Entity Node - Kore.ai Docs

The Entity Node extracts specific data from user utterances to fulfill intents in a dialog task. For every critical data point you need from a user, create a corresponding entity node with a prompt. Example: For the utterance “Book me a flight from LA to NYC on Sunday”, the agent needs to extract source_city (LA), destination_city (NYC), and departure_date (Sunday) — each via a separate entity node. The platform supports ~30 entity types (Address, Airport, Date, Currency, and more), custom regex patterns, and PII masking at the node level.

Chain multiple entity nodes to collect a series of inputs (username, location, amount, due date). Follow them with a Webhook node to process an API request.

Add the Node

Open the dialog task.
Add the Entity node. See Add a Node.
The Entity window opens with Component Properties selected by default.

Multilingual agents: Adding a new entity applies it across all languages (shared flow definition). Language-specific properties — prompts, error messages, synonym lists — require separate values per language.

Configure the Node

Component Properties

Changes here apply across all dialog tasks that use this node.

Enter a Name and Display Name. Entity names cannot include spaces.
Select an Entity Type from the dropdown. See Entity Types.
If the type supports it, enable Multi-item to allow multiple values separated by commas, semicolons, or “and”.
Under User Prompt, enter the prompt shown to the user (for example, Enter the Departure Date).
- Click Manage for channel-specific prompts. See Prompt Editor.
- Click Settings > enable Override for this instance to set an instance-specific prompt. Disabling the toggle deletes the override and restores the component-level prompt.
Under Error Prompts, review and modify default error messages.
- Click Manage to edit. See User Prompts.
- Enable Override for this instance for instance-specific error messages.
- Enable Present Prompts in the Order of Retries to reorder error messages using drag handles.
Toggle Rephrase Responses to rewrite agent replies using AI based on conversation context and user emotion. See GenAI Features.
Under Redaction of PII Data, set how PII entity values appear in prompts and messages at runtime:
- De-identify PII data — shows the redacted value.
- Use the original value — shows the actual value (useful for user confirmation).
These options affect only runtime display. PII data is always redacted in chat history and internal logs, regardless of this setting.
Enable Sensitive Entity to mask, redact, or replace sensitive input at the node level. Disabled by default. When enabled:
- Check Transient Entity to clear the data on session close. Chat history shows data_purge instead of the value.
- Click +Add Pattern to add a regex for identifying sensitive data (for example, [a-zA-Z]{3}[-]\d{4}).
- Select how to display sensitive data to unauthorized users:
  - Redaction — replace with a unique random alphanumeric value.
  - Replacement — replace with a static value you define.
  - Mask with character — mask leading and trailing characters with + or #.
See Redacting PII Data.
Under Variable Namespaces, associate namespaces to scope this node’s execution. Visible only when Variable Namespace is enabled for the agent. See Managing Namespace.

Instance Properties

Settings here apply only to this dialog task instance — not to other tasks using this node.

User Input

Option	Description
Mandatory	User must provide a valid value before the flow continues.
Optional	User is prompted once; flow continues with any input. Set a Default Value as fallback when the user provides no input.
Hidden	Agent does not prompt; captures value only if the user volunteered it in a prior utterance. Set a Default Value as fallback.

Mandatory-only settings:

Setting	Description
Allowed Retries	Max re-prompts for invalid input (1–5, default 5). After all retries are exhausted, triggers Behavior on Exceeding Retries.
Voice channel exception	On voice channels, when input doesn’t match the entity type and a Node Grammar is defined, the platform uses the No Match retry count from Voice Call Properties instead of Allowed Retries.
Behavior on Exceeding Retries	End of Dialog or Transition to a Node. Creates a “Behavior on Exceeding Retries” connection rule. Customize the message in Standard Responses.

User Input Correction (v7.3, String entities only)

Option	Description
Auto-correct user input (default)	Platform builds a custom dictionary from synonyms, task names, pattern words, and small talk utterances. Not supported in all languages.
Do not auto-correct user input	Disables autocorrection. Applied automatically for existing agents with pre-v7.3 entity extraction settings.

Entity Extraction

Option	Description
Evaluate unused text from previous utterance (default)	Uses text not already consumed by another entity in this dialog.
Evaluate unused and used text from previous utterances	Can reuse values already extracted by another entity node.
Do not evaluate previous utterances	Ignores prior utterances; always prompts the user explicitly.

