Dialogflow에서 작업과 호출 문구는 암시적 호출을 위해 지정하는 인텐트의 학습 문구로 정의됩니다. Dialogflow 통합 가이드의 설명대로 Dialogflow 콘솔에서 단일 단계 프로세스를 사용하여 작업에 딥 링크 및 암시적 호출에 대한 지원을 추가할 수 있습니다.
Actions SDK
Actions SDK에서는 작업 및 호출 구문이 쿼리 패턴으로 정의됩니다. 자세한 내용은 작업 정의를 참고하세요.
암시적 호출 분석 및 보기
이 섹션에서는 Actions 콘솔을 사용하여 작업과 일치하는 암시적 호출에 관한 정보를 분석하고 확인하는 방법을 설명합니다.
검색 분석
작업 콘솔의 애널리틱스 > 검색 섹션에서는 어시스턴트가 작업을 추천하도록 한 구문과 프롬프트에 일치된 인텐트에 관한 유용한 정보를 제공합니다.
호출 문구는 사용자에게 유용한 호출 및 검색 도구를 제공하지만 신중하게 선택해야 합니다. 작업의 호출 문구를 설계할 때는 다음 권장사항에 유의하세요.
호출 문구를 작업의 사용 사례에 맞게 유지합니다.
작업은 특정 용도를 제공해야 하므로 호출 구문이 작업의 기능을 실제로 설명해야 합니다. 예를 들어 작업이 항공편을 예약하는 데 사용되는 경우 '$location에서 $location까지 어떻게 가나요?'를 사용하지 마세요. 더 나은 대안은 '$location발 $location행 항공편 예약'입니다.
호출 구문에 항목이 포함된 경우 해당 항목의 모든 동의어가 해당 구문 및 작업의 전반적인 사용 사례와 관련이 있는지 확인해야 합니다.
동사와 목적어가 모두 포함된 구문 작성하기
호출은 완전한 문장으로 구성됩니다. 즉, 작업의 호출 구문이 언어적으로 자연스럽기 위해서는 동사-목적어 쌍이어야 합니다. 예를 들어 'Easy recipe.'는 'Ok Google, Easy recipe.'가 되므로 좋은 호출 구문이 아닙니다. 더 나은 대안은 '간단한 레시피를 만드는 방법'이 될 수 있으며 'Hey Google, 쉬운 레시피 어떻게 만들어 줘'로 호출할 수 있습니다.
올바르지 않은 호출 문구와 좋은 호출 문구의 예는 다음과 같습니다.
잘못된 호출 구문
구문에 동사가 부족합니다. "다섯 번째"
구문이 너무 일반적입니다. "뉴욕 여행"
좋은 호출 구문
문구에 명확한 동사와 목적어가 포함되어 있습니다. "아빠 농담 들려줘."
특정 요청을 제공하는 구문:
"오늘 뭘 입어야 할까?"
"퀴즈 게임을 하고 싶어."
"여자아이 이름을 짓고 싶은데."
이러한 호출 문구는 작업마다 고유하지 않을 수 있습니다. 사용자에게 추천할 작업을 결정하는 것은 어시스턴트의 몫입니다.
