@happyaccident/wink-sentiment
Version:
Accurate and fast sentiment scoring of phrases with emoticons :) & emojis 🎉
392 lines (359 loc) • 13.9 kB
JavaScript
// wink-sentiment
// Accurate and fast sentiment scoring of phrases with hashtags, emoticons & emojis.
//
// Copyright (C) 2017-18 GRAYPE Systems Private Limited
//
// This file is part of “wink-sentiment”.
//
// Permission is hereby granted, free of charge, to any person obtaining a
// copy of this software and associated documentation files (the "Software"),
// to deal in the Software without restriction, including without limitation
// the rights to use, copy, modify, merge, publish, distribute, sublicense,
// and/or sell copies of the Software, and to permit persons to whom the
// Software is furnished to do so, subject to the following conditions:
//
// The above copyright notice and this permission notice shall be included
// in all copies or substantial portions of the Software.
//
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
// OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
// THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
// FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
// DEALINGS IN THE SOFTWARE.
//
/* eslint-disable no-sync */
var chai = require( 'chai' );
var mocha = require( 'mocha' );
var ws = require( '../src/wink-sentiment.js' );
var fs = require( 'fs' );
var expect = chai.expect;
var describe = mocha.describe;
var it = mocha.it;
// Validate stem test cases given by Dr Martin F Porter for Porter Stemmer Algoritm V2.
describe( 'sentiment', function () {
it( 'should return a score of 0 with empty text', function () {
expect( ws( '' ) ).to.deep.equal( { score: 0, normalizedScore: 0 } );
expect( ws( ' ' ) ).to.deep.equal( { score: 0, normalizedScore: 0 } );
} );
it( 'should return a score of 4/2 with "I am feeling good"', function () {
expect( ws( 'I am feeling Good' ) ).to.deep.equal( {
score: 4,
normalizedScore: 2,
tokenizedPhrase: [
{ value: 'I', tag: 'word' },
{ value: 'am', tag: 'word' },
{ value: 'feeling', tag: 'word', score: 1 },
{ value: 'Good', tag: 'word', score: 3 }
]
} );
} );
it( 'should return a score of -3/-3 with "Not a good product"', function () {
// Case insensitive comparison at -2
expect( ws( 'Not a good product' ) ).to.deep.equal( {
score: -3,
normalizedScore: -3,
tokenizedPhrase: [
{ value: 'Not', tag: 'word' },
{ value: 'a', tag: 'word', negation: true },
{ value: 'good', tag: 'word', score: -3, negation: true },
{ value: 'product', tag: 'word' }
]
} );
} );
it( 'should return a score of -3/-3 with "Not good product"', function () {
// Case insensitive comparison at -1
expect( ws( 'Not good product' ) ).to.deep.equal( {
score: -3,
normalizedScore: -3,
tokenizedPhrase: [
{ value: 'Not', tag: 'word' },
{ value: 'good', tag: 'word', score: -3, negation: true },
{ value: 'product', tag: 'word', negation: true }
]
} );
} );
it( 'should return a score of 3/3 with "good product"', function () {
expect( ws( 'good product' ) ).to.deep.equal( {
score: 3,
normalizedScore: 3,
tokenizedPhrase: [
{ value: 'good', tag: 'word', score: 3 },
{ value: 'product', tag: 'word' }
]
} );
} );
it( 'should return a score of -2/-2 with "it was my bad luck"', function () {
// Test bi-gram config along with it's case insensitivity (**L**uck).
