Enhancement/577 Threshold values for Embeddings

includes/Classifai/Providers/OpenAI/EmbeddingCalculations.php

@@ -15,9 +15,11 @@ class EmbeddingCalculations {
 	 *
 	 * @param array $source_embedding Embedding data of the source item.
 	 * @param array $compare_embedding Embedding data of the item to compare.
+	 * @param float $threshold The threshold to use for the similarity calculation.
+	 *
 	 * @return bool|float
 	 */
-	public function similarity( array $source_embedding = [], array $compare_embedding = [] ) {
+	public function similarity( array $source_embedding = [], array $compare_embedding = [], $threshold = 1 ) {
 		if ( empty( $source_embedding ) || empty( $compare_embedding ) ) {
 			return false;
 		}
@@ -56,8 +58,15 @@ function( $x ) {
 		// Do the math.
 		$distance = 1.0 - ( $combined_average / sqrt( $source_average * $compare_average ) );
 
-		// Ensure we are within the range of 0 to 2.0.
-		return max( 0, min( abs( (float) $distance ), 2.0 ) );
+		/**
+		 * Filter the threshold for the similarity calculation.
+		 *
+		 * @param float $threshold The threshold to use.
+		 */
+		$threshold = apply_filters( 'classifai_threshold', $threshold );
+
+		// Ensure we are within the range of 0 to 1.0 (i.e. $threshold).
+		return max( 0, min( abs( (float) $distance ), $threshold ) );
 	}
 
 }

includes/Classifai/Providers/OpenAI/Embeddings.php

@@ -268,6 +268,15 @@ public function sanitize_settings( $settings ) {
 			} else {
 				$new_settings['taxonomies'][ $taxonomy_key ] = '0';
 			}
+
+			// Sanitize the threshold setting.
+			$taxonomy_key = $taxonomy_key . '_threshold';
+			if ( isset( $settings['taxonomies'][ $taxonomy_key ] ) && '0' !== $settings['taxonomies'][ $taxonomy_key ] ) {
+				$threshold_value = min( absint( $settings['taxonomies'][ $taxonomy_key ] ), 100 );
+				$new_settings['taxonomies'][ $taxonomy_key ] = $threshold_value ? $threshold_value : 75;
+			} else {
+				$new_settings['taxonomies'][ $taxonomy_key ] = 75;
+			}
 		}
 
 		// Sanitize the number setting.
@@ -354,6 +363,34 @@ public function supported_post_types() {
 		return apply_filters( 'classifai_openai_embeddings_post_types', $this->get_supported_post_types() );
 	}
 
+	/**
+	 * Get the threshold for the similarity calculation.
+	 *
+	 * @since 2.6.0
+	 *
+	 * @param string $taxonomy Taxonomy slug.
+	 * @return float
+	 */
+	public function get_threshold( $taxonomy = '' ) {
+		$settings  = $this->get_settings();
+		$threshold = 1;
+
+		if ( ! empty( $taxonomy ) ) {
+			$threshold = isset( $settings['taxonomies'][ $taxonomy . '_threshold' ] ) ? $settings['taxonomies'][ $taxonomy . '_threshold' ] : 75;
+		}
+
+		// Convert $threshold (%) to decimal.
+		$threshold = 1 - ( (float) $threshold / 100 );
+
+		/**
+		 * Filter the threshold for the similarity calculation.
+		 *
+		 * @param float $threshold The threshold to use.
+		 * @param string $taxonomy The taxonomy to get the threshold for.
+		 */
+		return apply_filters( 'classifai_threshold', $threshold, $taxonomy );
+	}
+
 	/**
 	 * The list of supported post statuses.
 	 *
@@ -468,6 +505,9 @@ private function set_terms( int $post_id = 0, array $embedding = [] ) {
 				continue;
 			}
 
