Question

我正在使用DialogFlow + Firebase修补Google上的操作。我们的想法是为默认AoG所不具备的物联网设备进行自定义构建。

就广播和电视而言，目前有两种意图：

1）频道：它接收3个参数：device_name（自定义实体），device_action（自定义实体）和值。

例如：请将无线电频道更改为22.3
或者将电视频道改为22

2）卷：它包含3个参数：device_name（自定义实体），device_action（自定义实体）和值。

例如：请将收音机音量更改为99 或者将电视的音量改为22

问题是代理似乎无法区分两者。

执行错误了吗？

//编辑2018年5月9日：

index.js

＆＃13;

// Copyright 2016, Google, Inc.
// Licensed under the Apache License, Version 2.0 (the 'License');
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//    http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an 'AS IS' BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

'use strict';

const util = require('util');
const functions = require('firebase-functions');
const {
  dialogflow,
  Suggestions,
  BasicCard,
  Button,
  SimpleResponse,
} = require('actions-on-google');

const {values, concat, random, randomPop} = require('./util');
const responses = require('./responses');

/** Dialogflow Contexts {@link https://dialogflow.com/docs/contexts} */
const AppContexts = {
  FACT: 'choose_fact-followup',
  CATS: 'choose_cats-followup',
  XXX: 'choose_xxx-followup',

};

/** Dialogflow Context Lifespans {@link https://dialogflow.com/docs/contexts#lifespan} */
const Lifespans = {
  DEFAULT: 5,
};

const app = dialogflow({
  debug: true,
  init: () => ({
    data: {
      // Convert array of facts to map
      facts: responses.categories.reduce((o, c) => {
        o[c.category] = c.facts.slice();
        return o;
      }, {}),
      cats: responses.cats.facts.slice(), // copy cat facts
    },
  }),
});

/**
 * Greet the user and direct them to next turn
 * @param {DialogflowConversation} conv DialogflowConversation instance
 * @return {void}
 */
app.intent('Unrecognized Deep Link Fallback', (conv) => {
  const response = util.format(responses.general.unhandled, conv.query);
  const suggestions = responses.categories.map((c) => c.suggestion);
  conv.ask(response, new Suggestions(suggestions));
});

// redirect to the intent handler for tell_fact
app.intent('choose_fact', 'tell_fact');

// Say a fact
app.intent('tell_fact', (conv, {category}) => {
  const {facts, cats, xxx} = conv.data;
  if (values(facts).every((c) => !c.length)) {
    // If every fact category facts stored in conv.data is empty,
    // close the conversation
    return conv.close(responses.general.heardItAll);
  }
  const categoryResponse =
    responses.categories.find((c) => c.category === category);
  const fact = randomPop(facts[categoryResponse.category]);
  if (!fact) {
    const otherCategory =
      responses.categories.find((other) => other !== categoryResponse);
    const redirect = otherCategory.category;
    const parameters = {
      category: redirect,
    };
    // Add facts context to outgoing context list
    conv.contexts.set(AppContexts.FACT, Lifespans.DEFAULT, parameters);
    const response = [
      util.format(responses.transitions.content.heardItAll, category, redirect),
    ];
    // If cat facts not loaded or there still are cat facts left
    if (cats.length) {
      response.push(responses.transitions.content.alsoCats);
    }
    response.push(responses.general.wantWhat);
    conv.ask(concat(...response));
    conv.ask(new Suggestions(otherCategory.suggestion));
    if (cats.length) {
      conv.ask(new Suggestions(responses.cats.suggestion));
    }
    return;
  }
  const {factPrefix} = categoryResponse;
  // conv.ask can be called multiple times to have the library construct
  // a single response itself the response will get sent at the end of
  // the function or if the function returns a promise, after the promise
  // is resolved.
  conv.ask(new SimpleResponse({
    speech: concat(factPrefix, fact),
    text: factPrefix,
  }));
  conv.ask(responses.general.nextFact);
  conv.ask(new BasicCard({
    title: fact,
    image: random(responses.content.images),
    buttons: new Button({
      title: responses.general.linkOut,
      url: responses.content.link,
    }),
  }));
  console.log('hiwwxxxxxww thi is aaron');
  conv.ask(responses.general.suggestions.confirmation);
});

// Redirect to the intent handler for tell_cat_fact
app.intent('choose_cats', 'tell_cat_fact');

// Say a cat fact
app.intent('tell_cat_fact', (conv) => {
  const {cats} = conv.data;
  console.log('this is cats data' + {cats});
  const fact = randomPop(cats);
  if (!fact) {
    conv.contexts.delete(AppContexts.FACT);
    conv.contexts.delete(AppContexts.CATS);
    conv.ask(responses.transitions.cats.heardItAll);
    return conv.ask(responses.general.suggestions.confirmation);
  }
  const {factPrefix, audio} = responses.cats;
  // conv.ask can be called multiple times to have the library construct
  // a single response itself. The response will get sent at the end of
  // the function or if the function returns a promise, after the promise
  // is resolved.
  const sound = util.format(audio, random(responses.cats.sounds));
  conv.ask(new SimpleResponse({
    // <speak></speak> is needed here since factPrefix is a SSML string
    // and contains audio.
    speech: `<speak>${concat(factPrefix, sound, fact)}</speak>`,
    text: factPrefix,
  }));
  conv.ask(responses.general.nextFact);
  conv.ask(new BasicCard({
    title: fact,
    image: random(responses.cats.images),
    buttons: new Button({
      title: responses.general.linkOut,
      url: responses.cats.link,
    }),
  }));
  console.log('hiwwxxxxxww thi is aaron');
  conv.ask(responses.general.suggestions.confirmation);
});



