For the complete documentation index, see llms.txt. This page is also available as Markdown.

Field Definitions

Defining which fields should appear in the output and whether and how they need to be used in matching.

fieldDefinition

This is a JSON array representing the fields from the source data to be used for matching, and the kind of matching they need.

Each field denotes a column from the input. Fields have the following JSON attributes:

fieldName

The name of the field from the input data schema.

fields

To be defined later. For now, please keep this as the fieldName

dataType

Type of the column - string, integer, double, etc.

matchType

  • The way to match the given field. Multiple match types can be combined (comma-separated), for example: FUZZY,NUMERIC.

  • Here are the supported match types and descriptions:

Match Type
Description
Applicable To

FUZZY

Broad matches tolerant to typos, abbreviations, and other variations. Uses fuzzy string similarity features.

string, integer, long, double, date

FUZZY_OPTIMISED

Same semantics as FUZZY but uses an optimized implementation. Provides similar matching quality with significantly lower CPU and memory usage on large datasets. Recommended for production/large-scale matching when FUZZY quality is desired but performance is critical. Zingg Enterprise Feature

string, integer, long, double, date

EXACT

Exact match with no tolerance for variation. Preferable for country codes, pin codes, and other categorical variables where you expect no variations.

string, integer, long, date, boolean

DONT_USE

Included in the output but no computation is done on this field. Useful for IDs required in the output. DONT_USE fields are not shown to the labeler when showConcise is set to true.

any

EMAIL

Matches only the local part (before the @) of email addresses.

string

EMAIL_OPTIMISED

Same semantics as EMAIL but uses an optimized implementation for much faster evaluation on large datasets while preserving match behavior. Recommended for large-scale runs where many emails are compared. Zingg Enterprise Feature

string

PINCODE

Matches postal / pin codes (supports typical local formats such as xxxxx or xxxxx-xxxx depending on data).

string

NULL_OR_BLANK

By default, Zingg treats nulls as matches. If NULL_OR_BLANK is added to a field that also has other match types (e.g., FUZZY), Zingg will build an explicit feature for null/blank values so the model can learn their effect.

string, integer, long, date, boolean

TEXT

Compares overlapping words between two strings. Good for descriptive or long-text fields without many typos.

string

NUMERIC

Extracts numbers from strings and compares how many are the same across both strings (useful for apartment numbers, building numbers, etc.).

string

NUMERIC_WITH_UNITS

Extracts product codes or numbers with units (e.g., 16gb) and compares how many are same across both strings

string

ONLY_ALPHABETS_EXACT

Compares only the alphabetic characters and requires an exact match. Useful when numeric parts should be ignored (e.g., building name vs flat number).

string

ONLY_ALPHABETS_FUZZY

Compares only the alphabetic characters using a fuzzy comparison; numeric characters are ignored. Useful for addresses when you want to handle street names fuzzily while numeric parts are handled separately (e.g., via NUMERIC).

string

ONLY_ALPHABETS_FUZZY_OPTIMISED

Same semantics as ONLY_ALPHABETS_FUZZY but uses optimized processing. Use this when you need the fuzzy alphabet-only behavior at production scale. Zingg Enterprise Feature

string

MAPPING_(FILENAME)

Maps input values to canonical values using a mapping file (e.g., nickname maps, company abbreviations, gender codes). Matching is tolerant to common variations defined in the mapping. See Advanced Match Types for mapping file format and examples.

string

Last updated

Was this helpful?