[[["이해하기 쉬움","easyToUnderstand","thumb-up"],["문제가 해결됨","solvedMyProblem","thumb-up"],["기타","otherUp","thumb-up"]],[["필요한 정보가 없음","missingTheInformationINeed","thumb-down"],["너무 복잡함/단계 수가 너무 많음","tooComplicatedTooManySteps","thumb-down"],["오래됨","outOfDate","thumb-down"],["번역 문제","translationIssue","thumb-down"],["샘플/코드 문제","samplesCodeIssue","thumb-down"],["기타","otherDown","thumb-down"]],["최종 업데이트: 2025-07-28(UTC)"],[[["\u003cp\u003eImplicit invocation allows users to discover and interact with your Action without explicitly invoking it by name, based on their requests and context.\u003c/p\u003e\n"],["\u003cp\u003eGoogle Assistant uses signals like user requests and context to match the user's need with suitable Actions, including yours.\u003c/p\u003e\n"],["\u003cp\u003eTo optimize for implicit invocation, ensure your Action's invocation phrases are specific to its purpose, include verbs and objects, and reflect natural language patterns.\u003c/p\u003e\n"],["\u003cp\u003eYou can analyze implicit invocation performance and manage invocation phrases within the Actions console.\u003c/p\u003e\n"],["\u003cp\u003eWhen using entities in Dialogflow training phrases, adhere to specific syntax rules for accurate entity recognition.\u003c/p\u003e\n"]]],["Implicit invocation allows users to discover Actions without naming them. The Google Assistant matches user requests to relevant Actions based on user context and configured intents. To enable this, define invocation phrases in Dialogflow training phrases or Actions SDK query patterns. Actions are recommended based on a recommendation algorithm. Analyzing \"Discovery\" in the Actions console reveals matched phrases and intents. Ensure phrases are specific, use verb-object pairs, and, in Dialogflow, use space separation for entities. Google can not guarantee the recommendation of an Action.\n"],null,["# Implicit invocation (Dialogflow)\n\nImplicit invocation is an interaction flow that occurs when a user makes a\nrequest to perform some task without invoking an Action by name. The\nGoogle Assistant attempts to match the user's request to a suitable fulfillment,\nsuch as an Action, search result, or mobile app, then presents recommendations\nto the user.\n\nIf you have deployed an Action that can help the user with their task,\nimplicit invocation provides a way for users to discover your Action\nvia the Assistant.\n| **Note:** If your Action is a game, you must register for the `actions.intent.PLAY_GAME` built-in intent in order to be eligible for implicit invocation. To learn how to implement the \"Play game\" intent, see the [built-in intents guide](/assistant/df-asdk/discovery/built-in-intents).\n\nHow it works\n------------\n\nTo find matching Actions, Google uses signals such as users telling\nthe Assistant to do something that's similar to an invocation phrase\nfor one of your configured intents, or when the user is in a context where your\nAction would be appropriate.\n\nThis interaction occurs as follows:\n\n1. A user asks the Assistant to perform a task.\n2. The recommendation algorithm determines that your Action can complete the user's task.\n3. The Assistant recommends your Action to the user.\n\nWhen the user selects your Action following implicit invocation,\nthe Assistant notifies your Action of the specific task that the\nuser requested.\n\nThe following example shows how a user can trigger an implicit invocation by\nrequesting a chicken soup recipe, without specifying an Action's invocation\nname.\n\n|---------------|--------------------------------------------------------------------------------------------------------------|\n| **User** | Ok Google, I need a chicken soup recipe. |\n| | **\\\u003cThe Assistant queries the Actions on Google platform to find suitable Actions for the user's request\\\u003e** |\n| **Assistant** | Sure. For that, you might like Personal Chef. Wanna give it a try? |\n\nIf your Action's invocation phrases can't be mapped to a variety of related\nuser queries, it won't be recognized as relevant. For more details, see\n[Best practices for writing useful invocation phrases](#best_practices_for_writing_useful_invocation_phrases).\n\nDue to the evolving nature of the recommendation algorithm, Google cannot\nguarantee that your Action will be recommended via implicit invocation.\n| **Note:** Implicit invocation behaves differently than a deep link. To use a deep link, users need to specify the Action explicitly by its name, along with a qualifying phrase that represents a task for your Action to handle (for example, \"*Ok Google, talk to Personal Chef to look up chicken soup recipes.*\").\n\nAdd invocation phrases\n----------------------\n\nThis section describes how you add invocation phrases for implicit invocation\nif you are using Dialogflow or Actions SDK to build your Action.\n\n### Dialogflow\n\nIf you are using entities in your training phrases, be sure to familiarize\nyourself with [requirements for phrases with entities](#syntax_Requirements).