Rosette API reference

New to Corsair? See API access, authentication, and error handling.

Name	Type	Required	Description
`categories`	`object[] \| null`	No	—

Name	Type	Required	Description
`entities`	`object[] \| null`	No	Entities, most confident first.

{
  type?: string | null,
  mention?: string | null,
  normalized?: string | null,
  count?: number | null,
  mentionOffsets?: {
    startOffset?: number | null,
    endOffset?: number | null
  }[] | null,
  entityId?: string | null,
  confidence?: number | null,
  linkingConfidence?: number | null,
  salience?: number | null,
  dbpediaTypes?: string[] | null,
  permId?: string | null
}[] | null

Info

info

info Read the server name, version and build, which explains a change in analysis output Risk: read

await corsair.rosette.api.info({});

Input: empty object Output

Name	Type	Required	Description
`name`	`string \| null`	No	—
`version`	`string \| null`	No	—
`buildNumber`	`string \| null`	No	—
`buildTime`	`string \| null`	No	Such as 20260922013523.
`licenseExpiration`	`string \| null`	No	When the key’s license ends, where the server reports it.

Language

language

language Detect the language of a text: every candidate, most likely first, with its confidence Risk: read

await corsair.rosette.api.language({ content: 'The quick brown fox jumps over the lazy dog.' });

Input

Name	Type	Required	Description
`content`	`string`	No	The text to analyse. Mutually exclusive with contentUri.
`contentUri`	`string`	No	A URL for Rosette to fetch, over http, https, ftp or ftps. Mutually exclusive with content.
`language`	`string`	No	An ISO 639 language code, two or three letters, such as eng or spa.
`profileId`	`string`	No	A saved language profile to apply.

Output

Name	Type	Required	Description
`languageDetections`	`object[] \| null`	No	Every candidate, most likely first.
`regionalDetections`	`object[] \| null`	No	Regional variants, when the server returns them.

languageDetections full type

{
  language?: string | null,
  confidence?: number | null
}[] | null

regionalDetections full type

{
  language?: string | null,
  confidence?: number | null
}[] | null

Morphology

morphology

morphology Analyse morphology: lemmas, part-of-speech tags, compound components and Han readings Risk: read

await corsair.rosette.api.morphology({ content: 'running', language: 'eng' });

Input

Name	Type	Required	Description
`content`	`string`	No	The text to analyse. Mutually exclusive with contentUri.
`contentUri`	`string`	No	A URL for Rosette to fetch, over http, https, ftp or ftps. Mutually exclusive with content.
`language`	`string`	No	An ISO 639 language code, two or three letters, such as eng or spa.
`profileId`	`string`	No	A saved language profile to apply.

Output

Name	Type	Required	Description
`tokens`	`string[] \| null`	No	—
`lemmas`	`string[] \| null`	No	—
`posTags`	`string[] \| null`	No	—
`compoundComponents`	`(string[] \| null)[] \| null`	No	The parts of each compound token, null for a non-compound.
`hanReadings`	`(string[] \| null)[] \| null`	No	Readings of each Han token, null for other tokens.

Name Similarity

nameSimilarity

nameSimilarity Score how likely two names are to be the same entity; the order of the two affects the score Risk: read

await corsair.rosette.api.nameSimilarity({ name1: { text: 'John Smith', entityType: 'PERSON' }, name2: { text: 'Jon Smith', entityType: 'PERSON' } });

Input

Name	Type	Required	Description
`name1`	`object`	Yes	—
`name2`	`object`	Yes	—
`parameters`	`Record<string, string \| number \| boolean>`	No	Scoring overrides by name, such as { deletionScore: “0.0” }. An unknown parameter name gets a 500 from the API.

name1 full type

{
  text: string,
  language?: string,
  script?: string,
  entityType?: PERSON | LOCATION | ORGANIZATION
}

name2 full type

{
  text: string,
  language?: string,
  script?: string,
  entityType?: PERSON | LOCATION | ORGANIZATION
}

Output

Name	Type	Required	Description
`score`	`number \| null`	No	—

Name Translation

nameTranslation

nameTranslation Translate a name into another language, optionally saying which script and entity type it is Risk: read

await corsair.rosette.api.nameTranslation({ name: 'John Smith', targetLanguage: 'rus', entityType: 'PERSON' });

Input

Name	Type	Required	Description
`name`	`string`	Yes	The name to translate.
`targetLanguage`	`string`	Yes	The ISO 639 language to translate into.
`entityType`	`PERSON \| LOCATION \| ORGANIZATION`	No	Defaults to PERSON.
`sourceLanguageOfOrigin`	`string`	No	—
`sourceLanguageOfUse`	`string`	No	—
`sourceScript`	`string`	No	ISO 15924 code.
`targetScript`	`string`	No	ISO 15924 code.
`targetScheme`	`string`	No	The transliteration scheme for the result, such as IC or BGN; an unknown scheme gets a 400.

