sentiment-parser

Sentiment Parser

Description

The sentiment-parser package is used to easily parse sentiments from sentences, using a simple one-liner.

Backend

The package currently uses both an offline package and the IBM Watson API to gather the results. It will also try to guess a best answer, based on the two differing results.

Prepare

You will need to secure credentials from IBM. To do that:

• Sign up for IBM Cloud.
• Once you have an accound - go to the Natural Language Understanding service page.
• Click the "Create" button.
• Click the "Service Credentials" tab on the Natural Language Understanding page in your IBM Cloud dashboard to view your credentials.
• Do the same for the ToneAnalyzer service

Initialization

const { Parser } = require('sentiment-parser');
const parser = new Parser();

    parser.setCredentials({
        nlAnalyzer: {
            username: '{your-api-username}',
            password: '{your-api-password}',
        },
        toneAnalyzer: {
            username: '{your-api-username}',
            password: '{your-api-password}',
        }
    });

Usage

parser.parseSentiment(userInput).then(response => {
    // response.score is a value between -1 to 1
    // response.label is "negative" or "positive"
}



parser.parseEmotion(`I hate this`).then(parsedTone => {
    // parsedTone.watsonTone.strongestTone contains the strongest tone
    // parsedTone.watsoneTone.allTones contains an array of all the recognized tones

    // Each tone has a tone_id, tone_name and a score (from 0 to 1)
}