Do not reuse: Prevents this entity’s input from being reused to extract other entities.
Entity Rules: Configure JSON-based validation rules. Click Add Rules and enter JSON. See Entity Rules.

Advanced Controls

Setting	Options
Intent Detection (String/Description entities only)	Accept as entity value (ignore intent) \| Prefer as intent and use Hold & Resume settings \| Ask the user
Interruptions Behavior	Use task-level setting \| Customize for this node. See Interruption Handling.
Analytics - Containment Type	Use task-level default \| Customize (Self Service or Drop-off)
Precedence (all types except String/Description)	Intent over Entity or Entity over Intent when the user’s input contains both. See NLU Configurations.
Custom Tags	Add tags for custom analytics profiles. See Custom Meta Tags.

User Input Flow

At each entity prompt, the platform processes input as follows:

Valid value → entity populated; dialog continues.
Ambiguous value → platform shows an ambiguity resolution prompt.
Invalid input at the ambiguity prompt:
- If valid for the entity type → used as-is; dialog continues.
- If invalid and triggers an intent/FAQ/small talk → handles per Hold & Resume; entity is re-prompted when dialog resumes.
- If invalid and no intent triggered → entity left blank; dialog continues from transitions.
Retry limit exceeded → triggers Behavior on Exceeding Retries. For child tasks set to End of Dialog, the parent task also terminates.

NLP Properties

Under Suggested Synonyms, enter synonyms for the entity and press Enter after each. See Managing Synonyms.
Under Suggested Patterns, click + Add Pattern to add extraction patterns. See Managing Patterns.
Under Manage Context > Context Output, define context tags set when this entity is populated.
Enable Auto emits the entity values captured to add entity values to the context object automatically. See Context Management.

IVR Properties

Configure input mode, grammar, prompts, and call behavior for IVR channels. See Configure Dialog Node IVR Properties.

Connection Properties

If this node is last in the sequence, only the connection property is visible.

Define which node executes next using conditions based on entity values or context objects. These conditions are instance-specific and do not affect other dialog tasks. See Adding IF-Else Conditions to Node Connections. To delete a condition, hover over it and click the bin icon. Deleted conditions cannot be restored. Default connection variants:

Variant	Behavior
Not Connected	No next node defined.
End of Dialog	Explicitly ends the dialog.
Return to Flow	Ends the Dialog Task; returns control to Flow Builder at the next node.

Deflect to Chat (Return to Flow only): Transitions voice interactions to chat. Types: Automation or Agent Transfer.

Deflect to Chat works only with Kore Voice Gateway channels (Phone number or SIP Transfer).

Entity Types

The entity type tells the NLP interpreter what data to expect from user input, improving recognition accuracy.

Entity Type	Returns	Description
Address	String	US and Germany standard address formats
Airport	JSON	City, airport name, IATA, ICAO
Attachment	String	File, image, or email up to 25 MB
City	String	City name (population > 5,000)
Color	String	Named colors with variants
Company	String	Company or organization names
Composite	Mixed	Multiple entity values in one entity
Country	JSON	Country name with alpha-2, alpha-3, and numeric codes
Currency	JSON	Amount and currency type
Custom	String	User-defined regex, concept, or variable
Date	String (ISO8601)	Date: `YYYY-MM-DD`
Date Period	JSON	Start and end date
Date Time	String (ISO8601)	Date + time: `YYYY-MM-DDThh:mm:ss.sTZD`
Description	String	Multi-sentence free text
Email	String	Email address
List of Items (enumerated)	String	Static or context-based selection list
List of Items (lookup)	String	Static or remote lookup list
Location	JSON	City/state with address and coordinates
Number	Number	Numeric value (max 18 digits)
Person Name	String	Full name of a person
Percentage	Float (0.0–1.0)	Percentage value
Phone Number	String	3–12 digit phone numbers
Quantity	JSON	Amount, unit, and quantity type
String	String	Single-sentence free text
Time	String (ISO8601)	Time: `HH:MM:SS.sZD`
Time Zone	String	Timezone as GMT offset
URL	String	Web URL
Zip Code	String	Postal/zip code

Address

Captures US and Germany standard address formats. For other countries, captures strings ending with a recognizable city or country name. Format: Street Name, City (required), Country (required), Zip Code

{ "AddressEntity": "200 E Main ST Phoenix AZ USA 85123" }

Airport

Captures airport details via city name, airport name, IATA/ICAO code, or US city abbreviations. Data source: opentraveldata.