Output

Name	Type	Required	Description
`translation`	`string \| null`	No	—
`confidence`	`number \| null`	No	—
`targetLanguage`	`string \| null`	No	—
`targetScript`	`string \| null`	No	—
`targetScheme`	`string \| null`	No	—
`sourceScript`	`string \| null`	No	—
`sourceLanguageOfOrigin`	`string \| null`	No	—
`sourceLanguageOfUse`	`string \| null`	No	—
`extendedInformation`	`Record<string, JsonValue> \| null`	No	Extra server details, such as processedByBabel.

Ping

ping

ping Check that the host is reachable Risk: read

await corsair.rosette.api.ping({});

Input: empty object Output

Name	Type	Required	Description
`message`	`string \| null`	No	A greeting such as “Babel Street Analytics Server at your service”.
`time`	`number \| null`	No	Server time in epoch milliseconds.

Relationships

relationships

relationships Extract relationships between the entities in a text, such as a person employed by an organisation Risk: read

await corsair.rosette.api.relationships({ content: 'John Smith works for Acme Corp.' });

Input

Name	Type	Required	Description
`content`	`string`	No	The text to analyse. Mutually exclusive with contentUri.
`contentUri`	`string`	No	A URL for Rosette to fetch, over http, https, ftp or ftps. Mutually exclusive with content.
`language`	`string`	No	An ISO 639 language code, two or three letters, such as eng or spa.
`profileId`	`string`	No	A saved language profile to apply.

Output

Name	Type	Required	Description
`relationships`	`object[] \| null`	No	—

relationships full type

{
  predicate?: string | null,
  predicateId?: string | null,
  arg1?: string | null,
  arg1Id?: string | null,
  arg2?: string | null,
  arg2Id?: string | null,
  arg3?: string | null,
  arg3Id?: string | null,
  adjuncts?: string[] | null,
  locatives?: string[] | null,
  temporals?: string[] | null,
  modalities?: string[] | null,
  confidence?: number | null
}[] | null

Semantics

similarTerms

semantics.similarTerms Find terms near the input in embedding space, in each requested result language Risk: read

await corsair.rosette.api.semantics.similarTerms({ content: 'car', resultLanguages: ['eng'] });

Input

Name	Type	Required	Description
`content`	`string`	No	The term to find neighbours for. Mutually exclusive with contentUri.
`contentUri`	`string`	No	A URL for Rosette to fetch, over http, https, ftp or ftps. Mutually exclusive with content.
`language`	`string`	No	An ISO 639 language code, two or three letters, such as eng or spa.
`profileId`	`string`	No	A saved language profile to apply.
`resultLanguages`	`string[]`	Yes	The languages to return similar terms in, such as [“eng”, “spa”]. Required by the API.

Output

Name	Type	Required	Description
`similarTerms`	`object \| null`	No	Similar terms keyed by result language, closest first.

similarTerms full type

Record<string, {
  term?: string | null,
  similarity?: number | null
}[]> | null

vector

semantics.vector Compute a text embedding for the whole document Risk: read

await corsair.rosette.api.semantics.vector({ content: 'A sentence to embed.' });

Input

Name	Type	Required	Description
`content`	`string`	No	The text to analyse. Mutually exclusive with contentUri.
`contentUri`	`string`	No	A URL for Rosette to fetch, over http, https, ftp or ftps. Mutually exclusive with content.
`language`	`string`	No	An ISO 639 language code, two or three letters, such as eng or spa.
`profileId`	`string`	No	A saved language profile to apply.

Output

Name	Type	Required	Description
`documentEmbedding`	`number[] \| null`	No	The embedding vector for the whole text.
`tokens`	`string[] \| null`	No	—
`tokenEmbeddings`	`number[][] \| null`	No	A vector per token, when per-token output was requested.

Sentences

sentences

sentences Split a text into sentences, returned as text Risk: read

await corsair.rosette.api.sentences({ content: 'The quick brown fox. It jumped over the dog.' });

Input

Name	Type	Required	Description
`content`	`string`	No	The text to analyse. Mutually exclusive with contentUri.
`contentUri`	`string`	No	A URL for Rosette to fetch, over http, https, ftp or ftps. Mutually exclusive with content.
`language`	`string`	No	An ISO 639 language code, two or three letters, such as eng or spa.
`profileId`	`string`	No	A saved language profile to apply.

Output

Name	Type	Required	Description
`sentences`	`string[] \| null`	No	Each sentence as text, with its trailing whitespace.

