Entering questions

Available in Classic and VPC

Once the conversation information is entered, a field for registering questions will be enabled. Enter questions you expect the users to ask in the question field. The conversation model learns the entered questions, so that it can understand similar questions in addition to the questions already entered.
The more questions you have, better the result is. Registering 10 or more general questions with similar intent is recommended. It is more helpful to have different utterance styles entered rather than just increasing the number of questions by slightly tweaking the previous sentence's endings. For example, let's say a user asks about a specific product's usage fee. Consider how the users will ask their questions, and enter various types of utterance as below.

• How much is 00
• How much will 00 cost me
• Is 00 expensive
• How much does 00 cost
• Show me a pricing chart
• Can you provide a quote for 00?
• What's the employee discount price for 00?

Register question

The following describes how to add a question.

1. Enter the conversation information by referring to Register conversation.
2. Select a question type.
   • General question: Questions entered in natural language
   • Regular expression question: Questions using regular expression syntax. For how to enter regular expression questions, refer to Regular expression input methods.
   • Intent: Enter the intent name to be sent from the intent classifier linked to the chatbot. When a user's expected utterance is entered and learned, the chatbot doesn't directly classify the intention. Instead, it receives the classified intention (intent) from the linked intention classifier and uses this information to analyze the conversation.
   • Call event: This feature allows you to proceed with the conversation using a call event as a trigger. The conversations are matched using the event sent when speech recognition starts in the AiCall service as a trigger. The menu is only enabled when the domain is an AiCall domain. For more information, refer to Call event.
3. Enter a question, and click the [Add] button.
4. Enter at least 10 questions.
   • Question marks and exclamation points can be omitted since they don't affect the conversation model learning.
   • Enter various conversations rather than entering multiple endings of the same conversation.

Add batch questions

You can add a large number of questions in an Excel file and upload it. It's convenient to use the upload feature as you can register many questions at once rather than entering them one by one. The file must be created in line with the specified template if you're uploading the questions in an Excel file, and if the file doesn't follow the template, then the upload may fail.

The following describes how to add batch questions.

1. From the NAVER Cloud Platform console, click the Services > CLOVA Chatbot > Domain menus, in that order.
2. Click the [Run builder] button of the domain to create a conversation in to run the chatbot builder.
3. From the chatbot builder, click the Conversation list > [Upload] button.
4. Once the "Upload conversation" window appears, click the [Download template] button by upload type.
5. Enter the question type (select from "general question" or "regular expression question") in column A in the downloaded Excel template file, and the question's content in column B, and then save the file.
6. Select the upload type from the chatbot builder's "Upload conversation" window, and upload the Excel file with the questions filled in.
7. Click the [Start upload] button.

Move question

You can move the uploaded questions to another conversation.
The following describes how to move a question.

1. Select the question to move from the list of questions, and then click the [Move] button.
   

2. When the Move question window appears, select the conversation to move the question to, and then click the [Move] button.

Call event

This is a feature that allows you to proceed with the conversation using a call event as a trigger. The conversations are matched using the event sent when speech recognition starts in the AiCall service as a trigger. This feature is only supported for AiCall domains. Only Speech recognition resumption event is supported among the call events at the moment.

How it works

The following describes how a call event works.

• When a Speech recognition resumption event occurs as a call event, the chatbot will search for a conversation where the Speech recognition resumption event is registered among the conversations in the chatbot builder.
• If there's an answer which fits the context or conditions, then the answer from the conversation will be sent to the user as the response. If such an answer couldn't be found, then the call event won't be processed.
• The Speech recognition resumption event will occur every time a speech is recognized in the AiCall service, so the event won't be processed if the answer to respond doesn't exist.
• Please note that it works differently to a case of responding with a failure message when there's no answer that fits the context or conditions for the questions whose type is general question/regular expression question/intent.

Example

Call event can be used when speech recognition is started again and the chatbot sends an answer which encourages the user to utter first. For instance, if getting information entered through the app page rather than speech is appropriate while going through an AiCall scenario, or if such a scenario must be performed for security reasons, then the scenario can be designed as follows.

Let's say you want to take the user's information through an app or on the web for a change of delivery address for a parcel, rather than through speech recognition.

1. Create a conversation about changing the parcel delivery location.
   • Insert action method $2{app page to change parcel delivery location} in the answer.
   • Add "app page to enter address" for the output context.
   • Set the maintenance time to "30 minutes." This is to maintain the context, even if the user stays on the app or web page for a long time.
     
2. Perform the following tasks in the API of the $2{app page to change parcel delivery location} action method.
   • Process the action of opening the app or web page to the user's mobile phone.
   • Request to stop speech recognition while the user enters the information to the AiCall server.
3. Once the user completes entering the information on the app or web page, perform the following tasks.
   • Process the information entered in the internal server.
   • Request the AiCall server to resume speech recognition to go back to the call.
4. Once the input is finished on the app or web, register the conversation in which the rest of the scenario will go on in the call environment.
   • Register the Speech recognition resumption event as the trigger.
   • Add "app page to enter address" to the input context, so that the context is continued from the "changing the parcel delivery location" conversation.
   • Enter an answer appropriate for the scenario, such as "The address input has been completed." as the answer.
     

Request to stop recognizing speech

Refer to the following to send a request to stop speech recognition to the AiCall server.

Request form

• Http Method : PUT
• Path : /recognition/{sessionId}/control/pause
• parameter

Request example

curl --location --request PUT 'https://ac86e44bd82542c39b6bfbc2fb6f488a.test-apigw.ntruss.com/write/v1/recogni- tion/8a9c75f8-1ed1-4925-8bed-7d741baf3abe/control/pause?contactCenterId=1' \
--header 'X-CLOVA-AICALL-API-KEY: 90d97f24-6f0b-4d9a-bbf9-ed391b44c6ea' \
--header 'X-CLOVA-AICALL-API-VERSION: 2' \
--header 'X-CLOVA-AICALL-TRACE-ID: caab126fedf14575b66b554f21b84a57'

Response form

• Http Status : 204
• There is no response body.

Request to start recognizing speech

Refer to the following to send a request to restart speech recognition to the AiCall server.

Request form

• Http Method : PUT
• Path : /recognition/{sessionId}/control/resume
• parameter

Request example

curl --location --request PUT 'https://ac86e44bd82542c39b6bfbc2fb6f488a.test-apigw.ntruss.com/write/v1/recogni- tion/8a9c75f8-1ed1-4925-8bed-7d741baf3abe/control/resume?contactCenterId=1' \
--header 'X-CLOVA-AICALL-API-KEY: 90d97f24-6f0b-4d9a-bbf9-ed391b44c6ea' \
--header 'X-CLOVA-AICALL-API-VERSION: 2' \
--header 'X-CLOVA-AICALL-TRACE-ID: caab126fedf14575b66b554f21b84a57'

Response form

• Http Status : 204
• There is no response body.