| 1: | <?php declare(strict_types=1); |
| 2: | |
| 3: | namespace Salient\Contract\Core; |
| 4: | |
| 5: | /** |
| 6: | * Text comparison algorithms |
| 7: | * |
| 8: | * Text comparison algorithms take two strings and calculate a value between `0` |
| 9: | * and `1`, where `0` indicates the strings could not be more similar, and `1` |
| 10: | * indicates they could not be more different. |
| 11: | */ |
| 12: | interface TextComparisonAlgorithm |
| 13: | { |
| 14: | /** |
| 15: | * Uncertainty is 0 if values are identical, otherwise 1 |
| 16: | */ |
| 17: | public const SAME = 1; |
| 18: | |
| 19: | /** |
| 20: | * Uncertainty is 0 if the longer value contains the shorter value, |
| 21: | * otherwise 1 |
| 22: | */ |
| 23: | public const CONTAINS = 2; |
| 24: | |
| 25: | /** |
| 26: | * Uncertainty is derived from levenshtein() |
| 27: | * |
| 28: | * String length cannot exceed 255 characters. |
| 29: | * |
| 30: | * @see levenshtein() |
| 31: | */ |
| 32: | public const LEVENSHTEIN = 4; |
| 33: | |
| 34: | /** |
| 35: | * Uncertainty is derived from similar_text() |
| 36: | * |
| 37: | * @see similar_text() |
| 38: | */ |
| 39: | public const SIMILAR_TEXT = 8; |
| 40: | |
| 41: | /** |
| 42: | * Uncertainty is derived from shared ngrams relative to the longest string |
| 43: | */ |
| 44: | public const NGRAM_SIMILARITY = 16; |
| 45: | |
| 46: | /** |
| 47: | * Uncertainty is derived from shared ngrams relative to the shortest string |
| 48: | */ |
| 49: | public const NGRAM_INTERSECTION = 32; |
| 50: | } |
| 51: |