次进入对话式界面",
      "instruction": "让系统能做什么、不能做什么在对话开始处或附近可见，不要只留一个空白输入框让用户猜。",
      "strength": "must",
      "basis": "editorial"
    },
    {
      "when": "用户已经进行了多轮对话",
      "instruction": "保留对话历史，并允许用户编辑或重新发送某一轮，而不必重来整段对话。",
      "strength": "recommend",
      "basis": "editorial"
    },
    {
      "when": "对话中出现由系统生成的信息",
      "instruction": "不要把生成内容当作已核实的事实直接呈现，要给用户可以核实或追溯来源的途径。",
      "strength": "must",
      "basis": "editorial"
    }
  ],
  "behavior": {
    "states": [
      "empty",
      "composing",
      "awaiting-response",
      "responded",
      "error"
    ],
    "a11y": [
      "消息流对屏幕阅读器是可读的活动区域，新消息会被播报",
      "建议的操作按钮和输入框都可用键盘单独聚焦和激活"
    ],
    "motion": "新消息出现不依赖动画即可被感知"
  },
  "demo": null,
  "version": "0.1.0",
  "editorialStatus": "drafted",
  "requirements": [],
  "demoCapabilities": [],
  "implementationNote": "尚无可操作的示例；配图为编辑性说明，用于展示概念而非实际产品截图。在具体产品中落地前，仍需验证：系统的能力边界说明是否真实反映其能力、自由文本的歧义处理效果、生成内容的可核实机制，以及历史记录的可编辑与可重试性。",
  "image": {
    "src": "/images/conversational-interface.webp",
    "alt": "虚构的 FIELDNOTES 聊天面板，交替排列的对话气泡以抽象色块表示消息内容，输入框上方是一排展示示例能力的建议按钮，底部有一行说明系统能力范围的小字。"
  },
  "searchTerms": [
    "chat",
    "chatbot",
    "natural language",
    "voice assistant",
    "turn-taking"
  ]
}