Update main.py

main.py

@@ -21,14 +21,10 @@ class TextPayload(BaseModel):
 async def analyze_text(payload: TextPayload):
     doc = nlp(payload.text)
     
-    # We use the raw tokens to verify frequencies accurately
     entity_freq = Counter([ent.text.lower() for ent in doc.ents])
     
-    # BRAND GUARDIAN: Names in this list will NEVER be split
     brand_safety = [
-        'johnson & johnson', 'ben & jerry', 'tiffany & co', 'h&m', 'at&t', 
-        'dolce & gabbana', 'marks & spencer', 'bed bath & beyond', 'a&p', 'c&a',
-        'p&g', 'procter & gamble'
+        'johnson & johnson', 'ben & jerry', 'tiffany & co', 'h&m', 'at&t'
     ]
     
     processed_entities = []
@@ -38,15 +34,10 @@ async def analyze_text(payload: TextPayload):
         raw_text = ent.text.strip()
         e_lower = raw_text.lower()
         
-        # Decide if we should split
         sub_items = [raw_text]
-        
-        # Only split if NOT a protected brand and contains 'and' or '&'
         if e_lower not in brand_safety:
             if " and " in e_lower or " & " in raw_text:
-                # Split the text into parts
                 parts = raw_text.replace(" and ", " & ").split(" & ")
-                # Only split if each part is a short name (1-2 words)
                 if all(1 <= len(p.strip().split()) <= 2 for p in parts):
                     sub_items = [p.strip() for p in parts if len(p.strip()) > 1]
 
@@ -57,16 +48,20 @@ async def analyze_text(payload: TextPayload):
 
             label = ent.label_
             
-            # Smart Category Overrides
-            tech_orgs = ['gravatar', 'hostwinds', 'facebook', 'instagram', 'bing', 'yandex', 'google']
-            tech_prods = ['vps', 'server', 'hosting', 'cloud', 'schema', 'json', 'api', 'plugin']
-            concepts = ['free', 'booster', 'get', 'noticed', 'top', 'best', 'simple']
-
-            if any(t in f_lower for t in tech_orgs): label = "ORG"
-            elif any(p in f_lower for p in tech_prods): label = "PRODUCT"
-            elif any(c in f_lower for c in concepts): label = "CONCEPT"
+            # --- UPDATED LOGIC HERE ---
+            # 1. Marketing/General Topics (Force to CONCEPT)
+            concepts = ['privacy', 'cookies', 'cookie', 'booster', 'get', 'noticed', 'free', 'terms']
             
-            # Length filter to catch those long "Privacy & Cookie" strings
+            # 2. Technical Data/Tools (Force to PRODUCT)
+            tech_items = ['exif', 'mediaif', 'cookiesif', 'json', 'api', 'schema', 'vps', 'server']
+
+            if any(c in f_lower for c in concepts):
+                label = "CONCEPT"
+            elif any(t in f_lower for t in tech_items):
+                label = "PRODUCT"
+            # ---------------------------
+
+            # Keep the length filter to avoid messy UI strings
             if len(final_text.split()) >= 4:
                 continue
 
@@ -78,7 +73,6 @@ async def analyze_text(payload: TextPayload):
             })
             seen.add(f_lower)
 
-    # Return up to 50 entities so the list feels rich and full
     sorted_entities = sorted(processed_entities, key=lambda x: x['frequency'], reverse=True)
 
     return {