fix: reduce false positives in award detection
Filter SPARQL query to only return entertainment awards (film, TV, music, theater) and add a canonical award map to normalize variants (e.g. all Oscar/Academy Award → "Oscar", all Golden Globe → "Golden Globe"). Non-entertainment awards (orders, medals, honorary degrees) are excluded both at SPARQL level and via PHP keyword filter. Also restart messenger container on cache:clear to avoid stale DI container errors. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
thibaud-leclere
@@ -91,6 +91,18 @@ class WikidataGateway
|
||||
?awardStatement ps:P166 ?award .
|
||||
?awardStatement pq:P585 ?date .
|
||||
BIND(YEAR(?date) AS ?year)
|
||||
|
||||
# Only keep entertainment awards (film, TV, music, theater, performing arts)
|
||||
VALUES ?awardSuperclass {
|
||||
wd:Q4220920 # film award
|
||||
wd:Q1407443 # television award
|
||||
wd:Q2235858 # music award
|
||||
wd:Q15056993 # film festival award
|
||||
wd:Q15383322 # theater award
|
||||
wd:Q29461289 # performing arts award
|
||||
}
|
||||
?award wdt:P31/wdt:P279* ?awardSuperclass .
|
||||
|
||||
SERVICE wikibase:label { bd:serviceParam wikibase:language "fr,en" . }
|
||||
}
|
||||
ORDER BY ?name DESC(?year)
|
||||
|
||||
@@ -14,6 +14,76 @@ use Psr\Log\LoggerInterface;
|
||||
|
||||
readonly class AwardImporter
|
||||
{
|
||||
/**
|
||||
* Canonical award name => keywords to match (case-insensitive).
|
||||
* Checked in order — first match wins.
|
||||
*/
|
||||
private const AWARD_MAP = [
|
||||
'Oscar' => ['Academy Award', 'Oscar'],
|
||||
'Golden Globe' => ['Golden Globe', 'Golden Globes'],
|
||||
'BAFTA' => ['BAFTA', 'British Academy Film Award', 'British Academy Television Award', 'British Academy Games Award'],
|
||||
'César' => ['César'],
|
||||
'SAG' => ['Screen Actors Guild'],
|
||||
'Emmy' => ['Emmy Award', 'Primetime Emmy'],
|
||||
'Tony' => ['Tony Award', 'Tony award'],
|
||||
'Grammy' => ['Grammy'],
|
||||
'Cannes' => ['Festival de Cannes', 'Cannes', "Palme d'or", "Caméra d'or"],
|
||||
'Sundance' => ['Sundance'],
|
||||
'Berlinale' => ['Berlinale', 'Berliner Bär', "Ours d'argent", "Ours d'or"],
|
||||
'Mostra de Venise' => ['Mostra', 'Venice Film Festival', 'Coupe Volpi', "Lion d'or"],
|
||||
'Saturn' => ['Saturn Award'],
|
||||
'MTV' => ['MTV Movie', 'MTV Video'],
|
||||
"Critics' Choice" => ["Critics' Choice"],
|
||||
'Independent Spirit' => ['Independent Spirit'],
|
||||
'Annie' => ['Annie Award'],
|
||||
'Goya' => ['prix Goya', 'Goya Award'],
|
||||
'Laurence Olivier' => ['Laurence Olivier', 'Olivier Award'],
|
||||
'David di Donatello' => ['David di Donatello'],
|
||||
'Gotham' => ['Gotham Award', 'Gotham Independent'],
|
||||
'NAACP Image' => ['NAACP Image'],
|
||||
"People's Choice" => ["People's Choice"],
|
||||
'Teen Choice' => ['Teen Choice'],
|
||||
'BET' => ['BET Award', 'BET Her', 'BET YoungStars'],
|
||||
'Black Reel' => ['Black Reel'],
|
||||
'National Board of Review' => ['National Board of Review'],
|
||||
'New York Film Critics Circle' => ['New York Film Critics Circle'],
|
||||
'Los Angeles Film Critics' => ['Los Angeles Film Critics'],
|
||||
'San Sebastián' => ['Donostia', 'San Sebastián'],
|
||||
'Sitges' => ['Sitges'],
|
||||
'Satellite' => ['Satellite Award'],
|
||||
'Lucille Lortel' => ['Lucille Lortel'],
|
||||
'Golden Raspberry' => ['Golden Raspberry', 'Razzie'],
|
||||
'Drama Desk' => ['Drama Desk'],
|
||||
'Genie' => ['Genie Award'],
|
||||
'European Film Award' => ['prix du cinéma européen', 'European Film Award'],
|
||||
'AACTA' => ['AACTA'],
|
||||
];
|
||||
|
||||
/**
|
||||
* Keywords indicating non-entertainment awards (case-insensitive).
|
||||
* These slip through even with the SPARQL filter.