{
  "AirportEntity": {
    "IATA": "LHR",
    "AirportName": "London Heathrow Airport",
    "City": "London",
    "ICAO": "EGLL"
  }
}

Input	Behavior
City name	Identifies airport; shows list if the city has multiple airports
Airport name	Matches full or partial name with a prominent keyword
IATA code	Identifies airport by IATA code
ICAO code	Identifies airport by ICAO code
City abbreviation	Uses geonames.org abbreviations

Attachment

Captures a file, image, or email attachment up to 25 MB.

{ "AttachmentEntity": "send" }

Supported channels only: Facebook, Twitter, Web/Mobile, Slack.

City

Captures any city with population > 5,000 as a string. Data source: geonames.org.

{ "CityEntity": "New York" }

City disambiguation is not performed. Cities are prioritized by population (for example, Portland OR ranks above Portland ME).

Color

Captures color names from utterances. Returns as a string.

{ "ColorEntity": "green" }

Color	Recognized variants (sample)
green	celadon, chartreuse, emerald, lime, mint, olive, sage
brown	amber, bronze, mahogany, russet, tan, chocolate, cinnamon
orange	auburn, orangish
blue	aqua, azure, cerulean, navy, teal, turquoise, sapphire
purple	magenta, mauve, orchid, lavender, lilac, violet
red	crimson, fuchsia, maroon, pink, scarlet, vermilion
yellow	beige, gold, lemon, mustard, peach, saffron
black	ebony, jet black, licorice
white	chalk, cream, ivory, off-white
grey	ashen, gray, grayish, hoary, taupe
metallic	argent, brass, bronze, copper, gold, platinum, silver
transparent	—
reflective	—

Company

Captures company or organization names. Returns as a string.

{ "OrganizationEntity": "amazon" }

Recognizes:

A large corpus of known global companies, mapping name variants to a canonical name (for example, Amazon, Amazon.com, Amazon Inc → “amazon”).
Any capitalized word followed by a legal suffix: Inc, Incorporated, Corp, Corporation, Group, Ltd, Limited, Co, Company, LP, LLP, LLLP, LLC, PLLC.

Composite

Captures multiple entity values in a single entity. Use when a user utterance contains a combination of related data points. Example: “I want to fly first class from LA to NYC tomorrow” — captures flight class, departure city, destination city, and travel date in one composite entity.

Country

Captures country names from utterances. Returns as JSON.

{
  "CountryEntity": {
    "alpha3": "USA",
    "alpha2": "US",
    "localName": "United States of America",
    "shortName": "United States",
    "numericalCode": 840
  }
}

Field	Description	Example
alpha3	Three-letter code	USA, GBR, IND
alpha2	Two-letter code	US, GB, IN
localName	Full country name	United States of America
shortName	Short name	United States
numericalCode	UN M49 code	840, 826, 356

Avoid special characters (&, ()) in inputs. Use commas or “and” to separate list items.

Currency

Captures amount and currency type from utterances. Returns as JSON.

{ "CurrencyEntity": { "code": "SGD", "amount": 20 } }

Recognizes: full names (Dollar, Rupees), symbols ($, £), abbreviations (INR, USD), and slang (Buck, Quid, Loonie, Benjamin).

Currency is not disambiguated by prior usage. USD may rank above SGD for “dollar” unless the user explicitly says SGD.

Custom

Validates user input against a regex, concept variable, or content/context/environment variable.

Type	Example input	Acceptable input
Regular Expression	`[a-zA-Z]{3}[-]\d{4}`	NLP-1234
Concept Variable	`~car_brands`	Tesla
Content/Context/Env Variable	`content.TicketNumber` (containing a regex)	NLP-1234

The dialog fails if the expression resolves to null. Use the preConditions entity rule to conditionally skip the node instead.

See Regex Expressions.

Date

Captures a date from utterances. Returns ISO8601 format: YYYY-MM-DD.

{ "DateEntity": "1982-04-13" }

Recognizes formatted dates, number-only dates, named months, relative dates (today, in 3 days, last Thanksgiving), and weekdays (next Saturday, first Sunday of next month).

Date Period

Captures a start date and end date. Prompts separately for each missing date.

