我为我的Alexa琐事游戏创建了自定义插槽类型。琐事问题的答案可以是任何东西：单词，数字，短语，名称。我将所有正确答案插入了“广告位价值”。但问题是，技能无法识别任何错误答案。响应意图只是不会被触发。而是说立即提示文本。有没有可能教它理解任何单词作为回应？ 我的互动模型是：
"intents": [
{
"name": "AMAZON.CancelIntent",
"samples": []
},
{
"name": "AMAZON.HelpIntent",
"samples": []
},
{
"name": "AMAZON.StopIntent",
"samples": []
},
{
"name": "AnswerIntent",
"slots": [
{
"name": "one",
"type": "RoundAnswer"
},
{
"name": "two",
"type": "RoundAnswer"
}
],
"samples": [
"The answers are First {one} Second {two}",
"my answers are First {one} Second {two}",
"first answer is {one} Second answer is {two} ",
"{one} and {two}",
"{one} {two} ",
"{one}",
"The answer is {one} "
]
},
{
"name": "DontKnowIntent",
"slots": [],
"samples": [
"i don't know",
"don't know",
"i don't know that one",
"dunno",
"skip",
"i don't know that",
"who knows",
"i don't know this question"
]
},
{
"name": "AMAZON.StartOverIntent",
"samples": [
"let's start from the beginning",
"start from the beginning",
"start game",
"new game",
"start",
"start new game"
]
},
{
"name": "AMAZON.RepeatIntent",
"samples": []
},
{
"name": "AMAZON.YesIntent",
"samples": []
},
{
"name": "AMAZON.NoIntent",
"samples": []
},
{
"name": "AMAZON.NavigateHomeIntent",
"samples": []
}
],
"types": [
{
"name": "RoundAnswer",
"values": [
{
"name": {
"value": "thirty six"
}
},
{
"name": {
"value": "Could not swim"
}
},
{
"name": {
"value": "Dentist"
}
},
{
"name": {
"value": "Black mud"
}
},
{
"name": {
"value": "Fireball"
}
},
{
"name": {
"value": "Comet"
}
},
{
"name": {
"value": "13,000"
}
}
...many more here
]
}
]
},
"dialog": {
"intents": [
{
"name": "AnswerIntent",
"confirmationRequired": false,
"prompts": {},
"slots": [
{
"name": "one",
"type": "RoundAnswer",
"confirmationRequired": false,
"elicitationRequired": true,
"prompts": {
"elicitation": "Elicit.Slot.1498167634260.856164267644"
}
},
{
"name": "two",
"type": "RoundAnswer",
"confirmationRequired": false,
"elicitationRequired": false,
"prompts": {}
}
]
}
],
"delegationStrategy": "SKILL_RESPONSE"
},
"prompts": [
{
"id": "Confirm.Slot.1498167634260.856164267644",
"variations": [
{
"type": "PlainText",
"value": "ok"
}
]
},
{
"id": "Confirm.Slot.1498167634260.534517403250",
"variations": [
{
"type": "PlainText",
"value": "ok"
}
]
},
{
"id": "Confirm.Slot.1498167634260.145008767498",
"variations": [
{
"type": "PlainText",
"value": "ok"
}
]
},
{
"id": "Confirm.Slot.1498167634260.1214922796775",
"variations": [
{
"type": "PlainText",
"value": "ok"
}
]
},
{
"id": "Elicit.Slot.1498167634260.856164267644",
"variations": [
{
"type": "PlainText",
"value": "what's the answer?"
}
]
},
{
"id": "Confirm.Slot.1498167634260.1519619016259",
"variations": [
{
"type": "PlainText",
"value": "ok"
}
]
},
{
"id": "Confirm.Slot.1498167634260.1412749795564",
"variations": [
{
"type": "PlainText",
"value": "ok"
}
]
},
{
"id": "Confirm.Slot.1498167634260.1335829309782",
"variations": [
{
"type": "PlainText",
"value": "ok"
}
]
},
{
"id": "Confirm.Intent.654381803412",
"variations": [
{
"type": "PlainText",
"value": "are your answers final?"
}
]
}
]
}
答案 0 :(得分：0)
仅当发声中包含的广告位与交互模型中硬编码的广告位值匹配时，Alexa才调用正确的意图。
我对您没有什么建议
与其直接猜测答案，不如让用户说出正确的选项 例如： Alexa：印度的首都是什么？选项a）德里选项b）海得拉巴 用户：选项a
是将选项a和选项b注册为交互模型中的广告位值并进行匹配 答案的这些选项取决于您如何设计代码/ json或db（我建议 答案应该在键值对行{a：Delhi，b：Hyderabad，答案：a}）
如果您注册了所有正确和错误的答案，即问题的所有选项 在您的互动模型中，那么alexa会识别广告位值并调用 正确的意图。但是有时用户可能会说MJ而不是Michael Jackson，所以您有 也添加同义词，最好选择第一个点