← All changes | app/AI/Embedding/DataEmbedding.php +11 -12 3.4.1 → 3.2.1 View file →

		@@ -41,13 +41,8 @@
41	41	*
42	42	* @return bool True on success, false on failure.
43	43	*/
44	44	public function generate_and_store( $product_id, $title, $content ) {
45		- $ai_data = AIFns::activated_ai_data();
46		- if ( empty( $ai_data['api_key'] ) ) {
47		- return false;
48		- }
49		-
50	45	$text = $title . ' ' . wp_strip_all_tags( $content );
51	46	$ai_service = AIFns::initializeAIService();
52	47	$embedding = $ai_service->generateEmbedding( $text );
53	48	if ( empty( $embedding ) || ! is_array( $embedding ) ) {
		@@ -76,8 +71,9 @@
76	71	return [];
77	72	}
78	73	// Log query embedding.
79	74	$rows = AIDB::get_all();
	75	+
80	76	if ( empty( $rows ) ) {
81	77	return [];
82	78	}
83	79	$results = $this->find_similar( $query_embedding, $rows, $limit );
		@@ -97,8 +93,9 @@
97	93	* @return array List of matched items with product ID, title, and similarity score.
98	94	*/
99	95	public function find_similar( array $embedding, array $rows, int $limit = 5 ): array {
100	96	$minimum_match = AIFns::get_embedding_minimum_accuracy();
	97	+ $query_norm = sqrt( array_sum( array_map( static fn( $x ) => $x * $x, $embedding ) ) );
101	98	$matches = [];
102	99	foreach ( $rows as $row ) {
103	100	if ( empty( $row['embedding'] ) ) {
104	101	continue;
		@@ -107,9 +104,9 @@
107	104	if ( ! is_array( $vector ) ) {
108	105	continue;
109	106	}
110	107	// Avoid redundant norm computation if similarity calc includes it.
111		- $score = $this->calculate_similarity_score( $embedding, $vector );
	108	+ $score = $this->calculate_similarity_score( $embedding, $vector, $query_norm );
112	109	if ( $score >= $minimum_match ) {
113	110	$matches[] = [
114	111	'post_id' => isset( $row['product_id'] ) ? (int) $row['product_id'] : 0,
115	112	'post_title' => isset( $row['title'] ) ? sanitize_text_field( $row['title'] ) : '',
		@@ -133,20 +130,22 @@
133	130	* query norm and computes the dot product and target norm in one pass.
134	131	*
135	132	* @param array $query_vec The query embedding vector.
136	133	* @param array $target_vec The stored embedding vector.
	134	+ * @param float $query_norm Precomputed L2 norm of the query vector.
137	135	*
138	136	* @return float Cosine similarity score (0.0–1.0).
139	137	*/
140		- protected function calculate_similarity_score( array $query_vec, array $target_vec ): float {
	138	+ protected function calculate_similarity_score( array $query_vec, array $target_vec, float $query_norm ): float {
141	139	$dot = 0.0;
142		- $normA = 0.0;
143	140	$normB = 0.0;
144		- $count = min( count( $query_vec ), count( $target_vec ) );
145		- for ( $i = 0; $i < $count; $i++ ) {
	141	+ $len = min( count( $query_vec ), count( $target_vec ) );
	142	+ for ( $i = 0; $i < $len; $i++ ) {
146	143	$dot += $query_vec[ $i ] * $target_vec[ $i ];
147		- $normA += $query_vec[ $i ] ** 2;
148	144	$normB += $target_vec[ $i ] ** 2;
149	145	}
150		- return $dot / ( sqrt( $normA ) * sqrt( $normB ) );
	146	+ if ( $query_norm <= 0.0 || $normB <= 0.0 ) {
	147	+ return 0.0;
	148	+ }
	149	+ return $dot / ( $query_norm * sqrt( $normB ) );
151	150	}
152	151	}