Field types on INFINI Pizza

Full-text types

Mon, 01 Jan 0001 00:00:00 +0000

Full-text types #

`text` #

Field type to index full-text values.

Parameters for `text` fields #

analyzer
search_analyzer
realtime
index
norms
index_options
fields (multi-fields; a dense_vector sub-field declares schema-driven semantic embedding)
profile
source / column / store
null_value

Support indexing multiple values #

Yes

`keyword` #

Used for structured content such as IDs, email addresses, hostnames, status codes, zip codes, or tags.

Parameters for `keyword` fields #

realtime
index
norms
index_options
normalizer
fields (multi-fields; a dense_vector sub-field declares schema-driven semantic embedding)
profile
source / column / store
null_value

Support indexing multiple values #

Yes

Numeric and boolean types

Mon, 01 Jan 0001 00:00:00 +0000

Numeric and boolean types #

All numeric types and boolean accept the same parameter set; only the value range differs per type.

Type	Description
`tiny_int`	signed 8-bit integer
`small_int`	signed 16-bit integer
`integer`	signed 32-bit integer
`big_int`	signed 64-bit integer
`byte`	unsigned 8-bit integer
`unsigned_small_int`	unsigned 16-bit integer
`unsigned_integer`	unsigned 32-bit integer
`unsigned_big_int`	unsigned 64-bit integer
`float`	single precision floating-point number
`double`	double precision floating-point number
`boolean`	logical Boolean (`true`/`false`)

Parameters #

realtime
index
inplace — O(1) in-place column writes for hot counters and flags
fields (multi-fields)
profile
source / column / store
null_value

Support indexing multiple values #

Yes

Date and geo types

Mon, 01 Jan 0001 00:00:00 +0000

Date and geo types #

`date` #

A date-time value. Internally stored as epoch milliseconds; accepted input depends on the field’s format (default strict_date_optional_time||epoch_millis), so both ISO-8601 strings ("2024-06-15T10:30:00Z") and raw millisecond integers work. Used by date range, date histogram aggregations and sort.

Parameters for `date` fields #

Support indexing multiple values #

Yes

`geo_point` #

A latitude/longitude point ({"lat": 39.9, "lon": 116.4}), indexed for geo bounding box and geo distance queries. Distances use haversine on WGS-84 in metres.

Vector types

Mon, 01 Jan 0001 00:00:00 +0000

Vector types #

`semantic_text` #

The ES-style field type for semantic search (Elasticsearch 8.15+ alignment): declare one field, writes embed it server-side, and the semantic query searches it with plain text — no field names, no manual vectors.

PUT /my-collection
{
 "schema": {
 "properties": {
 "body": {
 "type": "semantic_text",
 "service": "zai",
 "model": "GLM-4.6V",
 "dims": 3072,
 "similarity": "cosine"
 }
 }
 }
}

Documents carry plain text ({"body": "some text"}); the vector derives at write time into the internal body.embedding sub-field. The field is pure semantic (ES semantic_text alignment): the text is kept in _source but is not inverted-indexed — match/term queries do not hit it; search goes through the derived vectors. A sibling regular text field plus bool.should composition gives hybrid retrieval.

Object and nested types

Mon, 01 Jan 0001 00:00:00 +0000

Object and nested types #

`object` #

JSON documents are hierarchical in nature: the document may contain inner objects which, in turn, may contain inner objects themselves. The object type introduces inner objects.

{
 "album": {
 "artists": ["Bill Danoff", "Taffy Nivert", "John Denver"]
 }
}

In the above JSON document, field “album” is of type object.

Parameters for `object` fields #

realtime — an object field’s realtime parameter would overwrite its sub-fields’ realtime parameter, even when they are explicitly set.
index — an object field’s index parameter would overwrite its sub-fields’ index parameter, even when they are explicitly set.
properties
profile
source / column / store

Support indexing multiple values #

Yes

Relation type

Mon, 01 Jan 0001 00:00:00 +0000

Relation type #

`relation` #

A graph edge value ({"from": ..., "to": ...}, by document id or key) that materializes adjacency for the traverse query and the graph projection block. The relation parameter declares the connected types.

Parameters for `relation` fields #

realtime
index
relation (from_type / to_type meta)

Support indexing multiple values #

Yes — a document may carry multiple edges.

Field types on INFINI Pizza

Full-text types

Full-text types #

text #

Parameters for text fields #

Support indexing multiple values #

keyword #

Parameters for keyword fields #

Support indexing multiple values #

Numeric and boolean types

Numeric and boolean types #

Parameters #

Support indexing multiple values #

Date and geo types

Date and geo types #

date #

Parameters for date fields #

Support indexing multiple values #

geo_point #

Vector types

Vector types #

semantic_text #

Object and nested types

Object and nested types #

object #

Parameters for object fields #

Support indexing multiple values #

Relation type

Relation type #

relation #

Parameters for relation fields #

Support indexing multiple values #

`text` #

Parameters for `text` fields #

`keyword` #

Parameters for `keyword` fields #

`date` #

Parameters for `date` fields #

`geo_point` #

`semantic_text` #

`object` #

Parameters for `object` fields #

`relation` #

Parameters for `relation` fields #