Definition:
An Entity is a specific word or phrase that the bot detects and extracts from user messages to use as a variable in its operations. For example, in the sentence "I need a flight to Paris," "Paris" would be recognized as an entity representing the "arrival_city."
Types of Entities:
Custom Entities: These are user-defined entities that you can control, either as a list of specific values or patterns (e.g., cities, flight numbers).
System Entities: Pre-trained AI models that detect entities based on conversational patterns without predefined keywords (e.g., dates, locations).
Entity Management:
Download: Rule-based entities can be downloaded as CSV files for easy management.
Creation: To create a new entity, use the "Create entity" button, then define rule-based or system-based detection methods.
Custom Entities
Custom entities can be rule-based, where specific values are predefined for detection, or normalized, where values are transformed using an API. For example, a custom entity detecting cities in the US might include various aliases for a city like "New York" (e.g., "NYC," "The Big Apple").
System Entities
System entities use AI models trained on conversational data to detect entities like dates, emails, or locations. These entities are more flexible, recognizing a broader range of inputs without predefined keywords.
Wrappers
Wrappers help qualify entities by checking for specific keywords before or after the entity. For example, in the phrase "I leave from Paris to Tokyo," a wrapper can distinguish "Paris" as the departure city and "Tokyo" as the arrival city.
These elements allow for precise detection and extraction of information from user interactions, improving the bot's ability to understand and respond appropriately.