Global entities are automatically detected by the system, so you don’t have to add them manually. Global entities include things like POSTAL_CODE, where enumerating the full list would be difficult, and STREET, where predefining a format would be impossible. Global entities include:

Entity Description Example
AIRPORT An airport code SEATAC
CITY A city Paris
Los Angeles
COLOR A color blue
COUNTRY A country Canada
United States
DATE Dates and absolute date&timestamps today
6 p.m. tomorrow
DURATION A time period 2 weeks
2 weeks and 3 days
half a day
EMAIL An email address
MONEY Numbers with currency $2000
23 dollars
fifty bucks
ten pounds
ORG Names of institutions Nike factory
World Health Organization
PERCENT A percentage 100%
forty percent
PERSON_NAME Names of persons John
PHONE A phone number 800-555-1212
+66 11-222-3344
POSITION_IN_SERIES A number used in the context of order 15th
POSTAL_CODE United States postal code 10001
QUANTITY A quantity 5
SET A group
In, “The meeting with Bob is weekly on Tuesdays,” PERSON_NAME = Bob, SET = weekly, DATE = Tuesdays
every day
STATE United States state NY
New York
STREET United States descriptors for street names Main St.
123 Main St. NE
123 East-West Highway Apt. 107
TIME Time of day 2 p.m.
URL A URL<param>

Keep in mind that the detection of global entities is highly dependent on context. As a result, the system is powerful and capable of detecting the following:

  • Message: My name is Paris and I live in Paris
  • Entities: PERSON_NAME = Paris, CITY = Paris

  • Message: Hi Tuesday, can you arrive at 2pm on Tuesday?
  • Entities: PERSON_NAME = Tuesday, TIME = 2pm, DATE = Tuesday

Detection of entities is trained on commercial messages, so depending on context, you might get results that you don’t expect:

  • Non-commercial message: Washington cherry trees are beautiful this time of year
  • Entities: CITY = Washington but
  • Commercial message: Do you ship this product to Washington?
  • Entities: STATE = Washington

Using the Assist tool, you can assign global entities to user interactions and have the bot populate a slot with the user's input to the question to which the entity was assigned.