expect( ws( 'it was my bad Luck' ) ).to.deep.equal( {
score: -2,
normalizedScore: -2,
tokenizedPhrase: [
{ value: 'it', tag: 'word' },
{ value: 'was', tag: 'word' },
{ value: 'my', tag: 'word' },
{ value: 'bad', tag: 'word', score: -2, grouped: 1 },
{ value: 'Luck', tag: 'word' }
]
} );
} );
it( 'should return a score of 2/2 with "it was not my bad luck"', function () {
expect( ws( 'it was not my bad luck' ) ).to.deep.equal( {
score: 2,
normalizedScore: 2,
tokenizedPhrase: [
{ value: 'it', tag: 'word' },
{ value: 'was', tag: 'word' },
{ value: 'not', tag: 'word' },
{ value: 'my', tag: 'word', negation: true },
{ value: 'bad', tag: 'word', score: 2, negation: true, grouped: 1 },
{ value: 'luck', tag: 'word' }
]
} );
} );
it( 'should return a score of 6/3 with "love you <3"', function () {
expect( ws( 'love you <3' ) ).to.deep.equal( {
score: 6,
normalizedScore: 3,
tokenizedPhrase: [
{ value: 'love', tag: 'word', score: 3 },
{ value: 'you', tag: 'word' },
{ value: '<3', tag: 'emoticon', score: 3 }
]
} );
} );
it( 'should return a score of 6/3 with "love you<3"', function () {
expect( ws( 'love you<3' ) ).to.deep.equal( {
score: 6,
normalizedScore: 3,
tokenizedPhrase: [
{ value: 'love', tag: 'word', score: 3 },
{ value: 'you', tag: 'word' },
{ value: '<3', tag: 'emoticon', score: 3 }
]
} );
} );
it( 'should return a score of 8/2.6666666666666665 with "love you<3 :)"', function () {
expect( ws( 'love you<3 :)' ) ).to.deep.equal( {
score: 8,
normalizedScore: 2.6667,
tokenizedPhrase: [
{ value: 'love', tag: 'word', score: 3 },
{ value: 'you', tag: 'word' },
{ value: '<3', tag: 'emoticon', score: 3 },
{ value: ':)', tag: 'emoticon', score: 2 }
]
} );
} );
it( 'should return a score of 12/3 with "love you<3 😍😃"', function () {
expect( ws( 'love you<3 😍😃' ) ).to.deep.equal( {
score: 12,
normalizedScore: 3,
tokenizedPhrase: [
{ value: 'love', tag: 'word', score: 3 },
{ value: 'you', tag: 'word' },
{ value: '<3', tag: 'emoticon', score: 3 },
{ value: '😍', tag: 'emoji', score: 3 },
{ value: '😃', tag: 'emoji', score: 3 }
]
} );
} );
it( 'should return a score of 0/0 with "unknownword"', function () {
expect( ws( 'unknownword' ) ).to.deep.equal( {
score: 0,
normalizedScore: 0,
tokenizedPhrase: [
{ value: 'unknownword', tag: 'word' }
]
} );
} );
it( 'should return a score of 0/0 with unknown emoji "🚀"', function () {
expect( ws( '🚀' ) ).to.deep.equal( {
score: 0,
normalizedScore: 0,
tokenizedPhrase: [
{ value: '🚀', tag: 'emoji' }
]
} );
} );
it( 'should return a score of 0/0 with unknown emoji ";/"', function () {
expect( ws( ';/' ) ).to.deep.equal( {
score: 0,
normalizedScore: 0,
tokenizedPhrase: [
{ value: ';/', tag: 'emoticon' }
]
} );
} );
it( 'should return a handle a sentence with unknown emoji & emoticon', function () {
expect( ws( 'useless🚀 product;/' ) ).to.deep.equal( {
score: -2,
normalizedScore: -2,
tokenizedPhrase: [
{ value: 'useless', tag: 'word', score: -2 },
{ value: '🚀', tag: 'emoji' },
{ value: 'product', tag: 'word' },
{ value: ';/', tag: 'emoticon' }
]
} );
} );
it( 'should return a score of 12/3 with "#love you<3 😍😃 #unknown"', function () {
// This will ensure both known & unknown hashtags are tested.
expect( ws( '#love you<3 😍😃 #unknown' ) ).to.deep.equal( {
score: 12,
normalizedScore: 3,
tokenizedPhrase: [
{ value: '#love', tag: 'hashtag', score: 3 },
{ value: 'you', tag: 'word' },
{ value: '<3', tag: 'emoticon', score: 3 },
{ value: '😍', tag: 'emoji', score: 3 },
{ value: '😃', tag: 'emoji', score: 3 },
{ value: '#unknown', tag: 'hashtag' }
]
} );
} );
it( 'should return a score of 12/3 with "It was a #fail product"', function () {
// This will ensure both known & unknown hashtags are tested.
expect( ws( 'it was a #Fail product' ) ).to.deep.equal( {
score: -2,
normalizedScore: -2,
tokenizedPhrase: [
{ value: 'it', tag: 'word' },
{ value: 'was', tag: 'word' },
{ value: 'a', tag: 'word' },
{ value: '#Fail', tag: 'hashtag', score: -2 },
{ value: 'product', tag: 'word' }
]
} );
} );
it( 'should handle a phrase like cool stuff', function () {
// This will ensure both known & unknown hashtags are tested.
expect( ws( 'This a cool stuff!' ) ).to.deep.equal( {
score: 3,
normalizedScore: 3,
tokenizedPhrase: [
{ value: 'This', tag: 'word' },
{ value: 'a', tag: 'word' },
{ value: 'cool', tag: 'word', grouped: 1, score: 3 },
{ value: 'stuff', tag: 'word' },
{ value: '!', tag: 'punctuation' }
]
} );
} );
it( 'should handle phrase with negation', function () {
// This will ensure both known & unknown hashtags are tested.
