|
| 1 | +<?php |
| 2 | + |
| 3 | +namespace Tests\TextAnalysis\NGrams; |
| 4 | + |
| 5 | +use TextAnalysis\NGrams\StatisticFacade; |
| 6 | +use TextAnalysis\Tokenizers\RegexTokenizer; |
| 7 | +use TextAnalysis\NGrams\NGramFactory; |
| 8 | + |
| 9 | +/** |
| 10 | + * Description of NGramFactoryTest |
| 11 | + * |
| 12 | + * @author yooper <yooper> |
| 13 | + */ |
| 14 | +class StatisticFacadeTest extends \PHPUnit_Framework_TestCase |
| 15 | +{ |
| 16 | + private $text; |
| 17 | + private $tokens; |
| 18 | + |
| 19 | + public function __construct() { |
| 20 | + $this->text = file_get_contents(TEST_DATA_DIR . DS . 'Text'.DS.'Analysis'.DS.'text_ngrams.txt'); |
| 21 | + |
| 22 | + $tokenizer = new RegexTokenizer('/([\p{L}]+[\/\-_\']?[\p{L}]+)+|[\p{L}]+/iu'); |
| 23 | + $this->tokens = normalize_tokens($tokenizer->tokenize($this->text)); |
| 24 | + } |
| 25 | + |
| 26 | + public function testBigrams() |
| 27 | + { |
| 28 | + $ngrams = NGramFactory::create($this->tokens, 2, '<>'); |
| 29 | + $ngrams = NGramFactory::getFreq($ngrams, '<>'); |
| 30 | + |
| 31 | + //test frequency |
| 32 | + $this->assertEquals($ngrams['know<>something'], array( 0=>2, 1=> 3, 2 => 2)); |
| 33 | + |
| 34 | + //test tmi measure |
| 35 | + $ngrams_stats = StatisticFacade::calculate($ngrams, 'tmi', 2); |
| 36 | + $this->assertEquals(round($ngrams_stats['know<>something'], 4), 0.1612); |
| 37 | + |
| 38 | + //test ll measure |
| 39 | + $ngrams_stats = StatisticFacade::calculate($ngrams, 'll', 2); |
| 40 | + $this->assertEquals(round($ngrams_stats['know<>something'], 4), 13.8516); |
| 41 | + |
| 42 | + //test pmi measure |
| 43 | + $ngrams_stats = StatisticFacade::calculate($ngrams, 'pmi', 2); |
| 44 | + $this->assertEquals(round($ngrams_stats['know<>something'], 4), 4.3692); |
| 45 | + |
| 46 | + //test dice measure |
| 47 | + $ngrams_stats = StatisticFacade::calculate($ngrams, 'dice', 2); |
| 48 | + $this->assertEquals(round($ngrams_stats['know<>something'], 4), 0.8000); |
| 49 | + |
| 50 | + //test x2 measure |
| 51 | + $ngrams_stats = StatisticFacade::calculate($ngrams, 'x2', 2); |
| 52 | + $this->assertEquals(round($ngrams_stats['know<>something'], 4), 40.6444); |
| 53 | + |
| 54 | + //test tscore measure |
| 55 | + $ngrams_stats = StatisticFacade::calculate($ngrams, 'tscore', 2); |
| 56 | + $this->assertEquals(round($ngrams_stats['know<>something'], 4), 1.3458); |
| 57 | + |
| 58 | + //test phi measure |
| 59 | + $ngrams_stats = StatisticFacade::calculate($ngrams, 'phi', 2); |
| 60 | + $this->assertEquals(round($ngrams_stats['know<>something'], 4), 0.6556); |
| 61 | + |
| 62 | + //test odds measure |
| 63 | + $ngrams_stats = StatisticFacade::calculate($ngrams, 'odds', 2); |
| 64 | + $this->assertEquals(round($ngrams_stats['know<>something'], 4), 118.0000); |
| 65 | + |
| 66 | + //test leftFisher measure |
| 67 | + $ngrams_stats = StatisticFacade::calculate($ngrams, 'leftFisher', 2); |
| 68 | + $this->assertEquals(round($ngrams_stats['know<>something'], 4), 1.0000); |
| 69 | + |
| 70 | + //test rightFisher measure |
| 71 | + $ngrams_stats = StatisticFacade::calculate($ngrams, 'rightFisher', 2); |
| 72 | + $this->assertEquals(round($ngrams_stats['know<>something'], 4), 0.0016); |
| 73 | + } |
| 74 | + |
| 75 | + public function testTrigrams() |
| 76 | + { |
| 77 | + $ngrams = NGramFactory::create($this->tokens, 3, '<>'); |
| 78 | + $ngrams = NGramFactory::getFreq($ngrams, '<>'); |
| 79 | + |
| 80 | + //test frequency |
| 81 | + $this->assertEquals($ngrams['the<>know<>something'], array( 0 => 1, 1 => 4, 2 => 3, 3 => 2, 4 => 1, 5 => 1, 6 => 2)); |
| 82 | + |
| 83 | + //test tmi measure |
| 84 | + $ngrams_stats = StatisticFacade::calculate($ngrams, 'tmi', 3); |
| 85 | + $this->assertEquals(round($ngrams_stats['the<>know<>something'], 4), 0.2002); |
| 86 | + |
| 87 | + //test ll measure |
| 88 | + $ngrams_stats = StatisticFacade::calculate($ngrams, 'll', 3); |
| 89 | + $this->assertEquals(round($ngrams_stats['the<>know<>something'], 4), 16.9283); |
| 90 | + } |
| 91 | +} |
0 commit comments