{ "DatePeriodEntity": { "from": "2024-11-05", "to": "2024-11-10" } }

Input	Behavior
Neither date	Prompts for From Date
One date missing	Prompts for the missing date
Implicit reference + duration (5 days from Tuesday)	Calculates both dates
From date + duration (5 days from Nov 15)	Calculates both dates
Both dates explicit (from 5th to 10th)	Captures both

Date Period supports two separate user and error prompt sets: one for From Date and one for To Date.

Date Time

Captures date and time combined. Returns ISO8601 format: YYYY-MM-DDThh:mm:ss.sTZD.

{ "DateEntity": "2017-10-10T18:00:00+05:30" }

Description

Captures statements or paragraphs of free text. Returns as a string (supports wildcard characters).

{ "Description": "text here" }

Multi-item is not available for Description.

Email

Captures email addresses from utterances. Returns as a string.

{ "Email": "help@example.com" }

List of Items (enumerated)

Displays a defined list of values to the user for selection. Click Customise The List Of Items to configure.

List type	Configuration
Static List	Enter Display Name, Value, and Synonyms per item. Set Auto-Correction threshold.
List from Context	Specify a context variable, display name key, value key, and synonyms key.

Auto-Correction: Accepts closest-match inputs based on a configurable character-similarity threshold. Does not apply to dictionary words or alphanumeric inputs. Context object keys (added post-v7.1):

ambiguousEntityValues — array of ambiguous matches (title, value, synonym). Reset on re-prompt.
synonymsUsed — synonym used to identify the matched item. Reset on re-prompt.

To show the list of values in the prompt, set Display List of Values to the channel-formatted option. Works with Plain Text prompts only — not JavaScript templates.

List of Items (lookup)

Displays a list defined via static configuration or a remote service call. Click Customise The List Of Items to configure.

Static List

Tab	Description
List of Values	Enter Display Name, Value, and Synonyms per item
JSON Preview	Enter key/value/synonym pairs as JSON
Upload	Upload a JSON or CSV file

Remote List

Use for large datasets or security-restricted lookups via an external service. Step 1 — Define the service call. The platform populates context.inputData before calling the service:

{
  "inputData": {
    "input": ["get account"],
    "usedUp": ["0-x-x"],
    "isMultiItem": false
  }
}

Field	Description
`input`	Array of user inputs for the current dialog
`usedUp`	Index of words already used by other entities. Format: `sentenceIndex-startIndex-endIndex`
`isMultiItem`	Set to `true` if multiple values are expected

Step 2 — Map the response. Configure how to read the service response:

Field	Access path
Context variable (array)	Contains service response data
Display Key Name	`{{context.entities.<entity-name>.title}}`
Value Key	`{{context.entities.<entity-name>.value}}`
Synonyms Key	`{{context.entities.<entity-name>.synonym}}`
Matched Word Index	Same format as `usedUp` in `context.inputData`

Step 3 — Runtime flow.

Platform populates context.inputData and calls the service.
One value returned → assigned to entity.
Multiple values returned → platform extracts match from the list. No match → user selects from the list.
Ambiguous values → user selects one.
Multi-item entity → all valid values assigned; invalid values discarded.
Empty or malformed response → falls back to User Prompt.

Context keys added post-v7.1: ambiguousEntityValues, synonymsUsed (same as enumerated).

Location

Captures city or state location with address and coordinates. Returns as JSON.

{
  "Location": {
    "formatted_address": "Las Vegas, NV, USA",
    "lat": 36.1699412,
    "lng": -115.1398296
  }
}

Number

Captures a numeric value from utterances. Recognizes spelled-out numbers and abbreviations (1M). Consecutive number words are combined (“one two three” → 123).

{ "NumberEntity": 16 }

Maximum 18 digits.

Person Name

Captures a person’s full name. Returns as a string. Platform identifies capitalized words as names, taking up to three consecutive capitalized words.

{ "PersonName": "John Smith" }

Percentage

Captures percentage values. Returns as a float in range 0.0–1.0.

{ "PercentageEntity": 0.6 }

Supports: percent, percentage, %.

Phone Number

Captures 3–12 digit phone numbers. Returns as a string. Handles non-standard formats like (407)-407-4077 and 407.407.4077. Supports Arabic phone numbers spelled out in words.

{ "PhoneNumber": "+4075551212" }

Quantity

Captures amount, unit, and quantity type from utterances.

{
  "Quantity": {
    "unit": "milliliter",
    "amount": 500,
    "type": "volume",
    "source": "500 ml"
  }
}

Select a Unit Type and Default Unit when configuring the entity.

