Min score for kNN searches + sys prompt change

Author: felipeelia

includes/classes/Feature/KnnSearch/KnnSearch.php

@@ -18,6 +18,15 @@
  * kNN Search Feature
  */
 class KnnSearch extends Feature {
+	/**
+	 * Default settings
+	 *
+	 * @var array $default_settings.
+	 */
+	public $default_settings = [
+		'ep_knn_search_min_score' => 0.7,
+	];
+
 	/**
 	 * Initialize feature setting it's config
 	 */
@@ -43,4 +52,35 @@ public function setup() {
 		\ElasticPress\SearchAlgorithms::factory()->register( new SearchAlgorithm\KnnCosine() );
 		\ElasticPress\SearchAlgorithms::factory()->register( new SearchAlgorithm\Hybrid() );
 	}
+
+	/**
+	 * Set the `settings_schema` attribute
+	 */
+	public function set_settings_schema() {
+		$this->settings_schema = [
+			[
+				'key'     => 'ep_knn_search_min_score',
+				'label'   => __( 'Minimum score', 'elasticpress-labs' ),
+				'help'    => __( 'The minimum score to be used by kNN searches. Input a number between 0 and 1.', 'elasticpress-labs' ),
+				'type'    => 'number',
+				'default' => $this->default_settings['ep_knn_search_min_score'],
+			],
+		];
+	}
+
+	/**
+	 * Get the min_score setting.
+	 *
+	 * @return float
+	 */
+	public function get_min_score() {
+		/**
+		 * Filters the minimum score for KNN (k-Nearest Neighbors) search.
+		 *
+		 * @since 2.5.0
+		 * @param {float} $min_score The minimum score for KNN search. Default is retrieved from the 'ep_knn_search_min_score' setting.
+		 * @return {float} The minimum score for KNN search.
+		 */
+		return (float) apply_filters( 'ep_knn_search_min_score', $this->get_setting( 'ep_knn_search_min_score' ) );
+	}
 }

includes/classes/Feature/KnnSearch/SearchAlgorithm/Hybrid.php

@@ -63,10 +63,13 @@ public function get_es_query( $formatted_args, $args, $query ): array {
 			return $formatted_args;
 		}
 
+		$knn_search_feature = \ElasticPress\Features::factory()->get_registered_feature( 'knn_search' );
+
 		return [
 			'from'        => $formatted_args['from'],
 			'size'        => $formatted_args['size'],
 			'post_filter' => $formatted_args['post_filter'],
+			'min_score'   => $knn_search_feature->get_min_score(),
 			'query'       => $formatted_args['query'],
 			'knn'         => [
 				'field'          => 'chunks.vector',

includes/classes/Feature/KnnSearch/SearchAlgorithm/Knn.php

@@ -63,10 +63,13 @@ public function get_es_query( $formatted_args, $args, $query ): array {
 			return $formatted_args;
 		}
 
+		$knn_search_feature = \ElasticPress\Features::factory()->get_registered_feature( 'knn_search' );
+
 		return [
 			'from'        => $formatted_args['from'],
 			'size'        => $formatted_args['size'],
 			'post_filter' => $formatted_args['post_filter'],
+			'min_score'   => $knn_search_feature->get_min_score(),
 			'knn'         => [
 				'field'          => 'chunks.vector',
 				'query_vector'   => array_map( 'floatval', $query_embedding ),

includes/classes/Feature/KnnSearch/SearchAlgorithm/KnnCosine.php

@@ -63,10 +63,13 @@ public function get_es_query( $formatted_args, $args, $query ): array {
 			return $formatted_args;
 		}
 
+		$knn_search_feature = \ElasticPress\Features::factory()->get_registered_feature( 'knn_search' );
+
 		return [
 			'from'        => $formatted_args['from'],
 			'size'        => $formatted_args['size'],
 			'post_filter' => $formatted_args['post_filter'],
+			'min_score'   => $knn_search_feature->get_min_score(),
 			'query'       => [
 				'bool' => [
 					'must' => [

includes/classes/Feature/RAG.php

@@ -403,7 +403,7 @@ public function ai_api_request( $prompt, $search_term ) {
 				],
 				[
 					'role'    => 'system',
-					'content' => 'Send your response as a JSON object with the following keys: "response" (the asnwer, in HTML format) and "references" (an array of objects with the URLs you used to build the response, having "url" and "title" as attributes). Do not wrap the response in any other tags or limiters like "```json". Make sure the JSON object returned is properly escaped.',
+					'content' => __( 'Send your response as a JSON object with the following keys: "response" (the asnwer, in HTML format) and "references" (an array of objects with the URLs you used to build the response, having "url" and "title" as attributes). Do not wrap the response in any other tags or limiters like "```json". Make sure the JSON object returned is properly escaped. Do not append the list of URLs used to the "response" value, as it will be displayed using the values in "references".', 'elasticpress-labs' ),
 				],
 				[
 					'role'    => 'user',