@expertai/nlapi v2.0.0

nlapi-nodejs

nlapi-nodejs is a NodeJS client for the expert.ai Natural Language API.
Natural Language API provides a lot of natural language understanding, document classification and information detection capabilities. You can therefore use this NodeJS client to add all those capabilities to NodeJS applications.
Useful links for Natural Language API:

• Live demo
• Documentation
• Developer portal with API's Swagger UI

This NodeJS client provides a class whose methods map to the resources of Natural Language API REST interface.
It has methods corresponding to text-processing resources and methods to access API self-documentation resources.

Installation

Install the client package with this command:

npm install @expertai/nlapi

Usage

Set credentials

You need a developer account to use expert.ai Natural Language API.
Go to the expert.ai developer portal and sign up.
Expert.ai Natural Language API can be used for free up to a monthly limit of 10M characters. If you want to use it above that limit, subscribe a payment plan from inside the expert.ai developer portal or from AWS Marketplace. Find more information on the product policies site.

Your developer account credentials must be specified as environment variables:

EAI_USERNAME=YOUR_USERNAME
EAI_PASSWORD=YOUR_PASSWORD

Replace YOUR_USERNAME with the email address you specified during registration on the developer portal and YOUR_PASSWORD with the account password.

Create the client

To use this client in your code, import the NLClient class:

// ES5 example
const { NLClient } = require("@expertai/nlapi")

// ES6+ example
import {NLClient} from "@expertai/nlapi"

Then create an instance of the client object:

let nlClient = new NLClient()

You can then invoke the object methods to use expert.ai Natural Language API.

To use Language enum type:

// ES5 example
const { Language } = require("@expertai/nlapi")

// ES6+ example
import { Language } from "@expertai/nlapi";

To use Analysis enum type:

// ES5 example
const { Analysis } = require("@expertai/nlapi")

// ES6+ example
import { Analysis } from "@expertai/nlapi";

Methods

NLClient class methods reproduce the behavior of Natural Language API endpoints and have corresponding parameters.

Full analysis

Use the analyze() method to perform all-inclusive (or "full") document analysis.

nlClient.analyze("Put your text here.", {
  language: Language.EN,
  context: "standard"
}).then((result) => {
  console.log(result.data)
})

Language is an enum-like object with all the possible language codes:

EN = 'en', // English
ES = 'es', // Spanish
FR = 'fr', // French
DE = 'de', // German
IT = 'it' // Italian

context is the name of the context to be used for the analysis. Available contexts can be listed with the context() method (see below).

The structure of the returned object is described here.

Specific/partial document analysis

Use the analyze() method with the analysis parameter to perform a specific/partial document analysis choosing from:

• Deep linguistic analysis
• Keyphrase extraction
• Named entity recognition
• Relation extraction
• Sentiment analysis

nlClient.analyze("Put your text here.", {
  language: Language.EN,
  context: "standard",
  analysis: Analysis.Relations
}).then((result) => {
  console.log(result.data)

Analysis is an enum-like object with the possible specific/partial analyses:

Disambiguation = "disambiguation", // Deep linguistic analysis
Relevants = "relevants", // Keyphrase extraction
Entities = "entities", // Named entity recognition
Sentiment = "sentiment", // Sentiment analysis
Relations = "relations"  // Relation extraction

The structure of the returned object is described in these pages:

• disambiguation analysis (deep linguistic analysis)
• relevants analysis (keyphrase extraction)
• entities analysis (named entity recognition)
• relations analysis (relation extraction)
• sentiment analysis (sentiment analysis)

Document classification

Use the categorize() method to perform document classification with one of the pre-defined taxonomies.

nlClient.categorize("test", {
  taxonomy: "iptc",
  language: Language.EN
}).then((result) => {
  console.log(result.data)
})

taxonomy is one of the Natural Language API taxonomies. Use the taxonomies() method (see below) to get the list of available taxonomies together with the language they support.
The structure of the returned object is described here.
In the case of geotax and emotional-traits taxonomies, the returned object has an additional extraData property. In the former case it contains GeoJSON, in the latter it contains the main groups of emotional traits.

Use the taxonomy() method (see below) to get the category tree for a given taxonomy. i.e. to know about all the possible output categories.

Information detection

Use the detect() method to perform information detection with one of the available detectors (for example: PII, Writeprint, Temporal information)

nlClient.detect("test", {
  language: Language.EN,
  detector: "pii"
}).then((result) => {
  console.log(result.data)
})

detector is one of the Natural Language API detectors. Use the detectors() method (see below) to get the list of available detectors together with the language they support.
Follow the links here to see the structure of the returned object for each detector.

API self-documentation: contexts

nlClient.context().then((result) => {
  console.log(result)
})

The method returns the list of available document analysis contexts. The returned object is described here.

API self-documentation: taxonomies

nlClient.taxonomies().then((result) => {
  console.log(result)
})

The method returns the list of available document classification taxonomies. The returned object is described here.

API self-documentation: category tree

nlClient.taxonomy({
  taxonomy: "iptc",
  language: Language.EN
}).then((result) => {
  console.log(result.data)
})

The method returns the category tree for a given taxonomy in a given language.
Use the taxonomies() method to get the list of available taxonomies together with the language they support.
The returned object is described here.

API self-documentation: detectors

nlClient.detectors().then((result) => {
  console.log(result)
})

The method returns the list of detectors that can be used for information detection. The returned object is described here.