Type	Supported units
Length	kilometer, meter, centimeter, inch, foot, yard, mile
Area	square kilometer, square meter, square mile, square yard, square foot
Volume	cubic meter, liter, milliliter, gallon, ounce
Time	hour, minute, second, millisecond
Speed	meters/second, km/hour, miles/hour
Pressure	pascal, atmosphere, bar
Energy	calorie, kilocalorie, watt-hour, kilowatt-hour
Memory	bit, byte, kilobyte, megabyte, gigabyte, terabyte
Weight	ton, kilogram, gram, pound, ounce
Angle	degree
Age	day, week, month, year, decade, century
Temperature	celsius, fahrenheit

String

Identical to Description but limited to one sentence. No validation is applied unless the entity is trained. Use as a last resort when no other entity type fits.

Time

Captures time from utterances. Returns ISO8601 format: HH:MM:SS.sZD.

{ "TimeEntity": "T06:00:00+05:30" }

Recognizes: am/pm, spelled-out numbers (Six AM), and context phrases (Six in the evening).

Time Zone

Captures a timezone and converts it to GMT offset.

{ "TimeZoneEntity": "-06:00" }

Recognizes standard timezone names (Eastern Standard Time → -6:00).

URL

Captures web URLs from utterances. Returns as a string.

{ "URLEntity": "www.kore.ai" }

Zip Code

Captures postal/zip codes. Returns as a string. Handles hyphens, spaces, and leading zeros. Does not extract codes from regex-like patterns (for example, #$A1B2%).

{ "ZipcodeEntity": "32746" }

Entity Rules

Entity rules add validation and processing hints beyond basic type restrictions. Apply rules in two ways: Via the UI: Go to Instance Properties > Entity Rules and enter JSON in the editor. Via a Script Node (placed before the entity node in the dialog):

context.entityRules.<entityName> = {
  "ruleName": "value"
}

For subentity rules in a composite entity:

context.entityRules.<compositeEntityName> = {
  <subentityName>: {
    "ruleName": "value"
  }
}

Generic Rules

Apply to all entity types.

Rule	Values	Description
`allowConfirmation`	true/false	Presents the extracted value to the user for confirmation before continuing. Currently applies to LoV enumerated types only.
`confirmYesSynonyms`	`["~concept1", "~concept2"]`	Additional words/phrases to confirm an entity value. Use with `allowConfirmation`.
`confirmNoSynonyms`	`["~concept1", "~concept2"]`	Words/phrases to cancel confirmation; sets entity value to null. Use with `allowConfirmation`.
`processLatestSentence`	true/false	Restricts entity extraction to sentences from the current volley only.
`patternsOnly`	true/false	Restricts extraction to defined entity patterns only; disables the default NLP fallback.
`preConditions`	`["contextTag"]` or `["( pattern )"]`	Extracts the entity only when a specified context tag/trait is set or a pattern matches. Skips extraction entirely if the condition is not met.
`isCurrencyCode`	true/false	Enables (true) or disables (false) currency code identification in multilingual setups.

preConditions examples:

{ "preConditions": ["isCustomerX"] }

Extracts entity only if the isCustomerX context tag is set.

{ "preConditions": ["( buy )"] }

Extracts entity only if the word “buy” appears in the utterance. Utterance buy 10 apples → 10; utterance search for 10 apples → no extraction.

String and Description Rules

Rule	Values	Description
`discardAsCommand`	true	Treats discard commands (abort, cancel, exit, quit, stop, and synonyms) as dialog discard, not as string input.
`stripLeading`	`["~concept"]`	Removes words in the concept from the start of the extracted string.
`stripTrailing`	`"~concept ~concept2"`	Removes words in the concept from the end of the extracted string. Value can be a string, space-separated concepts, or an array.
`avoidSingleWord`	`"~concept"`	Ignores any single-word value belonging to the concept, unless it is the entire input.
`avoidSingleVerb`	true	Ignores single-verb values unless the entire input is a verb.
`extractOnlyNumbers`	true	Extracts only the numbers present in the string.
`letters`	`2` or `"5-8"`	Extracts words composed of exactly N alphabetic letters, or within a range. Punctuation and numbers are excluded. Returns uppercase.

letters examples:

{ "letters": 2 }

Utterance	Extracted
`234 XY`	`XY`
`234 x y`	`XY`
`234 alpha bravo`	`AB` (phonetic alphabet)
`234 X Y Z`	nothing (no 2-letter match)

