CLOVA Chatbot glossary

Available in Classic and VPC

The key terms required to use CLOVA Chatbot are as follows.

API Gateway

A service provided by NAVER Cloud Platform that enables you to manage tasks related to API calls.

Common message

A chatbot message that is configured separately from conversation datasets and used globally. These messages include welcome messages, failure messages, no-response messages, fixed menus, feedback, similar answers, and push messages.

Conversation inputter

Can use Conversation Management, Conversation Option Management, and the Test interface. Can access the chatbot domain through the invitation URL sent by the domain administrator.

Domain

The basic unit that constitutes a chatbot service. One domain must be created for each chatbot, and conversation creation, channel integration, and analytics information are managed by domain.

Domain group

A feature that groups 2 or more chatbot domains to provide an answer from the domain best suited to the user's question. If multiple services are provided, you can configure a domain for each service and then connect the domains through a domain group.

For example, if a company provides 3 services, it is recommended to configure a separate domain for each service and group them through a domain group, rather than providing all 3 services within a single domain.

Domain administrator

Can create and delete domains and change domain settings. A domain administrator can invite conversation inputters, manage their permissions, and directly access the NAVER Cloud Platform console.

User

In this guide, a user refers to a person who uses the chatbot.

User variable

A value that is stored and reused while a conversation session is maintained. By setting information such as the user's name, phone number, and location as user variables, the chatbot can use the information in conversations without asking the user for the same information again.

Slot

An item of information that must be obtained from the user in a task. A task can be performed when all required slots are filled, and each slot is mapped to an entity.

Failure message

A message sent when the chatbot cannot understand the user's question or find an appropriate answer.

Action method

A feature used to retrieve data from an external system and include it in a chatbot answer. The result of calling a registered URL can be used in an answer and is entered in the ${action-method} format.

Entity

A dictionary containing domain-specific words such as names of people, organizations, places, dates, and products. By defining words with the same or similar meanings as entities, the chatbot can analyze entities in user utterances to identify user intent. Entities are also used to configure slots in a task.

Welcome message

A message sent to the user when the chatbot is launched.

Natural language

A language used by people in everyday life, as opposed to artificial languages understood by computers.

Retrain

The process of analyzing users' conversation history, adding questions that the chatbot could not answer to the conversation dataset, and retraining the model.

Regular expression parameter

A feature that defines frequently used patterns as variables when creating regular expression questions or pattern entities. There are system variables commonly used across Korean domains and domain variables used only in specific domains.

Creator

In this guide, a creator refers to an administrator who designs and builds chatbots.

Chatbot builder

A tool that provides the features required to build and operate chatbot services. When a conversation dataset is entered and trained, the trained model identifies user intent and provides an appropriate answer. It provides advanced features such as contexts, forms, and action methods for implementing various types of business logic, as well as operational features such as testing, retraining, and metrics.

Context

Information used to connect conversations and understand the context of connected conversations.

Task

A feature used for conversations that must proceed sequentially. It obtains the information required for slots from the user and provides a final answer or executes an action when all required slots are filled.

Form

A feature that provides multiple-choice or open-ended questions and continues the conversation based on the user's answers. It is entered in the #{form-name} format and can be integrated with an action method to execute a specified action when the user selects an answer.