diff --git a/brancheneinstufung.py b/brancheneinstufung.py
index 04507f6a..6195128c 100644
--- a/brancheneinstufung.py
+++ b/brancheneinstufung.py
@@ -95,45 +95,61 @@ class WikipediaScraper:
     def __init__(self):
         wikipedia.set_lang(Config.LANG)
         
+    def _extract_domain_hint(self, website):
+        """Extrahiert den Domain-Schlüssel aus der Website-URL"""
+        if not website:
+            return ""
+        # Entferne Protokoll und www, zerlege in Teile
+        clean_url = website.lower().replace("https://", "").replace("http://", "").replace("www.", "")
+        domain_parts = clean_url.split(".")
+        return domain_parts[0] if domain_parts else ""
+
     def _generate_search_terms(self, company_name, website_hint=""):
         """Generiert Suchbegriffe aus Firmenname und Website"""
         search_terms = [company_name.strip()]
         
-        # Zusatzbegriffe aus Firmennamen
-        name_parts = company_name.split()
-        if len(name_parts) > 1:
-            search_terms.append(" ".join(name_parts[:2]))
+        # Bereinigung von Rechtsformen und Sonderzeichen
+        clean_name = re.sub(
+            r'\s+(?:GmbH|AG|KG|OHG|e\.V\.|mbH|& Co\. KG| GmbH & Co\. KG).*$', 
+            '', 
+            company_name
+        ).strip()
         
-        # Bereinigung von Rechtsformen
-        clean_name = re.sub(r'\s+(?:GmbH|AG|KG|OHG|e\.V\.|mbH).*$', '', company_name)
-        if clean_name != company_name:
+        # Füge bereinigten Namen hinzu, wenn unterschiedlich
+        if clean_name and clean_name != company_name:
             search_terms.append(clean_name)
             
-        # Extraktion aus Website
-        if website_hint:
-            domain_parts = website_hint.replace("https://", "").replace("http://", "").replace("www.", "").split(".")
-            if len(domain_parts) > 1 and domain_parts[0] not in ["de", "com", "org"]:
-                search_terms.append(domain_parts[0])
-                
+        # Extrahiere erste zwei relevante Wörter
+        name_words = [w for w in re.split(r'\W+', clean_name) if w]
+        if len(name_words) >= 2:
+            search_terms.append(" ".join(name_words[:2]))
+            
+        # Domain-Hint hinzufügen
+        domain_hint = self._extract_domain_hint(website_hint)
+        if domain_hint and domain_hint not in ["de", "com", "org", "net"]:
+            search_terms.append(domain_hint)
+            
         debug_print(f"Generierte Suchbegriffe: {search_terms}")
-        return search_terms
-    
+        return list(set(search_terms))  # Duplikate entfernen
+
     def _validate_article(self, page, company_name, domain_hint=""):
         """Überprüft ob der Artikel zum Unternehmen passt"""
-        # Ähnlichkeitsprüfung des Titels
-        title_similarity = SequenceMatcher(
-            None, 
-            page.title.lower(), 
-            company_name.lower()
-        ).ratio()
+        # Normalisiere beide Namen
+        page_title = re.sub(r'\(.*?\)', '', page.title).strip().lower()
+        search_name = re.sub(r'[^a-zA-Z0-9äöüß ]', '', company_name).strip().lower()
+        
+        # Ähnlichkeitsprüfung
+        similarity = SequenceMatcher(None, page_title, search_name).ratio()
+        debug_print(f"Ähnlichkeit '{page_title}' vs '{search_name}': {similarity:.2f}")
         
         # Zusätzliche Domain-Prüfung
         if domain_hint:
-            html_content = requests.get(page.url).text
-            if domain_hint.lower() not in html_content.lower():
+            html_content = requests.get(page.url).text.lower()
+            if domain_hint not in html_content:
+                debug_print(f"Domain-Hint '{domain_hint}' nicht im Artikel gefunden")
                 return False
                 
-        return title_similarity >= Config.SIMILARITY_THRESHOLD
+        return similarity >= Config.SIMILARITY_THRESHOLD
     
     @retry_on_failure
     def search_company_article(self, company_name, website_hint=""):