Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 23 |
|
0.00% |
0 / 2 |
CRAP | |
0.00% |
0 / 1 |
| UnderlinkedFunctionScoreBuilder | |
0.00% |
0 / 23 |
|
0.00% |
0 / 2 |
6 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| append | |
0.00% |
0 / 21 |
|
0.00% |
0 / 1 |
2 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace GrowthExperiments\NewcomerTasks\TaskSuggester; |
| 4 | |
| 5 | use CirrusSearch\Search\Rescore\BoostFunctionBuilder; |
| 6 | use Elastica\Query\FunctionScore; |
| 7 | use Elastica\Script\Script; |
| 8 | |
| 9 | /** |
| 10 | * A CirrusSearch rescore function which prioritizes underlinked articles and is otherwise random. |
| 11 | * @see https://www.mediawiki.org/wiki/Extension:CirrusSearch/Scoring#Rescoring |
| 12 | */ |
| 13 | class UnderlinkedFunctionScoreBuilder implements BoostFunctionBuilder { |
| 14 | |
| 15 | /** Function type used in the rescore profile */ |
| 16 | public const TYPE = 'growth_underlinked'; |
| 17 | |
| 18 | /** @var float */ |
| 19 | private $weight; |
| 20 | |
| 21 | /** @var int */ |
| 22 | private $minimumLength; |
| 23 | |
| 24 | /** |
| 25 | * @param float $weight Weight of the underlinkedness metric (vs. a random factor) in sorting. |
| 26 | * @param int $minimumLength Do not consider articles shorter than this underlinked. |
| 27 | */ |
| 28 | public function __construct( float $weight, int $minimumLength ) { |
| 29 | $this->weight = $weight; |
| 30 | $this->minimumLength = $minimumLength; |
| 31 | } |
| 32 | |
| 33 | /** |
| 34 | * @inheritDoc |
| 35 | */ |
| 36 | public function append( FunctionScore $container ) { |
| 37 | // For articles shorter than the minimum length, the underlinkedness score is 0. |
| 38 | // Otherwise, it is the chance that a randomly picked word in the |
| 39 | // article is not a link. (Approximately - doesn't take multi-word links into account.) |
| 40 | // Since this is very close to 1, a power function is used to smooth it to the [0,1] range. |
| 41 | // See https://phabricator.wikimedia.org/T317546#8246903 for why .length is used |
| 42 | |
| 43 | // Chosen arbitrarily because it gave nice values for a few sample articles. |
| 44 | $smoothingFactor = 4; |
| 45 | |
| 46 | // Keep this in sync with tests/api-testing/underlinked-rescore.js |
| 47 | $script = /** @lang JavaScript */ |
| 48 | <<<'SCRIPT' |
| 49 | doc['text_bytes'] >= minimumLength |
| 50 | ? pow( |
| 51 | max( |
| 52 | 0, |
| 53 | 1 - ( |
| 54 | doc['outgoing_link.token_count'].length |
| 55 | / max( 1, doc['text.word_count'] ) |
| 56 | ) |
| 57 | ), |
| 58 | smoothingFactor |
| 59 | ) |
| 60 | : 0 |
| 61 | SCRIPT; |
| 62 | $script = trim( preg_replace( '/\s+/', ' ', $script ) ); |
| 63 | $params = [ |
| 64 | 'minimumLength' => $this->minimumLength, |
| 65 | 'smoothingFactor' => $smoothingFactor, |
| 66 | ]; |
| 67 | $container->addScriptScoreFunction( |
| 68 | new Script( $script, $params, Script::LANG_EXPRESSION ), |
| 69 | null, |
| 70 | $this->weight |
| 71 | ); |
| 72 | // Mix with a random factor so everyone doesn't get the same list of articles. |
| 73 | $container->addRandomScoreFunction( |
| 74 | random_int( 1, PHP_INT_MAX ), |
| 75 | null, |
| 76 | 1 - $this->weight, |
| 77 | '_seq_no' |
| 78 | ); |
| 79 | $container->setScoreMode( FunctionScore::SCORE_MODE_SUM ); |
| 80 | } |
| 81 | |
| 82 | } |