{ "letters": "5-8" }

Utterance: Kore.ai is the best platform in artificial intelligence of recent times. Extracted: platform, recent, times

Number Rules

Rule	Values	Description
`asString`	true	Captures the number as a string, preserving leading zeros.
`digits`	`"3"` or `"13-17"`	Enforces exact digit count or a valid range. Preserves leading zeros. Implies `asString=true` and `integerOnly=true`. Decimals and ordinals are not matched.

digits does not split spoken number groups. For input "two thousand and twenty three", the rule "digits": 4 will not match. Use individual spoken digit groups like "twenty twenty three" for a 4-digit match.

Examples:

{ "asString": true }

Utterance: OTP is 009944 → Extracted: "009944" (without rule: 9944)

{ "digits": "3" }

Utterance: CVV is 034 → Extracted: "034"

{ "digits": "13-17" }

Utterance: My credit card number is 4012 8888 8888 1881 → Extracted: "4012 8888 8888 1881" (16 digits, within range)

Currency Rules

Rule	Values	Description
`defaultCode`	`"NZD"` or `"NZ"`	Default currency when none is specified. Accepts 3-letter currency code or 2-letter country alpha-2 code.
`maxDigits`	number	Max digits allowed in the amount. Exceeding amounts are discarded.
`currencyCodes`	`["USD", "INR", "NZD"]`	Restricts accepted currencies to this list.
`ignoreOnlyCode`	true/false	`true`: requires both code and amount; code alone is rejected. `false`: accepts code without amount.

Examples:

{ "defaultCode": "NZD" }

Utterance: Pay 30 → Extracted: NZD30; utterance Pay USD30 → Extracted: USD30

{ "maxDigits": 3 }

Utterance: Pay USD3000 → Prompts for value (exceeds 3 digits)

{ "currencyCodes": ["USD", "INR", "NZD"] }

Utterance: Pay AUD30 → Prompts for value (AUD not in list)

{ "ignoreOnlyCode": true }

Utterance: Pay AUD → Error: amount required

Composite Entity Rules (Model Number)

Rule	Values	Description
`modelNumber`	true or `<separator char>`	Captures structured IDs (membership number, SSN, seat) from a composite entity. Concatenates matched subentity values into a single string instead of a JSON object.

modelNumber is a voice-compatible alternative to custom regex. Regex requires exact character matching, which breaks on ASR transcription variations (for example, “C” → “sea”, or a pause splitting “23” → “20 3”). Using Number and String entity types with the digits and letters rules handles all these variations. Example: Pattern @AccountNumbers @AccountSuffix where:

@AccountNumbers: Number entity with "digits": 8
@AccountSuffix: String entity with "letters": 2

Utterance	Extracted
`3 3 1 3 4 2 1 2 X Y`	`33134212XY`
`33134212 – XY`	`33134212XY`
`33134212 alpha bravo`	`33134212AB`
`thirty three thirteen forty two twelve a like apple b as in boy`	`33134212AB`

A modelNumber composite entity is a voice-compatible replacement for a custom regex entity like \d{8}\-?[a-zA-Z]{2}.

Person Name Rules

Rule	Values	Description
`disablePatterns`	`["possessive"]`	Disables the possessive pattern so “Bob’s” extracts “Bob” instead of “Bob’s”.
`ignoreWords`	`["word", "~concept"]`	Words/concepts not treated as names even when capitalized. Value can be a string, space-separated concepts, or an array.
`negativePatterns`	`["about *"]`	Patterns that prevent extraction (for example, “about Philip” → Philip is not captured).
`ignorePunctuation`	`["-", ","]`	Punctuation to ignore in person name extraction. Applies to non-English languages (excludes German, French, Spanish).

Examples:

{ "disablePatterns": ["possessive"] }

Utterance: schedule Bob's review at 9 am → Extracted: Bob

{ "ignoreWords": ["review", "~prepositionList"] }

Utterance: meeting for Bob Review → Extracted: Bob (without rule: Bob Review)

{ "negativePatterns": ["about *"] }

Utterance: schedule a meeting about Philip with Fred → Extracted: Fred (without rule: Philip)

Company Rules

Rule	Values	Description
`ignoreWords`	`["atm"]`	Words/concepts not treated as company names even when capitalized.
`negativePatterns`	array of patterns	Patterns that prevent company name extraction in specific scenarios.