//say a tv channel
app.intent('volume', (conv, {device_name,device_action, value}) => {

  var no_device_name = device_name;
  var no_value = value;
  var no_device_action = device_action;

  var this_device_value = util.inspect(value, false, null);
  var this_device_name = util.inspect(device_name, false, null);
  var this_device_action = util.inspect(device_action, false, null);
  console.log(this_device_action[0]);

  if (no_device_name[0] == 'channel'){
    console.log('inside tv but CHANNEL');



  }

  else{
    console.log('VOLUME');
  }
  console.log('THIS IS VOL');
  console.log(no_device_action[0]);
   conv.ask(`Alright, ${device_name} VOLUM is now ${value}! `);
  console.log('inside volume ' + value[0]);
  console.log('inside volume' + device_name[0]);
  console.log('inside volume' + device_action[0]);
  console.log('hiwwxxxxxww thi is aaron');


});

//say a tv channel
app.intent('channel', (conv, {device_channel_name, device_channel_action, channel_value}) => {


  console.log('THIS IS CHANNEL');



  conv.ask(`Alright, ${device_channel_name} CHANNEL is now ${channel_value}! `);
  console.log('inside CHANNEL ' + channel_value[0]);
  console.log('inside CHANNEL' + device_channel_name[0]);
  console.log('inside CHANNEL' + device_channel_action[0]);
  console.log('hiwwxxxxxww thi is aaron');


});



app.intent('no_input', (conv) => {
  const repromptCount = parseInt(conv.arguments.get('REPROMPT_COUNT'));
  if (repromptCount === 0) {
    conv.ask(`What was that?`);
  } else if (repromptCount === 1) {
    conv.ask(`Sorry I didn't catch that. Could you repeat yourself?`);
  } else if (conv.arguments.get('IS_FINAL_REPROMPT')) {
    conv.close(`Okay let's try this again later.`);
  }
});



// The entry point to handle a http request
exports.factsAboutGoogle = functions.https.onRequest(app);

＆＃13;

Answer 1

如果您有一个示例短语，其中该短语的一部分设置为表示参数，则该短语的该部分可以采用任何值，该参数被定义为。＆＃34;已解决的价值＆＃34;列只会在您的示例短语中显示已解决的内容。

所以在你的＆＃34;频道＆＃34;意图，短语＆＃34;电视频道到3＆＃34;也可以匹配＆＃34;无线电音量到4＆＃34;自从＆＃34; tv＆＃34;和＆＃34;无线电＆＃34;是device_name实体和＆＃34;频道＆＃34;和＆＃34;音量＆＃34;都是device_action实体。

您有几个解决方案：

您可以将它们变成一个接受所有短语的Intent，并在您的webhook中检查device_action的值，看看您应该做什么。
您可以将它们保持为单独的Intent并完全删除device_action参数。你不需要它们。只需使用各种示例短语中的同义词，学习系统就哪些同义词与哪个Intent一起工作进行训练。
如果您仍然担心同义词，或者它仍然有意义，请将它们设为不同的实体类型。

例如，您可能希望拥有＆＃34;频道＆＃34;和＆＃34;预设＆＃34;作为同一实体类型下的不同实体。您关心的是差异，因为它会改变您对待号码的方式，但您仍在更改频道。＆＃34;卷＆＃34;将是一个不同的实体，因为它代表了完全不同的东西。

在您的评论中，您询问了设备的个性化别名。你没有询问不止一种具有不同名称的设备类型（所以你想要处理＆＃34; tv1＆＃34;和＃34; tv2＆＃34;使用相同的Intent，但是能够检测到使用上面的选项（3）可以最好地处理这两种方法，但方式略有不同。

如果你想要多个＆＃34; tv＆＃34;，每个都是相同的实体类型（例如，＆＃34;设备电视＆＃34;），但每一个都是不同的实体价值。它可能看起来像这样：

＆＃34;但等等！＆＃34;我听到你说＆＃34;如果用户有三台电视怎么办？还是九点？我必须设置所有这些吗？如果他们想把一台卧室电视叫做“卧室电视”怎么办呢？另一个是电视＆＃39;？＆＃34;

这表明你会将电视及其别名的数量保存在某种数据库中（无论如何你都需要这样做 - 你可能需要将他们的电视映射到一些独特的设备ID以实际打开它们），以及当用户与您的座席进行通话时，请使用User Entity进行更新。

用户实体具有仅为该用户设置的用户特定名称和别名。当用户第一次与您的代理商交谈时，您将使用Dialogflow API设置它们。如果您使用的是V1，则会使用/userEntities端点。如果您使用的是V2，则会使用projects.agent.sessions.entityTypes资源。

谷歌上的行动：类似的意图？

1 个答案: