# Pre-built Slots
# What are pre-built slots?
# Pre-built slots v1
Template slots are pre-built slot extractors to help you quickly extract values that belong to that specific slot. It doesn't require training data upfront and is non-configurable. The Platform currently supports money and date.
Let's take a deeper look of the fields inside the DATE and MONEY list (you can find it in the query sidebar or full response payload):
"slot values": {
"_DATE_": [
{
"type": "DATE",
"startDate": "2020-02-07T00:00:00Z",
"endDate": "2020-02-07T23:59:59Z",
"period": "P1D",
"altDates": [
{
"descriptor": "PAST_RESOLVE",
"startDate": "2019-02-07T00:00:00Z",
"endDate": "2019-02-07T23:59:59Z"
},
{
"descriptor": "FUTURE_RESOLVE",
"startDate": "2021-02-07T00:00:00Z",
"endDate": "2021-02-07T23:59:59Z"
}
],
"anchor": null,
"tokens": "today"
}
],
"_MONEY_": [
{
"tokens": "$ 200",
"value": 200,
"currency": "dollars",
"visual": "$200.00",
"speakable": "two hundred dollars"
}
]
},
# DATE
| Key | Value | Description |
|---|---|---|
| type | string: DATE, DURATION, TIME, or SET | DATE: a calendar date or period. It can be last week, this year or a specific holiday. DURATION: a span of time (or range) for example from March to May, since June, after September. TIME: a time of day or offset for example 9am, 2 o’clock, morning evening etc. The Z in the startDate or endDate field is short for "zulu" time i.e. GMT/UTC time. SET: a recurrence of dates or times for example every month, weekly etc. |
| startDate/ endDate | string | Identified start time and end time respectively. The format come out as standard python date format. It consists of year, month, day, hour, minute, second and time offset. |
| period | string | Amount of time between the startDate and endDate. "P1W" stands for period of 1 week; "P1M" stands for period of 1 month; "PT24H" stands for period of time of 24 hours. |
| altDates | list | altDates conditionally contains extra dates when a date can be cast into the past or future. This is especially useful for ambiguous dates. For example, "June" may return the month of June for the current year as well as last June, next June, or both depending on where we are at in the year [show past_resolve and future resolve in the payload]. |
| anchor | string | anchor is used with recurrences (like yearly, each month, etc.) to give a point of reference for the recurrence. These anchor values conform to the TimeML standard. |
| tokens key | string | Tokens extracted from the query to build the date object |
# MONEY
| Key | Value | Description |
|---|---|---|
| tokens | string | Value extracted from the query |
| value | integer | The numerical value converted from the token extracted |
| currency | string | The currency type The platform currently supports USD and synonyms such as dollar, buck, and euros, pounds and Turkish lira. |
| visual | string | The human readable format meant for visual display |
| speakable | string | The extended format meant to be used in text to speech services |
# Pre-built slots v2
Version 2 pre-built slots offers more capabilities number, position, phone, and email.
"slot values": {
"_DATE_": [
{
"value": {
"grain": "day",
"from": "2021-02-06T00:00:00+00:00",
"to": "2021-02-06T23:59:59+00:00",
"normalized": {
"unit": "second",
"value": 86400
},
"visual": "on February 6th, 2021",
"speakable": "on february sixth"
},
"alt_values": [
{
"grain": "day",
"from": "2021-02-13T00:00:00+00:00",
"to": "2021-02-13T23:59:59+00:00",
"normalized": {
"unit": "second",
"value": 86400
},
"descriptor": "future",
"visual": "on February 13th, 2021",
"speakable": "on february thirteenth"
},
{
"grain": "day",
"from": "2021-01-30T00:00:00+00:00",
"to": "2021-01-30T23:59:59+00:00",
"normalized": {
"unit": "second",
"value": 86400
},
"descriptor": "past",
"visual": "on January 30th, 2021",
"speakable": "on january thirtieth"
}
],
"type": "time",
"tokens": "on saturday",
"version": "2"
}
],
"_MONEY_": {
"type": "dict",
"values": [
{
"currency": "dollars",
"speakable": "thirty five dollars",
"status": "MAPPED",
"tokens": "$ 35",
"value": 35,
"visual": "$35.00"
}
]
},
"_EMAIL_": [
{
"value": "john@example.com",
"tokens": "john@example.com",
"version": "2"
}
],
"_PHONE_": [
{
"value": "5555555555",
"tokens": "(555)555-5555",
"version": "2"
}
],
"_POSITION_": [
{
"offset": 0,
"span": 3,
"anchor": "start",
"type": "position",
"tokens": "first three",
"version": "2"
}
],
"_NUMBER_": [
{
"value": 22,
"words": "twenty two",
"digits": "two two",
"type": "number",
"tokens": "twenty two",
"version": "2"
}
]
}
# Date
Date is either a time, duration, or recurrence (depending on the type).