\n\nIn Dialogflow, your Actions and their invocation phrases are defined as\ntraining phrases in intents that you specify for implicit invocation. You can\nadd support for deep links and implicit invocations to your Action using a\nsingle-step process in the [Dialogflow console](//console.dialogflow.com), as described in the\n[Dialogflow integration guide](//dialogflow.com/docs/integrations/actions/build#specify_actions).\n| **Note:** The maximum number of implicit invocation intents should not exceed 10. The maximum number of training phrases in each invocation intent should not exceed 10.\n\n### Actions SDK\n\nIn the Actions SDK, your Actions and their invocation phrases are defined as\nquery patterns. See [Define Actions](/assistant/df-asdk/actions-sdk/define-actions) for more\ninformation.\n\nAnalyze and view implicit invocations\n-------------------------------------\n\nThis section describes how to use the [Actions console](//console.actions.google.com/) to analyze and view\ninformation about implicit invocations that matched to your Action.\n\n### Discovery analytics\n\nThe **Analytics \\\u003e Discovery** section of the [Actions console](//console.actions.google.com) provides useful\ninformation about what phrases prompted the Assistant to recommend your\nAction and which of your intents was matched to the prompt.\n\nYou can find more information on our documentation about [console analytics and\nhealth](/assistant/console/analytics#discovery).\n\n### View invocation phrases\n\nIn the [Actions console](//console.actions.google.com), you can view your Action and its invocation\nphrases under **Build \\\u003e Actions**.\n| **Note:** If you're using Dialogflow and any of your Actions are not listed on this page, those missing Actions are most likely not set as additional triggering intents. Click the **Add Actions from Dialogflow** button for instructions that explain how to set your intents correctly in Dialogflow.\n\nYou can click on any of the Actions listed to view their invocation phrases.\nIf you're using Dialogflow, you have the option to **Add more phrases**,\nwhich links you directly to your Action's intents in Dialogflow.\n\nRequirements for phrases with entities\n--------------------------------------\n\nIf you are using Dialogflow, the following rules apply to [entities](//dialogflow.com/docs/entities).\nThese syntax rules also apply to languages (for example, Korean and Thai) that\nin practice do not use space separation between words.\n\n- If you place an entity in a phrase, it must have a space before and after.\n- You do not need to add space before the entity if it is placed at the *beginning* of a phrase.\n- You do not need to add space after the entity if it is placed at the *end* of a phrase.\n\nIf an entity is not space-separated from text, Actions on Google will read the\nentire phrase as a string with no value substitution for the entity.\n\nYou should test phrases that contain entities and text in the Actions simulator\nto ensure that the intended meaning is retained.\n\nTo learn more, see the Dialogflow [entities overview](//dialogflow.com/docs/entities).\n\nBest practices for writing useful invocation phrases\n----------------------------------------------------\n\nInvocation phrases offer a helpful invocation and discovery tool\nfor users, but you must choose them carefully. Keep the following best practices\nin mind when designing your Action's invocation phrases:\n\n- **Keep your invocation phrases specific to your Action's use cases**\n\n Your Action should [serve a specific purpose](/assistant/df-asdk/design), so make sure\n your invocation phrases actually describe what your Action does. For\n example, if your Action is used to book flights, don't use \"*How do I get\n from $location to $location?* \". A better alternative is \"*Book a flight from\n $location to $location.*\".\n - If your invocation phrases contain entities, make sure all synonyms of those entities are relevant to its phrase(s) and your Action's use case overall.\n- **Build phrases that include both a *verb* and an *object***\n\n An invocation consists of a complete sentence, which means your Action's\n invocation phrase needs to be a verb-object pair in order to be\n linguistically natural. For example, \"*Easy recipe.* \" wouldn't be a good\n invocation phrase because the full invocation would be \"*Ok Google, Easy\n recipe.* \". A better alternative could be \"*How do I make an easy recipe?* \"\n which can be invoked by \"*Ok Google, how do I make an easy recipe*\".\n\nConsider the following examples of bad and good invocation phrases:\n\n**Bad invocation phrases**\n\n- Phrase lacks a verb: *\"The number five.\"*\n- Phrase is too general: *\"Travel to New York.\"*\n\n**Good invocation phrases**\n\n- Phrase has a clear verb and object: *\"Hear a dad joke.\"*\n- Phrase provides specific requests:\n\n - *\"What should I wear today?\"*\n - *\"I want to play a trivia game.\"*\n - *\"I want some baby names for girls.\"*\n\nThese invocation phrases may not be unique to your Action; it's up to\nthe Assistant to determine which Actions to suggest to users."]]
