remove unneeded embedding normalization

src/services/embedding_service.py

@@ -74,14 +74,9 @@ class EmbeddingService:
             # Get the image embedding vector
             embedding_vector = embeddings.image_embedding
             
-            # Convert to numpy array and normalize
+            # Convert to numpy array - DO NOT normalize Vertex AI embeddings
             embeddings_array = np.array(embedding_vector, dtype=np.float32)
             
-            # Normalize the feature vector
-            norm = np.linalg.norm(embeddings_array)
-            if norm > 0:
-                embeddings_array = embeddings_array / norm
-            
             logger.info(f"Generated image embeddings with shape: {embeddings_array.shape}")
             return embeddings_array.tolist()
             
@@ -112,14 +107,9 @@ class EmbeddingService:
             # Get the text embedding vector
             embedding_vector = embeddings.text_embedding
             
-            # Convert to numpy array and normalize
+            # Convert to numpy array - DO NOT normalize Vertex AI embeddings
             embeddings_array = np.array(embedding_vector, dtype=np.float32)
             
-            # Normalize the feature vector
-            norm = np.linalg.norm(embeddings_array)
-            if norm > 0:
-                embeddings_array = embeddings_array / norm
-            
             logger.info(f"Generated text embeddings with shape: {embeddings_array.shape}")
             return embeddings_array.tolist()
             
@@ -175,29 +165,5 @@ class EmbeddingService:
             logger.error(f"Error deleting embedding: {e}")
             return False
     
-    def calculate_similarity(self, embedding1: List[float], embedding2: List[float]) -> float:
-        """
-        Calculate cosine similarity between two embeddings
-        
-        Args:
-            embedding1: First embedding
-            embedding2: Second embedding
-            
-        Returns:
-            Cosine similarity (0-1)
-        """
-        try:
-            # Convert to numpy arrays
-            vec1 = np.array(embedding1)
-            vec2 = np.array(embedding2)
-            
-            # Calculate cosine similarity
-            similarity = np.dot(vec1, vec2) / (np.linalg.norm(vec1) * np.linalg.norm(vec2))
-            
-            return float(similarity)
-        except Exception as e:
-            logger.error(f"Error calculating similarity: {e}")
-            raise
-
 # Create a singleton service
 embedding_service = EmbeddingService() 

src/services/vector_db.py

@@ -49,7 +49,7 @@ class VectorDatabaseService:
         
         # Handle HTTPS setting properly
         if https is None:
-#TODO ensure SSL
+            #TODO ensure SSL
             https = False
         
         logger.info(f"Connecting to Qdrant at {self.host}:{self.port} (HTTPS: {https}, gRPC: {prefer_grpc})")
@@ -331,26 +331,4 @@ class VectorDatabaseService:
             return False
 
 
-# Utility functions for vector operations
-def normalize_vector(vector: List[float]) -> List[float]:
-    """Normalize a vector to unit length."""
-    vector_array = np.array(vector)
-    norm = np.linalg.norm(vector_array)
-    if norm == 0:
-        return vector
-    return (vector_array / norm).tolist()
 
-
-def cosine_similarity(vector1: List[float], vector2: List[float]) -> float:
-    """Calculate cosine similarity between two vectors."""
-    v1 = np.array(vector1)
-    v2 = np.array(vector2)
-    
-    dot_product = np.dot(v1, v2)
-    norm_v1 = np.linalg.norm(v1)
-    norm_v2 = np.linalg.norm(v2)
-    
-    if norm_v1 == 0 or norm_v2 == 0:
-        return 0.0
-    
-    return dot_product / (norm_v1 * norm_v2)