| Key | Value | Description |
|---|---|---|
| from | string | the starting point of time |
| to | string | the ending point of time (inclusive) |
| grain | string | the granularity of the time (e.g. day, year) |
| value | integer | how many of the grain are in the recurrence |
| times | integer | how many times the recurrence happens given the grain and value |
| anchor | object | an anchor time which to base the recurrence off of |
| inner | object | a basis to build a recurrence which cannot be represented by an anchor from |
| inner.grain | string | the granularity of the recurrent time (e.g. day, year) |
| inner.day | integer | the day within inner.grain (e.g. 1 => first day) |
| inner.instance | integer | the instance of inner.grain within grain to repeat (e.g. 2 => 2nd instance) |
| total_seconds | string | amount of time in seconds between from/to |
| offset_seconds | string | amount of time in seconds since now |
| visual | string | the time for visual display (e.g. on December 2nd, 2019) |
| speakable | string | the time for speakable display (e.g. on december second) |
| alt_values | list | alternative values given ambiguous cases |
| type | string | time, duration or recurrence |
| tokens | string | extracted tokens |
| version | string | 2 (duckling version number) |
Alternative values can be helpful depending on your use case. You can always choose a future value for a transfer or a past value for transaction searching. It works the same as the altDates from the pre-built slots v1.
Note: Fields to/from will be null whenever it is an open-ended range (e.g. "since monday" or "before saturday")
# Money
| Key | Value | Description |
|---|---|---|
| value | integer/ float | the money value as a number |
| currency | string | the currency of the extraction |
| descriptor | string | (max |
| visual | string | the money amount for visual display (e.g. $20.00) |
| speakable | string | the money amount for speakable display (e.g. twenty dollars) |
| tokens | string | extracted tokens |
| version | string | 2 (duckling version number) |
Duckling for the most part follows the ISO 4217 standard for currencies (using EUR, etc. as the currency code), but uses symbols/words for ambiguous currencies (e.g. dollar or pound). The extraction can also be max/min value in which case the slot value has an extra descriptor field which is either max or min.
# Phone
| Key | Value | Description |
|---|---|---|
| value | string | the phone number value formatted properly |
| tokens | string | extracted tokens |
| version | string | 2 (duckling version number) |
| Key | Value | Description |
|---|---|---|
| value | string | the email value formatted properly |
| tokens | string | extracted tokens |
| version | string | 2 (duckling version number) |
# Number
Number is a decimal or ordinal number with several helpers.
| Key | Value | Description |
|---|---|---|
| value | integer/ float | the number value |
| words | string | the number written out as words (e.g. twenty third or fifty) |
| digits | string | the number written out as digits (e.g. 23rd or five zero) |
| type | string | number or ordinal |
| tokens | string | extracted tokens |
| version | string | 2 (duckling version number) |
# Position
Position is a list selection helper that enables the user to refer to a list of items given their position (as opposed to their content).
| Key | Value | Description |
|---|---|---|
| offset | integer | the starting position given the anchor |
| span | integer | how many indices to include given the offset |
| anchor | string | (start |
| type | string | ordinal or position |
| tokens | string | extracted tokens |
| version | string | 2 (duckling version number) |
To help visualize how offset works, here are a few examples:
first
offset: 0
span: 1
anchor: start
0 1 2 3 4 5
^
last two:
offset: -2
span: 2
anchor: end
-5 -4 -3 -2 -1
^ ^
three in the middle:
offset: -1
span: 3
anchor: middle
-5 -4 -3 -2 -1 0 1 2 3 4 5
^ ^ ^
# How to add/remove pre-built slots?
In the competency sidebar,
In the overflow menu
next to the competency name in the competency sidebar, select edit.In the modal popped up, select the pre-built slots you wish to add to your competency.
Click Save.
