130 lines
4.0 KiB
PHP
130 lines
4.0 KiB
PHP
<?php
|
|
|
|
declare(strict_types=1);
|
|
|
|
namespace App\Jobs;
|
|
|
|
use App\Models\Artwork;
|
|
use App\Models\RecArtworkRec;
|
|
use Illuminate\Bus\Queueable;
|
|
use Illuminate\Contracts\Queue\ShouldQueue;
|
|
use Illuminate\Foundation\Bus\Dispatchable;
|
|
use Illuminate\Queue\InteractsWithQueue;
|
|
use Illuminate\Queue\SerializesModels;
|
|
use Illuminate\Support\Facades\DB;
|
|
|
|
/**
|
|
* Compute behavior-based (co-like) similarity from precomputed item pairs.
|
|
*
|
|
* Spec §7.3 — runs nightly.
|
|
* For each artwork: read top pairs from rec_item_pairs, store top N.
|
|
*/
|
|
final class RecComputeSimilarByBehaviorJob implements ShouldQueue
|
|
{
|
|
use Dispatchable, InteractsWithQueue, Queueable, SerializesModels;
|
|
|
|
public int $tries = 2;
|
|
public int $timeout = 600;
|
|
|
|
public function __construct(
|
|
private readonly ?int $artworkId = null,
|
|
private readonly int $batchSize = 200,
|
|
) {
|
|
$queue = (string) config('recommendations.queue', 'default');
|
|
if ($queue !== '') {
|
|
$this->onQueue($queue);
|
|
}
|
|
}
|
|
|
|
public function handle(): void
|
|
{
|
|
$modelVersion = (string) config('recommendations.similarity.model_version', 'sim_v1');
|
|
$resultLimit = (int) config('recommendations.similarity.result_limit', 30);
|
|
$maxPerAuthor = (int) config('recommendations.similarity.max_per_author', 2);
|
|
|
|
$query = Artwork::query()->public()->published()->select('id', 'user_id');
|
|
|
|
if ($this->artworkId !== null) {
|
|
$query->where('id', $this->artworkId);
|
|
}
|
|
|
|
$query->chunkById($this->batchSize, function ($artworks) use ($modelVersion, $resultLimit, $maxPerAuthor) {
|
|
foreach ($artworks as $artwork) {
|
|
$this->processArtwork($artwork, $modelVersion, $resultLimit, $maxPerAuthor);
|
|
}
|
|
});
|
|
}
|
|
|
|
private function processArtwork(
|
|
Artwork $artwork,
|
|
string $modelVersion,
|
|
int $resultLimit,
|
|
int $maxPerAuthor,
|
|
): void {
|
|
// Fetch top co-occurring artworks (bi-directional)
|
|
$candidates = DB::table('rec_item_pairs')
|
|
->where('a_artwork_id', $artwork->id)
|
|
->select(DB::raw('b_artwork_id AS related_id'), 'weight')
|
|
->union(
|
|
DB::table('rec_item_pairs')
|
|
->where('b_artwork_id', $artwork->id)
|
|
->select(DB::raw('a_artwork_id AS related_id'), 'weight')
|
|
)
|
|
->orderByDesc('weight')
|
|
->limit($resultLimit * 3)
|
|
->get();
|
|
|
|
if ($candidates->isEmpty()) {
|
|
return;
|
|
}
|
|
|
|
$relatedIds = $candidates->pluck('related_id')->map(fn ($id) => (int) $id)->all();
|
|
|
|
// Fetch author info for diversity filtering
|
|
$authorMap = DB::table('artworks')
|
|
->whereIn('id', $relatedIds)
|
|
->where('is_public', true)
|
|
->where('is_approved', true)
|
|
->whereNotNull('published_at')
|
|
->where('published_at', '<=', now())
|
|
->whereNull('deleted_at')
|
|
->pluck('user_id', 'id')
|
|
->all();
|
|
|
|
// Apply diversity cap
|
|
$authorCounts = [];
|
|
$final = [];
|
|
foreach ($candidates as $cand) {
|
|
$relatedId = (int) $cand->related_id;
|
|
if (! isset($authorMap[$relatedId])) {
|
|
continue; // not public/published
|
|
}
|
|
$authorId = (int) $authorMap[$relatedId];
|
|
$authorCounts[$authorId] = ($authorCounts[$authorId] ?? 0) + 1;
|
|
if ($authorCounts[$authorId] > $maxPerAuthor) {
|
|
continue;
|
|
}
|
|
$final[] = $relatedId;
|
|
if (count($final) >= $resultLimit) {
|
|
break;
|
|
}
|
|
}
|
|
|
|
if ($final === []) {
|
|
return;
|
|
}
|
|
|
|
RecArtworkRec::query()->updateOrCreate(
|
|
[
|
|
'artwork_id' => $artwork->id,
|
|
'rec_type' => 'similar_behavior',
|
|
'model_version' => $modelVersion,
|
|
],
|
|
[
|
|
'recs' => $final,
|
|
'computed_at' => now(),
|
|
],
|
|
);
|
|
}
|
|
}
|