Example:

{ "ignoreWords": ["atm"] }

Utterance: find ATM → Extracted: nothing (without rule: ATM, recognized as an Italian company)

Date Rules

Rule	Values	Description
`range`	`{"from": "2020-01-01", "to": "today"}`	Accepts dates only within the range. Either endpoint is optional. Values: `YYYY-MM-DD` or keywords: `today`, `tomorrow`, `yesterday`. Endpoints are inclusive.
`referenceDate`	`"2020-07-09"`	Sets a fixed reference date for relative date calculations. Values: `YYYY-MM-DD` or keywords.
`preferredDateFormat`	`"mm-dd-yyyy"`	Resolves ambiguous date input using the specified format. Options: `yyyy-mm-dd`, `yyyy-dd-mm`, `dd-mm-yyyy`, `mm-dd-yyyy`. Overridden by the user’s in-conversation preference.
`returnOnlyMonthYear`	true/false	Captures only month and year; ignores the day even if provided.

Examples:

{ "range": {"from": "2020-01-01", "to": "today"} }

Utterance: show schedule for 2019-02-03 → Prompts for value (out of range) Utterance: show schedule for 2020-02-03 → Extracted: 2020-02-03

{ "referenceDate": "2020-07-09" }

Utterance: schedule after two days → Extracted: 2020-07-11

{ "returnOnlyMonthYear": true }

Utterance: 03-04-2021 → Extracted: 04-2021

Date Period Rules

Rule	Values	Description
`range`	`{"from": "...", "to": "..."}`	Same as Date `range` rule.
`range` with `strict`	`{"from": "...", "to": "...", "strict": true}`	`true`: accepts only dates strictly within the range. `false`: accepts dates overlapping the range.
`referenceDate`	`"<date>"`	Sets a reference date for relative calculations.
`tense`	`"past"` or `"future"`	Forces year resolution when the year is absent. Without this rule, the platform uses the current year if the date falls within 90 days, otherwise the previous year.
`preferredDateFormat`	`"mm-dd-yyyy"`	Same as Date rule. Options: `yyyy-mm-dd`, `yyyy-dd-mm`, `ddmmyyyy`, `mmddyyyy`.

Date Time Rules

Rule	Values	Description
`range`	`{"from": "2020-01-01T00:00:00+05:30", "to": "..."}`	Accepts date-times only within the range. Values: date (`YYYY-MM-DD`), datetime (`YYYY-MM-DDTHH:MM:SS`), or keywords: `today`, `tomorrow`, `yesterday`, `now`.
`preferredTimes`	`{"from": "T12:00:00", "to": "T18:00:00"}`	Default preferred time range for all days. Resolves ambiguous times (for example, “3” → 3 PM if range is 9 AM–6 PM). Times in ISO 8601 format `THH:MM`.
`preferredTimes` (per day)	`{"from": ["", "T09:00", ...], "to": ["", "T18:00", ...]}`	Arrays of 7 values (Sun–Sat) for day-specific time preferences. Empty string = no preference for that day.
`preferredTimes` (favor)	`{"favor": "pm"}`	Sets preference by keyword: `future`, `past`, `am`, or `pm`.
`timeRangePossible`	true/false	If true, “10 to 4” resolves to two times (10:00 and 16:00) instead of 3:50.
`preferredDateFormat`	`"mm-dd-yyyy"`	Same as Date rule. Options: `yyyy-mm-dd`, `yyyy-dd-mm`, `ddmmyyyy`, `mmddyyyy`.

Time Rules

Rule	Values	Description
`preferredTimes`	Same as DateTime	Sets preferred time range or day-specific preferences.
`timeRangePossible`	true/false	Same as DateTime rule.
`range`	`{"from": "now"}` or `{"to": "now"}`	Restricts to times after or before now. Currently, only `"now"` is supported as a value.
`timeOnly`	true/false	`true`: returns time without “T” prefix or offset (for example, `18:00:00`). `false`: returns full ISO format (for example, `T18:00:00+5:30`). Supported in multilingual setups except English, German, French, and Spanish.

City Rules

Rule	Values	Description
`ignoreWords`	`"Send"`	Words not treated as city names. Value can be a string, space-separated concepts, or an array.

