Core

Performs various natural language tasks over raw text input.

Request

Response

Returns the output of various natural language tasks on the input text (see below for more information).

Example

Tokenization

Tokens appear under the tokens key for each sentence as the first element in the array.

Note that multi-word entities like the song “Let It Be” will be tokenized to be treated as a single word. To turn this off, set the ner parameter to false. In cases where it is ambiguous as to whether multiple words should be interpreted as a multi-word entity, “alternatives” for each sentence parse are returned in the response.

Parts of Speech

Parts of speech tags appear under the tokens key for each sentence as the second element in the array.

Lemmatization

Lemmatized or normalized tokens appear under the tokens key for each sentence as the third element in the array.

Normalization includes making the token lowercase, transforming verbs to their infinitive form, transforming adjectives to their positive degree, and transforming nouns to their singular form.

Syntax Dependency Trees

Syntax dependency trees appear under the dependencies key for each sentence.

The first element in the array is the index of the dependent word. The second element in the array is the index of the head word (or governor). The third element in the array is the name of the dependency relation.

Sentence Graph of Entities and Relationships

A graph of entities and relationships will appear under the graph key for each sentence. This task provides similar information to the relation extraction task or open information extraction task in NLP.

Named Entity Recognition

Named entities will appear under the graph key for each sentence under any entity objects under the ner key.

The ner object provides a label (label) for the recognized entity, along with identifiers for the Freebase entity (freebaseIdentifier) and the Cortex Knowledge Graph entity (id). An example of the ner object is shown below:

1{
2  "freebaseIdentifier": "/m/01hg45m",
3  "id": "PWE199585",
4  "label": "Let It Be",
5  "type": "concept"
6}

Disambiguation

Coreference Resolution

Sentiment Analysis