update

app.py

@@ -126,19 +126,10 @@ def load_session_data(username):
 
 def load_samples(methods):
     logger.info(f"Loading samples for methods: {methods}")
-    samples = []
+    samples = set()  # Use a set to avoid duplicates
     categories = ["TP", "TN", "FP", "FN"]
 
-    method_dirs = []
-    for method in methods:
-        if method == 'No-XAI':
-            method_dirs.append('NO_XAI')
-        elif method == 'Dater':
-            method_dirs.append('DATER')
-        elif method == 'Chain-of-Table':
-            method_dirs.append('COT')
-        elif method == 'Plan-of-SQLs':
-            method_dirs.append('POS')
+    method_dirs = [get_method_dir(method) for method in methods]
 
     for category in categories:
         dir_a = f'htmls_{method_dirs[0].upper()}/{category}'
@@ -150,18 +141,29 @@ def load_samples(methods):
         matching_files = files_a & files_b
 
         for file in matching_files:
-            samples.append({
-                'category': category,
-                'file': file
-            })
+            samples.add((category, file))
 
-    return samples
+    # Convert set of tuples back to list of dictionaries
+    samples = [{'category': category, 'file': file} for category, file in samples]
 
+    logger.info(f"Loaded {len(samples)} unique samples across all categories")
+    return samples
 
 def select_balanced_samples(samples):
     try:
-        selected_samples = random.sample(samples, min(10, len(samples)))
-        logger.info(f"Selected balanced samples: {len(selected_samples)}")
+        # Ensure we have at least 10 unique samples
+        unique_samples = list({(s['category'], s['file']) for s in samples})
+
+        if len(unique_samples) < 10:
+            logger.warning(f"Not enough unique samples. Only {len(unique_samples)} available.")
+            selected_samples = unique_samples
+        else:
+            selected_samples = random.sample(unique_samples, 10)
+
+        # Convert back to dictionary format
+        selected_samples = [{'category': category, 'file': file} for category, file in selected_samples]
+
+        logger.info(f"Selected {len(selected_samples)} unique samples")
         return selected_samples
     except Exception as e:
         logger.exception("Error selecting balanced samples")
@@ -285,7 +287,6 @@ def get_method_dir(method):
     elif method == 'Plan-of-SQLs':
         return 'POS'
 
-
 def get_visualization_dir(method):
     if method == "No-XAI":
         return 'htmls_NO_XAI'