Example:

{ "ignoreWords": "Send" }

Utterance: Send destination to my email → Prompts for entity value (without rule: captures “Send” as a city)

Zip Code Rules

Rule	Values	Description
`preferredCountries`	`["GB"]`	Restricts zip code extraction to specified countries (alpha-2 codes) plus the user’s location country.

Example:

{ "preferredCountries": ["GB"] }

Utterance: check delivery to PO16 7GZ → Extracted: PO16 7GZ

Location Rules

Rule	Values	Description
`preferredCountries`	`["GB"]`	Restricts location extraction to specified countries (alpha-2 codes) plus the user’s location country.

List of Values (LoV) Rules

Rule	Values	Description
`ignoreLemmaWords`	true/false	`false` (default): lemmatization applied — “tickets” matches “ticket”. `true`: exact match only — “tickets” does not match “ticket”.
`ignoreWords`	`["~size"]`	Ignores specified words during LoV matching. Reduces false matches or ambiguity caused by common synonyms shared across multiple choices.

Examples:

{ "ignoreLemmaWords": true }

Input options: “ticket”, “tickets”

Input I need a ticket → Extracted: ticket
Input I need tickets → Error: not a valid option

{ "ignoreWords": ["~size"] }

Input options: “extra large size t-shirt”, “large size t-shirt”, “medium size t-shirt”

Input I need extra large size t-shirt → Extracted: extra large size t-shirt (ignores “size” during matching)

List of Items (enum) Rules

Rule	Values	Description
`ownership`	`"include"` or `"exclude"`	Detects ownership phrases (“is mine”, “belongs to me”) to include or exclude matched items from the entity value.
`includeWords`	`"great"` or `["word"]`	Additional words treated as ownership indicators. Use with `ownership: include`.
`excludeWords`	`"~lovConcept"` or `["word"]`	Additional words treated as non-ownership indicators. Use with `ownership: exclude`.
`noIndexMatch`	true	Disables selection by alphabetic or numeric index.

Examples: Items: pen, watch, bottle, book, cap

{ "ownership": "include" }

Utterance: first two are mine → Extracted: ["pen", "watch"]

{ "ownership": "exclude" }

Utterance: first two are mine → Extracted: ["bottle", "book", "cap"]

{ "ownership": "include", "includeWords": "great" }

Utterance: first two are great → Extracted: ["pen", "watch"]

{ "noIndexMatch": "true" }

Utterance: a → Prompts for input (without rule: extracts ["pen"])

Modules

Platform Services

Administration

References

​Add the Node

​Configure the Node

​Component Properties

​Instance Properties

​User Input

​User Input Correction (v7.3, String entities only)

​Entity Extraction

​Advanced Controls

​User Input Flow

​NLP Properties

​IVR Properties

​Connection Properties

​Entity Types

​Address

​Airport

​Attachment

​City

​Color

​Company

​Composite

​Country

​Currency

​Custom

​Date

​Date Period

​Date Time

​Description

​Email

​List of Items (enumerated)

​List of Items (lookup)

​Static List

​Remote List

​Location

​Number

​Person Name

​Percentage

​Phone Number

​Quantity

​String

​Time

​Time Zone

​URL

​Zip Code

​Entity Rules

​Generic Rules

​String and Description Rules

​Number Rules

​Currency Rules

​Composite Entity Rules (Model Number)

​Person Name Rules

​Company Rules

​Date Rules

​Date Period Rules

​Date Time Rules

​Time Rules

​City Rules

​Zip Code Rules

​Location Rules

​List of Values (LoV) Rules

​List of Items (enum) Rules

Add the Node

Configure the Node

Component Properties

Instance Properties

User Input

User Input Correction (v7.3, String entities only)

Entity Extraction

Advanced Controls

User Input Flow

NLP Properties

IVR Properties

Connection Properties

Entity Types

Address

Airport

Attachment

City

Color

Company

Composite

Country

Currency

Custom

Date

Date Period

Date Time

Description

Email

List of Items (enumerated)

List of Items (lookup)

Static List

Remote List

Location

Number

Person Name

Percentage

Phone Number

Quantity

String

Time

Time Zone

URL

Zip Code

Entity Rules

Generic Rules

String and Description Rules

Number Rules

Currency Rules

Composite Entity Rules (Model Number)

Person Name Rules

Company Rules

Date Rules

Date Period Rules

Date Time Rules

Time Rules

City Rules

Zip Code Rules

Location Rules

List of Values (LoV) Rules

List of Items (enum) Rules