Sentiment

sentiment

sentiment Score document and per-entity sentiment as a pos, neg or neu label with a confidence Risk: read

await corsair.rosette.api.sentiment({ content: 'I love this product, but the delivery was slow.' });

Input

Name	Type	Required	Description
`content`	`string`	No	The text to analyse. Mutually exclusive with contentUri.
`contentUri`	`string`	No	A URL for Rosette to fetch, over http, https, ftp or ftps. Mutually exclusive with content.
`language`	`string`	No	An ISO 639 language code, two or three letters, such as eng or spa.
`profileId`	`string`	No	A saved language profile to apply.

Output

Name	Type	Required	Description
`document`	`object \| null`	No	—
`entities`	`object[] \| null`	No	Sentiment per entity, when the analysis returned any.

document full type

{
  label?: string | null,
  confidence?: number | null
} | null

entities full type

{
  type?: string | null,
  mention?: string | null,
  normalized?: string | null,
  count?: number | null,
  mentionOffsets?: {
    startOffset?: number | null,
    endOffset?: number | null
  }[] | null,
  entityId?: string | null,
  confidence?: number | null,
  linkingConfidence?: number | null,
  salience?: number | null,
  dbpediaTypes?: string[] | null,
  permId?: string | null,
  sentiment?: {
    label?: string | null,
    confidence?: number | null
  } | null
}[] | null

Tokens

tokens

tokens Split a text into tokens; use morphology for part-of-speech tags and lemmas Risk: read

await corsair.rosette.api.tokens({ content: 'The quick brown fox', language: 'eng' });

Input

Name	Type	Required	Description
`content`	`string`	No	The text to analyse. Mutually exclusive with contentUri.
`contentUri`	`string`	No	A URL for Rosette to fetch, over http, https, ftp or ftps. Mutually exclusive with content.
`language`	`string`	No	An ISO 639 language code, two or three letters, such as eng or spa.
`profileId`	`string`	No	A saved language profile to apply.

Output

Name	Type	Required	Description
`tokens`	`string[] \| null`	No	The text split into tokens.

Topics

topics

topics Extract keyphrases and their broader concepts, with a salience rather than a probability Risk: read

await corsair.rosette.api.topics({ content: 'A new machine learning framework was released today.' });

Input

Name	Type	Required	Description
`content`	`string`	No	The text to analyse. Mutually exclusive with contentUri.
`contentUri`	`string`	No	A URL for Rosette to fetch, over http, https, ftp or ftps. Mutually exclusive with content.
`language`	`string`	No	An ISO 639 language code, two or three letters, such as eng or spa.
`profileId`	`string`	No	A saved language profile to apply.

Output

Name	Type	Required	Description
`keyphrases`	`object[] \| null`	No	—
`concepts`	`object[] \| null`	No	—

keyphrases full type

{
  phrase?: string | null,
  salience?: number | null
}[] | null

concepts full type

{
  phrase?: string | null,
  salience?: number | null,
  conceptId?: string | null
}[] | null

Transliteration

transliteration

transliteration Transliterate text between Latin and the native script of its language, which keeps the words rather than translating them Risk: read

await corsair.rosette.api.transliteration({ content: 'Mohammed Ali', language: 'ara' });

Input

Name	Type	Required	Description
`content`	`string`	No	The text to transliterate. Mutually exclusive with contentUri.
`contentUri`	`string`	No	A URL for Rosette to fetch, over http, https, ftp or ftps. Mutually exclusive with content.
`language`	`string`	No	The language of the text, such as ara; it sets which native script Latin text is moved into, or out of.

Output

Name	Type	Required	Description
`transliteration`	`string \| null`	No	—

Overview

Plugins

Rosette API reference

Categories

categories

Entities

entities

Info

info

Language

language

Morphology

morphology

Name Similarity

nameSimilarity

Name Translation

nameTranslation

Ping

ping

Relationships

relationships

Semantics

similarTerms

vector

Sentences

sentences

Sentiment

sentiment

Tokens

tokens

Topics

topics

Transliteration

transliteration

​Categories

​categories

​Entities

​entities

​Info

​info

​Language

​language

​Morphology

​morphology

​Name Similarity

​nameSimilarity

​Name Translation

​nameTranslation

​Ping

​ping

​Relationships

​relationships

​Semantics

​similarTerms

​vector

​Sentences

​sentences

​Sentiment

​sentiment

​Tokens

​tokens

​Topics

​topics

​Transliteration

​transliteration

Categories

categories

Entities

entities

Info

info

Language

language

Morphology

morphology

Name Similarity

nameSimilarity

Name Translation

nameTranslation

Ping

ping

Relationships

relationships

Semantics

similarTerms

vector

Sentences

sentences

Sentiment

sentiment

Tokens

tokens

Topics

topics

Transliteration

transliteration