expect( ws( 'Not so well done my boy! I am unhappy.' ) ).to.deep.equal( {
score: -5,
normalizedScore: -2.5,
tokenizedPhrase: [
{ value: 'Not', tag: 'word' },
{ value: 'so', tag: 'word', negation: true },
{ value: 'well', tag: 'word', grouped: 1, negation: true, score: -3 },
{ value: 'done', tag: 'word' },
{ value: 'my', tag: 'word' },
{ value: 'boy', tag: 'word' },
{ value: '!', tag: 'punctuation' },
{ value: 'I', tag: 'word' },
{ value: 'am', tag: 'word' },
{ value: 'unhappy', tag: 'word', score: -2 },
{ value: '.', tag: 'punctuation' }
]
} );
} );
it( 'should handle phrase without negation', function () {
// This will ensure both known & unknown hashtags are tested.
expect( ws( 'Sometimes I can be so short sighted.' ) ).to.deep.equal( {
score: -2,
normalizedScore: -2,
tokenizedPhrase: [
{ value: 'Sometimes', tag: 'word' },
{ value: 'I', tag: 'word' },
{ value: 'can', tag: 'word' },
{ value: 'be', tag: 'word' },
{ value: 'so', tag: 'word' },
{ value: 'short', tag: 'word', grouped: 1, score: -2 },
{ value: 'sighted', tag: 'word' },
{ value: '.', tag: 'punctuation' }
]
} );
} );
it( 'should return a score of 5/1.9764 with >15 words sentence', function () {
// This will trigger condition when # words > 15 (average sentence length).
expect( ws( 'Sound quality on both end is excellent, I use headset to call my wife and ask my wife to use headset to call me!' ) ).to.deep.equal( {
score: 5,
normalizedScore: 1.9764,
tokenizedPhrase: [
{ value: 'Sound', tag: 'word' },
{ value: 'quality', tag: 'word', score: 2 },
{ value: 'on', tag: 'word' },
{ value: 'both', tag: 'word' },
{ value: 'end', tag: 'word' },
{ value: 'is', tag: 'word' },
{ value: 'excellent', tag: 'word', score: 3 },
{ value: ',', tag: 'punctuation' },
{ value: 'I', tag: 'word' },
{ value: 'use', tag: 'word' },
{ value: 'headset', tag: 'word' },
{ value: 'to', tag: 'word' },
{ value: 'call', tag: 'word' },
{ value: 'my', tag: 'word' },
{ value: 'wife', tag: 'word' },
{ value: 'and', tag: 'word' },
{ value: 'ask', tag: 'word' },
{ value: 'my', tag: 'word' },
{ value: 'wife', tag: 'word' },
{ value: 'to', tag: 'word' },
{ value: 'use', tag: 'word' },
{ value: 'headset', tag: 'word' },
{ value: 'to', tag: 'word' },
{ value: 'call', tag: 'word' },
{ value: 'me', tag: 'word' },
{ value: '!', tag: 'punctuation' }
]
} );
} );
it( 'should throw error with undefined input', function () {
expect( ws.bind( null ) ).to.throw( 'wink-sentiment: input phrase must be a string, instead found: undefined' );
} );
it( 'should throw error with non-string input', function () {
expect( ws.bind( null, 10 ) ).to.throw( 'wink-sentiment: input phrase must be a string, instead found: number' );
} );
} );
describe( 'validate with amazon product review data from UCI', function () {
// To be on safe side with travis!
this.timeout(3600);
// Load Amazon Product Review [Sentiment Labelled Sentences Data Set](https://archive.ics.uci.edu/ml/machine-learning-databases/00331/)
// at [UCI Machine Learning Repository](https://archive.ics.uci.edu/ml/index.php).
var input = fs.readFileSync( './test/data/amazon_cells_labelled.txt', 'utf8' ).split( '\n' );
input.pop();
var fn = 0,
fp = 0,
tn = 0,
tp = 0;
input.forEach( function ( row ) {
var cols = row.split( '\t' );
if ( +cols[ 1 ] === 1 ) {
if ( ws( cols[0] ).score >= 0 ) tp += 1;
if ( ws( cols[0] ).score < 0 ) fn += 1;
} else {
if ( ws( cols[0] ).score < 0 ) tn += 1;
if ( ws( cols[0] ).score >= 0 ) fp += 1;
}
} );
var precision = tp / ( tp + fp );
var recall = tp / ( tp + fn );
// Reduce verbosity of test output by moving `it` outside the `forEach`.
it( 'it should achieve an accuracy of 77%', function () {
expect( Math.round( ( tp + tn ) * 100 / ( tp + tn + fp + fn ) ) ).to.be.at.least( 77 );
} );
it( 'it should achieve an recall of 98%', function () {
expect( Math.round( recall * 100 ) ).to.equal( 98 );
} );
it( 'it should achieve an precision of 69%', function () {
expect( Math.round( precision * 100 ) ).to.equal( 69 );
} );
it( 'it should achieve an f-measure of 0.81', function () {
expect( +( 2 * precision * recall / ( precision + recall ) ).toFixed( 2 ) ).to.equal( 0.81 );
} );
} );