|
||||
*/
|
||||
private const EXCLUDED_KEYWORDS = [
|
||||
// National orders and decorations
|
||||
'chevalier', 'officier', 'commandeur', 'compagnon',
|
||||
'ordre du', 'ordre de', 'order of the',
|
||||
'grand-croix', 'grand officier', 'grand cordon',
|
||||
'Knight Bachelor', 'Knight Commander',
|
||||
'croix d\'',
|
||||
// Honorary degrees and memberships
|
||||
'honoris causa',
|
||||
'membre de l\'', 'membre de la', 'membre honoraire', 'membre associé', 'membre élu',
|
||||
'Fellow of', 'fellow de',
|
||||
// Scholarships
|
||||
'bourse ',
|
||||
// Medals (military, scientific, etc.)
|
||||
'médaille', 'Medal',
|
||||
// Other non-entertainment
|
||||
'Time 100', '100 Women', 'All-NBA',
|
||||
'étoile du Hollywood Walk of Fame',
|
||||
'allée des célébrités',
|
||||
];
|
||||
|
||||
public function __construct(
|
||||
private WikidataGateway $wikidataGateway,
|
||||
private AwardTypeRepository $awardTypeRepository,
|
||||
@@ -54,6 +124,10 @@ readonly class AwardImporter
|
||||
$wikidataAwards = $allAwards[$actor->getName()] ?? [];
|
||||
|
||||
foreach ($wikidataAwards as $wikidataAward) {
|
||||
if ($this->isExcluded($wikidataAward['name'])) {
|
||||
continue;
|
||||
}
|
||||
|
||||
$awardType = $this->resolveAwardType($wikidataAward['name'], $knownTypes);
|
||||
|
||||
$award = new Award();
|
||||
@@ -74,12 +148,33 @@ readonly class AwardImporter
|
||||
*/
|
||||
private function resolveAwardType(string $awardName, array &$knownTypes): AwardType
|
||||
{
|
||||
// 1. Try canonical map first
|
||||
$canonicalName = $this->findCanonicalName($awardName);
|
||||
|
||||
if (null !== $canonicalName) {
|
||||
foreach ($knownTypes as $type) {
|
||||
if ($type->getName() === $canonicalName) {
|
||||
return $type;
|
||||
}
|
||||
}
|
||||
|
||||
$newType = new AwardType();
|
||||
$newType->setName($canonicalName);
|
||||
$newType->setPattern($canonicalName);
|
||||
$this->em->persist($newType);
|
||||
$knownTypes[] = $newType;
|
||||
|
||||
return $newType;
|
||||
}
|
||||
|
||||
// 2. Fall back to existing pattern matching
|
||||
foreach ($knownTypes as $type) {
|
||||
if (str_contains($awardName, $type->getPattern())) {
|
||||
return $type;
|
||||
}
|
||||
}
|
||||
|
||||
// 3. Create new type with prefix extraction
|
||||
$newType = new AwardType();
|
||||
$prefix = $this->extractPrefix($awardName);
|
||||
$newType->setName($prefix);
|
||||
@@ -91,10 +186,43 @@ readonly class AwardImporter
|
||||
return $newType;
|
||||
}
|
||||
|
||||
private function findCanonicalName(string $awardName): ?string
|
||||
{
|
||||
$normalized = mb_strtolower($awardName);
|
||||
|
||||
foreach (self::AWARD_MAP as $canonical => $keywords) {
|
||||
foreach ($keywords as $keyword) {
|
||||
if (str_contains($normalized, mb_strtolower($keyword))) {
|
||||
return $canonical;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return null;
|
||||
}
|
||||
|
||||
private function isExcluded(string $awardName): bool
|
||||
{
|
||||
$normalized = mb_strtolower($awardName);
|
||||
|
||||
foreach (self::EXCLUDED_KEYWORDS as $keyword) {
|
||||
if (str_contains($normalized, mb_strtolower($keyword))) {
|
||||
return true;
|
||||
}
|
||||
}
|
||||
|
||||
return false;
|
||||
}
|
||||
|
||||
private function extractPrefix(string $awardName): string
|
||||
{
|
||||
// Extract text before " for " or " pour " (common patterns in award names)
|
||||
if (preg_match('/^(.+?)\s+(?:for|pour)\s+/i', $awardName, $matches)) {
|
||||
// "X for Y", "X pour Y", "X du Y", "X de la Y", "X de l'Y", "X des Y"
|
||||
if (preg_match('/^(.+?)\s+(?:for|pour|du|de la|de l\'|des)\s+/iu', $awardName, $matches)) {
|
||||
return trim($matches[1]);
|
||||
}
|
||||
|
||||
// "... festival de cinéma de X" or "... festival de X"
|
||||
if (preg_match('/festival\s+(?:de\s+(?:cinéma\s+de\s+)?)?(.+?)$/iu', $awardName, $matches)) {
|
||||
return trim($matches[1]);
|
||||
}
|
||||
|
||||
|
||||
Reference in New Issue
Block a user