Update

model/email_feature_extractor.py

@@ -336,10 +336,22 @@ def get_function_words(text: str) -> Set[str]:
 # Phishing-related keywords (case-insensitive)
 PHISHING_KEYWORDS = {
     'account': r'\baccount\b',
+    'access': r'\baccess\b',
     'bank': r'\bbank\b',
+    'credit': r'\bcredit\b',
+    'click': r'\bclick\b',
+    'identity': r'\bidentity\b',
+    'inconvenience': r'\binconvenience\b',
     'information': r'\binformation\b',
+    'limited': r'\blimited\b',
+    'minutes': r'\bminutes?\b',
+    'password': r'\bpassword\b',
+    'recently': r'\brecently\b',
     'risk': r'\brisk\b',
+    'social': r'\bsocial\b',
     'security': r'\bsecurity\b',
+    'service': r'\bservice\b',
+    'suspended': r'\bsuspended\b',
 }
 
 
@@ -509,14 +521,16 @@ def extract_advanced_nlp_features(text: str) -> Dict[str, Any]:
 
 def extract_features(email_text: str, include_advanced: bool = False) -> Dict[str, Any]:
     """
-    Extract all 9 features from email content using enhanced NLP libraries.
+    Extract all 21 features from email content using enhanced NLP libraries.
 
     Features extracted (in exact order):
     1. Total Number of Characters C
     2. Vocabulary richness W/C
-    3-7. Keyword counts (Account, Bank, Information, Risk, Security)
-    8. Total number of Function words/W
-    9. Unique Words
+    3-19. Keyword counts (Account, Access, Bank, Credit, Click, Identity,
+          Inconvenience, Information, Limited, Minutes, Password, Recently,
+          Risk, Social, Security, Service, Suspended)
+    20. Total number of Function words/W
+    21. Unique Words
 
     Enhanced with:
     - Automatic text preprocessing and normalization (handles multi-line input)
@@ -529,7 +543,7 @@ def extract_features(email_text: str, include_advanced: bool = False) -> Dict[st
         include_advanced: If True, include advanced NLP features (not used by model)
 
     Returns:
-        dict: Dictionary containing all 9 features with exact column names
+        dict: Dictionary containing all 21 features with exact column names
               (plus optional advanced features if include_advanced=True)
     """
     # Handle empty or None input
@@ -572,15 +586,27 @@ def extract_features(email_text: str, include_advanced: bool = False) -> Dict[st
         'Total Number of Characters C': total_chars,
         'Vocabulary richness W/C': vocab_richness,
         'Account': keyword_counts['Account'],
+        'Access': keyword_counts['Access'],
         'Bank': keyword_counts['Bank'],
+        'Credit': keyword_counts['Credit'],
+        'Click': keyword_counts['Click'],
+        'Identity': keyword_counts['Identity'],
+        'Inconvenience': keyword_counts['Inconvenience'],
         'Information': keyword_counts['Information'],
+        'Limited': keyword_counts['Limited'],
+        'Minutes': keyword_counts['Minutes'],
+        'Password': keyword_counts['Password'],
+        'Recently': keyword_counts['Recently'],
         'Risk': keyword_counts['Risk'],
+        'Social': keyword_counts['Social'],
         'Security': keyword_counts['Security'],
+        'Service': keyword_counts['Service'],
+        'Suspended': keyword_counts['Suspended'],
         'Total number of Function words/W': function_word_ratio,
         'Unique Words': unique_words,
     }
 
-    logger.info(f"✓ Successfully extracted all 9 features from email (length: {total_chars} chars, words: {len(words)})")
+    logger.info(f"✓ Successfully extracted all 21 features from email (length: {total_chars} chars, words: {len(words)})")
     logger.debug(f"Core features: {features}")
 
     # Optionally include advanced NLP features