+			// Get threshold setting for this taxonomy.
+			$threshold = $this->get_threshold( $tax );
+
 			// Get embedding similarity for each term.
 			foreach ( $terms as $term_id ) {
 				if ( ! current_user_can( 'assign_term', $term_id ) && ( ! defined( 'WP_CLI' ) || ! WP_CLI ) ) {
@@ -477,9 +517,9 @@ private function set_terms( int $post_id = 0, array $embedding = [] ) {
 				$term_embedding = get_term_meta( $term_id, 'classifai_openai_embeddings', true );
 
 				if ( $term_embedding ) {
-					$similarity = $calculations->similarity( $embedding, $term_embedding );
+					$similarity = $calculations->similarity( $embedding, $term_embedding, $threshold );
 					if ( false !== $similarity ) {
-						$embedding_similarity[ $tax ][ $term_id ] = $calculations->similarity( $embedding, $term_embedding );
+						$embedding_similarity[ $tax ][ $term_id ] = $similarity;
 					}
 				}
 			}

includes/Classifai/Providers/Provider.php

@@ -404,10 +404,12 @@ public function render_checkbox_group( array $args = array() ) {
 		foreach ( $args['options'] as $option_value => $option_label ) {
 			$value       = '';
 			$default_key = array_search( $option_value, $args['default_values'], true );
+			$option_value_theshold = $option_value . '_threshold';
 
 			// Get saved value, if any.
 			if ( isset( $setting_index[ $args['label_for'] ] ) ) {
 				$value = $setting_index[ $args['label_for'] ][ $option_value ] ?? '';
+				$threshold_value = $setting_index[ $args['label_for'] ][ $option_value_theshold ] ?? '';
 			}
 
 			// Check for backward compatibility.
@@ -435,6 +437,11 @@ public function render_checkbox_group( array $args = array() ) {
 				checked( $value, $option_value, false ),
 				esc_html( $option_label )
 			);
+
+			// Render Threshold field.
+			if ( 'openai_embeddings' === $this->option_name && 'taxonomies' === $args['label_for'] ) {
+				$this->render_threshold_field( $args, $option_value_theshold, $threshold_value );
+			}
 		}
 
 		// Render description, if any.
@@ -446,6 +453,34 @@ public function render_checkbox_group( array $args = array() ) {
 		}
 	}
 
+	/**
+	 * Render a threshold field.
+	 *
+	 * @since 2.6.0
+	 *
+	 * @param array  $args          The args passed to add_settings_field
+	 * @param string $option_value  The option value.
+	 * @param string $value         The value.
+	 *
+	 * @return void
+	 */
+	public function render_threshold_field( $args, $option_value, $value ) {
+		printf(
+			'<p class="threshold_wrapper">
+				<label for="%1$s_%2$s_%3$s">%4$s</label>
+				<br>
+				<input type="number" id="%1$s_%2$s_%3$s" class="small-text" name="classifai_%1$s[%2$s][%3$s]" value="%5$s" />				
+			</p>',
+			esc_attr( $this->option_name ),
+			esc_attr( $args['label_for'] ),
+			esc_attr( $option_value ),
+			esc_html__( 'Threshold (%)', 'classifai' ),
+			esc_attr( $value )
+		);
+	}
+
+
+
 	/**
 	 * Renders the checkbox group for 'Generate descriptive text' setting.
 	 *

src/scss/language-processing.scss

@@ -171,3 +171,20 @@
 		color: rgb(117, 117, 117);
 	}
 }
+
+.form-table {
+    td {
+        p.threshold_wrapper {
+            margin: 10px 0 40px;
+
+            &:last-of-type {
+                margin-bottom: 20px;
+            }
+
+            label {
+                display: inline-block;
+                margin: 0 0 10px;
+            }
+        }
+    }
+}

tests/cypress/integration/language-processing/classify-content-openapi-embeddings.test.js

@@ -24,6 +24,7 @@ describe( '[Language processing] Classify Content (OpenAI) Tests', () => {
 		cy.get( '#openai_embeddings_post_types_post' ).check();
 		cy.get( '#openai_embeddings_post_statuses_publish' ).check();
 		cy.get( '#openai_embeddings_taxonomies_category' ).check();
+		cy.get( '#openai_embeddings_taxonomies_category_threshold' ).type( 80 ); // "Test" requires 80% confidence. At 81%, it does not apply.
 		cy.get( '#number' ).clear().type( 1 );
 		cy.get( '#submit' ).click();
 	} );