Search index data types - Tablestore - Alibaba Cloud Documentation Center

This topic describes the mappings between field data types in data tables and their corresponding types in search indexes. It also explains the additional properties and query features supported by different field data types.

Data type mappings

The value of a field in a search index is derived from the value of the corresponding field in the data table. The data types of these fields must match. The following table describes the mappings between field data types in search indexes and data tables.

Important

The data types in the table must have a one-to-one mapping. Geo-point and Nested types also require specific formats. If these requirements are not met, the data is discarded as dirty data. This can cause data to exist in the table but not be found in the search index.

Field data type in search indexes	Field data type in data tables	Description
Long	Integer	A 64-bit long integer.
Double	Double	A 64-bit double-precision floating-point number.
Boolean	Boolean	A Boolean value.
Keyword	String	A string that cannot be tokenized.
FuzzyKeyword	String	A string that supports high-performance fuzzy queries.
Text	String	A string or text that can be tokenized. For more information, see String types.
Date	Integer, String	The Date data type supports various custom formats for date data.
IP	String	The IP type supports IP addresses in IPv4 and IPv6 formats.
Geo-point	String	The coordinate information of a point. The format is `latitude,longitude`, with latitude first, then longitude. The latitude must be in the range of [-90, +90], and the longitude must be in the range of [-180, +180]. For example, `35.8,-45.91`.
Vector	String, Binary	The vector type. The value is a string in the format of a Float32 array. The length of the array is equal to the dimension of the field. For example, the vector string `[1, 5.1, 4.7, 0.08 ]` has a dimension of 4.
Nested	String	The nested type. For example, `[{"a": 1}, {"a": 3}]`.
JSON type	String	The JSON type. It supports the OBJECT and NESTED types.

Field attribute support

Search index fields support additional properties, such as array, virtual column, and highlighting. The supported properties vary based on the data type. The following table lists the applicable data types and provides a description for each additional property.

Property	Applicable data types	Description
Array	Long, Double, Boolean, Keyword, Text, Date, IP, and Geo-point	To store a series of data of the same type, set the field to the array type. When you write data, it must be in the JSON array format, such as `["a","b","c"]`. Nested, Vector, and JSON types are arrays by nature, so you do not need to set this property.
Virtual column	Long, Double, Keyword, FuzzyKeyword, Text, Date, IP, Geo-point, and Vector	If you want to query new fields with new types without changing the storage structure and data in Tablestore, set the field as a virtual column.
Date format	Date	When you use the Date type, you must specify the date format.
Tokenization	Text	To implement full-text search, configure tokenization for the field.
Summary and highlighting	Text	To highlight hit search queries in full-text searches, enable the Summary and Highlighting feature for the field.
Vector configuration	Vector	When you use a Vector field, you must specify the vector's measure algorithm and dimension.
JSON type configuration	JSON	When using a JSON field, specify the JSON type. The Object and Nested types are supported.

Query feature support

The following table describes the query features available for each data type.

Note

"✓" indicates that the feature is supported. "×" indicates that the feature is not supported.
The match all query feature does not require you to set fields.

Query feature

Long

Double

Boolean

Keyword

FuzzyKeyword

Text

Date

Geo-point

JSON Object

Nested/JSON Nested

Vector

✓

✓

✓

✓

✓

✓

✓

Wildcard query based on tokenization

✓

Geo query

✓

Nested type query

✓

Collapse (deduplicate)

✓

✓

✓

✓

✓